A method for evaluating the differential antibacterial activity of photocatalytic coatings against specific spoilage bacteria in chilled fresh meat.
By constructing a multidimensional discrete data matrix and fitting it into a continuous time-series response tensor using Weibull and Logistic models, and combining information entropy and clustering algorithms, the defense mechanism of photocatalytic coating against specific spoilage bacteria in chilled fresh meat is determined, and a dynamic illumination control curve is generated. This solves the shortcomings of existing photocatalytic coating evaluation methods and achieves precise antibacterial and energy-saving effects.
Patent Information
- Application Number
- CN202611215311.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-12
- Publication Date
- 2026-11-17
AI Technical Summary
Existing photocatalytic coating evaluation methods cannot accurately quantify and distinguish the microscopic defense mechanisms of specific putrefactive bacteria, resulting in a single light intervention strategy, insufficient sterilization precision, and high power consumption.
By constructing a multidimensional discrete data matrix, the Weibull distribution model and the Logistic growth model are fitted into a continuous time-series response tensor. Combining the information entropy algorithm and the one-dimensional clustering algorithm, the defense mechanism of the target bacteria resistant to the bottleneck is determined, and a step-like or constant light control curve is generated.
It enables precise evaluation of the antibacterial activity of specific spoilage bacteria in chilled fresh meat, improving sterilization accuracy and reducing equipment power consumption.
Smart Images

Figure CN123116205A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of antibacterial evaluation technology, specifically to a method for evaluating the differential antibacterial activity of photocatalytic coatings against specific spoilage bacteria in chilled fresh meat. Background Technology
[0002] Fresh meat can be contaminated by specific spoilage bacteria during cold chain storage and distribution, affecting meat quality and food safety. Photocatalytic coating technology, as a surface antibacterial method, has been applied in the field of fresh meat preservation. Accurately evaluating the antibacterial efficacy of photocatalytic coating and outputting corresponding light control strategies accordingly is a necessary step in maintaining the cleanliness of the fresh meat storage environment.
[0003] Current methods for evaluating the antibacterial properties of photocatalytic coatings typically involve extracting test matrix units at set time points and determining the antibacterial effect by detecting single indicators such as the number of macroscopic viable bacteria. This evaluation method, which relies on obtaining static test data at discrete time points, cannot calculate and reflect the dynamic response evolution of specific spoilage bacteria under photocatalytic stress throughout the entire cold chain storage period. Furthermore, existing evaluation methods mostly focus on macroscopic viable bacteria counts and cannot quantify the resistance pathways of specific spoilage bacteria at the microscopic level. In actual cold chain storage, specific spoilage bacteria exhibit different defense mechanisms when facing photocatalytic system stress. Some strains increase biofilm formation to create a physical shielding effect that blocks light quantum transmission, while others reduce the activity of key enzymes in the tricarboxylic acid cycle and enter an endogenous metabolic dormant state to maintain survival.
[0004] Because existing methods cannot accurately determine and distinguish the specific defense mechanisms of different specific putrefactive bacteria, the control system cannot output a matching light intervention curve. Generally, it can only operate with a single fixed light intensity or maintain the highest power output. The above-mentioned fixed light control mode can neither overcome the physical barrier of the biofilm of specific putrefactive bacteria, nor can it be reasonably adjusted according to the photon quantum demand of dormant strains. This results in insufficient sterilization accuracy of photocatalytic devices and a large amount of ineffective energy consumption. Summary of the Invention
[0005] To address the shortcomings of existing technologies, this invention provides a method for evaluating the differential antibacterial activity of photocatalytic coatings against specific spoilage bacteria in chilled fresh meat. This method solves the problems of existing evaluation methods relying on discrete static data and being unable to quantitatively distinguish the microscopic defense mechanisms of specific spoilage bacteria, resulting in a single light intervention strategy and high power consumption of the equipment.
[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for evaluating the differential antibacterial activity of a photocatalytic coating against specific spoilage bacteria in chilled fresh meat, the method comprising the following steps: Test matrix units of discrete time series nodes are extracted to obtain multidimensional discrete test data, and the multidimensional discrete test data is used to construct an initial discrete data matrix; The Weibull distribution model and the Logistic growth model are used to fit the multidimensional discrete test data in the initial discrete data matrix into continuous biological response decay functions and cumulative functions, respectively. The virtual continuous data points in the continuous time domain are extracted and recombined into a third-order continuous time series response tensor. Extract the continuous response values of biofilm formation from the third-order continuous time-series response tensor and calculate the interface shading attenuation coefficient; perform range normalization on the third-order continuous time-series response tensor to obtain normalized tensor data, and use the information entropy algorithm to calculate the information entropy and generate the feature index weight vector. Extract the static slice data corresponding to the end time of the set evaluation period from the third-order continuous time-series response tensor. Combine the feature index weight vector, static slice data and effective photocatalytic transmission factor converted from the interface shading attenuation coefficient to calculate the comprehensive antibacterial response index. Call the one-dimensional clustering algorithm to lock and mark the target bacteria that are tolerant to the bottleneck. Read the time-dimensional index data corresponding to the target bacteria of tolerance bottleneck and backtrack to extract the response trajectory of the dominant weighted index. Determine whether the target bacteria of tolerance bottleneck execute a physical shielding defense mechanism or an endogenous metabolic dormancy mechanism, and generate a step-like dynamic light control curve or a constant light intensity control curve as the evaluation result.
[0007] By mapping discrete detection point data into continuous time-series tensors using the Weibull distribution model and Logistic growth model, an evolutionary model of microbial stress response is constructed in the time dimension. Combining information entropy weights and interface shading attenuation coefficients, the attenuation process of photocatalytic excitation efficiency with microbial biofilm growth is calculated. By combining one-dimensional clustering algorithm to identify bottleneck-tolerant strains and retrospectively analyzing the dominant stress resistance mechanism of target bacteria with bottleneck tolerance, a matching light source output model is provided for strains that execute different defense mechanisms, thereby improving the sterilization accuracy and energy utilization rate of photocatalytic equipment in cold chain environments.
[0008] Furthermore, the step of extracting test matrix units from discrete time series nodes to obtain multidimensional discrete test data, and constructing an initial discrete data matrix from the multidimensional discrete test data, includes: Multidimensional discrete test data were extracted, including macroscopic decrease in the number of viable bacteria, microscopic increase in cell membrane permeability, biofilm formation, relative malondialdehyde content, and residual activity of key enzymes in the tricarboxylic acid cycle. The multidimensional discrete test data were used to construct an initial discrete data matrix according to the time dimension, strain dimension, detection index dimension, and treatment group dimension. The treatment group dimension included at least the blank control group, the light-only group, the coating-only group, and the light-and-coating synergistic group.
[0009] This step involves simultaneously collecting macroscopic count indicators and microscopic biochemical indicators to provide basic data support for subsequent analysis of different stress resistance mechanisms of specific putrefactive bacteria.
[0010] Furthermore, the steps of using the Weibull distribution model and the Logistic growth model to fit the multidimensional discrete test data in the initial discrete data matrix into continuous biological response decay functions and cumulative functions, respectively, include: The residual activity rate of key enzymes in the tricarboxylic acid cycle and the logarithmic value of residual viable bacteria obtained by converting macroscopic viable bacteria counts were classified into a set of attenuation indicators. The Weibull distribution model was called to perform nonlinear regression fitting calculation, and the discrete attenuation test data were fitted and mapped into a continuous biological response attenuation function. The macroscopic decrease in the logarithm of viable bacteria, the increase in microscopic cell membrane permeability, the measured value of biofilm formation, and the relative content of malondialdehyde were classified into a cumulative index set. The Logistic growth model was called to perform nonlinear regression fitting calculation, and the discrete cumulative test data were fitted and mapped to a continuous cumulative function.
[0011] This step selects the corresponding nonlinear regression model based on the changing patterns of different detection indicators, converting discrete test data into a function expression in the continuous time domain, thus avoiding the time blind spot caused by discrete sampling.
[0012] Furthermore, the steps of extracting the continuous response values of biofilm formation from the third-order continuous-time response tensor and calculating the generation interface shading attenuation coefficient include: The extinction coefficient at a specific wavelength is multiplied with the continuous response value of biofilm formation, which is extracted from the third-order continuous time response tensor. The product result is negative, and the negative product result is used as the exponential term of an exponential function with the natural constant as the base to obtain the exponential decay value. By subtracting the exponential decay value from the numerical value, the excitation efficiency decay coefficient of the photocatalytic coating caused by the physical shielding effect of the biofilm is calculated and generated in the continuous time dimension. The excitation efficiency decay coefficient of the photocatalytic coating is then output as the interface shielding decay coefficient.
[0013] This step uses the extinction coefficient and the continuous response numerical calculation of biofilm formation amount to calculate the attenuation coefficient, quantifying the physical shading phenomenon caused by the increase in biofilm thickness into an attenuation parameter of the photocatalytic coating excitation efficiency, which is used to correct the subsequent light intervention model.
[0014] Furthermore, the steps of performing range normalization on the third-order continuous time-series response tensor to obtain normalized tensor data, calculating information entropy using the information entropy algorithm, and generating feature index weight vectors include: Based on the positive and negative mapping relationship between various detection indicators and the final photocatalytic sterilization efficiency, the multidimensional stress detection indicators are divided into a set of positive feature indicators and a set of negative feature indicators. The system smoothing constant in the underlying database is extracted as an arithmetic compensation term. For the positive feature indicator set and the negative feature indicator set, the range normalization processing with the system smoothing constant is called to perform the dimensionless mapping operation and output normalized tensor data. The feature weight of each detection index at a specific spatiotemporal coordinate point is calculated relative to the sum of the same index in the entire spatiotemporal domain. A compensation weight set is constructed by injecting a minimal positive offset into the feature weight value. Based on the compensation weight set, the feature index information entropy of the multidimensional stress detection index is generated by calling the Shannon information entropy theory model.
[0015] This step utilizes the system smoothing constant and the minimum positive offset to prevent zero values from occurring during normalization and information entropy calculation, which could lead to program errors. The information entropy algorithm is used to automatically generate corresponding weights based on the dispersion of the detection index data, replacing the subjective weighting method.
[0016] Furthermore, the steps for extracting static slice data corresponding to the end time of the set evaluation period from the third-order continuous time-series response tensor, and calculating the comprehensive antibacterial response index by combining the feature index weight vector, static slice data, and effective photocatalytic transmission factor converted from the interface shading attenuation coefficient, include: Extract the static slice data corresponding to the end time of the set evaluation period from the third-order continuous time-series response tensor, and perform inner product summation operation on the static slice data of a single spoilage strain of fresh meat and the feature index weight vector to output the static basic response value. The interface shading attenuation coefficient corresponding to the end of the evaluation period is extracted and converted into an effective photocatalytic transmission factor. The effective photocatalytic transmission factor is multiplied with the static basic response value to generate the comprehensive antibacterial response index of the specific spoilage strain of chilled fresh meat at the end of the evaluation period.
[0017] This step combines static slice data, index weights, and effective photocatalytic transmission factors to calculate a comprehensive index, which is used to quantitatively determine the survival status of different specific putrefactive strains under photocatalytic stress at the end of the evaluation period.
[0018] Furthermore, the steps of using a one-dimensional clustering algorithm to locate and label target bacteria resistant to bottlenecks include: The one-dimensional clustering algorithm is configured as a standard K-means clustering model. The minimum, mean, and maximum values of the comprehensive antibacterial response index are extracted and established as the initial geometric centers of the low-sensitivity, medium-sensitivity, and high-sensitivity clusters, respectively. Based on the principle of minimizing Euclidean distance, the strains are assigned to the corresponding sensitivity clusters. The cluster center update and assignment operations are repeated until the cluster center coordinate offset is lower than the system tolerance constant. A set of specific spoilage strains belonging to the low-sensitivity cluster of chilled fresh meat is extracted. Within the low-sensitivity cluster, a minimum value search operation is performed on candidate strains to extract specific candidate strains with the lowest comprehensive antimicrobial response index and lock them as the target bacteria for tolerance bottleneck.
[0019] This step uses a one-dimensional clustering algorithm to classify mixed bacterial strains according to their sensitivity based on the comprehensive antibacterial response index. In the low-sensitivity cluster, the target bacteria with the strongest resistance to the bottleneck are identified, thus establishing the target computation object for subsequent determination of defense mechanisms.
[0020] Furthermore, the steps of reading the time-dimensional index data corresponding to the target bacteria of the tolerance bottleneck and backtracking to extract the response trajectory of the dominant weighted index to determine whether the target bacteria of the tolerance bottleneck are executing a physical shielding defense mechanism or an endogenous metabolic dormancy mechanism include: The weighted response trajectory corresponding to the biofilm formation measurement value is assigned to the external physical barrier feature group. The integral value of the external physical barrier feature group is the integral value of the product of the normalized response value of biofilm formation and the feature index weights of the biofilm formation measurement value in the continuous time domain. The weighted response trajectory corresponding to the residual activity rate of the key enzyme in the tricarboxylic acid cycle is assigned to the internal physiological dormancy feature group. The integral value of the internal physiological dormancy feature group is the integral value of the product of the normalized response value of the residual activity rate of the key enzyme in the tricarboxylic acid cycle and the feature index weights of the residual activity rate of the key enzyme in the tricarboxylic acid cycle in the continuous time domain. The integral ratio of the external physical barrier feature group and the internal physiological dormancy feature group in the continuous time domain is calculated. When the integral ratio is less than the decision constant 1.0, the target bacteria that tolerate the bottleneck are determined to execute the physical shielding defense mechanism. When the integral ratio is greater than or equal to the decision constant 1.0, the target bacteria that tolerate the bottleneck are determined to execute the endogenous metabolic dormancy mechanism.
[0021] This step calculates the weighted response trajectory integral of biofilm formation and the weighted response trajectory integral of enzyme activity residue rate in the continuous time domain, and compares the two integrals to determine whether the target bacteria that tolerate the bottleneck resist the photocatalytic system stress by increasing physical barriers or by reducing metabolic activity.
[0022] Furthermore, in the step of generating a stepped dynamic light control curve or a constant light intensity control curve as the evaluation result, when it is determined that the target bacteria resistant to the bottleneck are implementing a physical shielding defense mechanism, the specific process of generating the stepped dynamic light control curve includes: The interface shading attenuation coefficient is extracted, and a pulsed high-energy beam that is positively correlated with the interface shading attenuation coefficient is superimposed on a base constant light intensity. A step-type dynamic illumination control curve with superimposed pulse compensation waveform is calculated and generated, and the dynamic illumination output intensity in the continuous time domain is discretized and sampled.
[0023] For target bacteria that have reached the bottleneck of physical shielding defense mechanisms, this step adds a pulsed high-energy beam with an attenuation coefficient to the basic constant light intensity to compensate for the decrease in light quantum transmittance caused by the physical shielding of the biofilm.
[0024] Furthermore, in the step of generating a stepped dynamic light control curve or a constant light intensity control curve as the evaluation result, when it is determined that the target bacteria tolerant to the bottleneck are executing an endogenous metabolic dormancy mechanism, the specific process of generating the constant light intensity control curve includes: The residual safety threshold is extracted as an algebraic boundary condition. Based on the first-order kinetic lethality model, the residual safety threshold is converted into the total photon demand dose. The total photon demand dose is equal to the difference between the initial inoculated live bacteria logarithm and the residual safety threshold multiplied by the tenth-order dose constant of the specific spoilage bacteria. The total photon demand dose is divided by the set total cold chain storage time to generate a constant light intensity control curve. The constant light intensity obtained by the solution is limited to within the preset safe light intensity upper limit.
[0025] For target bacteria that have reached the bottleneck of endogenous metabolic dormancy, this step calculates the total dose of light required to meet the residual safety threshold based on the kinetic lethality model. The total dose of light required is then evenly distributed over the total cold chain storage time to output constant light control parameters, thereby continuously disrupting the metabolic function of dormant specific putrefactive bacteria.
[0026] This invention provides a method for evaluating the differential antibacterial activity of photocatalytic coatings against specific spoilage bacteria in chilled fresh meat. It offers the following beneficial effects: 1. This invention uses the Weibull distribution model and the Logistic growth model to fit the multidimensional discrete test data in the initial discrete data matrix into third-order continuous time-series response tensors. This solves the data discontinuity defect caused by existing evaluation methods that rely solely on static data at a single time point. This invention uses nonlinear regression fitting calculation to convert the multidimensional discrete test data into a function expression in the continuous time domain, realizing continuous calculation of the stress response state of specific spoilage bacteria throughout the entire cold chain storage cycle, and improving the continuity of photocatalytic antibacterial evaluation data.
[0027] 2. This invention uses a one-dimensional clustering algorithm to identify the target bacteria that tolerate the bottleneck from a multi-strain mixed system. By calculating the integral ratio between the integral values of the external physical barrier feature groups and the integral values of the internal physiological dormancy feature groups in the continuous time domain, it determines whether the target bacteria that tolerate the bottleneck are implementing a physical shielding defense mechanism or an endogenous metabolic dormancy mechanism. The above evaluation steps overcome the limitation of traditional macroscopic viable bacteria counting methods that cannot distinguish microscopic stress resistance pathways. It can accurately identify the specific pathways by which the target bacteria that tolerate the bottleneck resist photocatalytic stress, and provide a clear mechanism judgment standard for the subsequent formulation of antibacterial strategies.
[0028] 3. This invention outputs a matching control curve based on the determination of the defense mechanism of the target bacteria with tolerance bottleneck. When the defense mechanism is determined to be a physical shielding mechanism, a pulsed high-energy beam is superimposed on the basic constant light intensity to compensate for the transmission attenuation caused by the biofilm. When the defense mechanism is determined to be an endogenous metabolic dormancy mechanism, a constant light intensity control curve that meets the total photon requirement is output according to the first-order kinetic lethality model. The above-mentioned method of outputting control curves by classification changes the traditional single light intervention mode. It can provide corresponding light output parameters according to the actual defense mechanism of specific putrefactive bacteria, and reduce the ineffective power consumption of the equipment while ensuring the antibacterial efficacy of the photocatalytic system. Attached Figure Description
[0029] Figure 1 This is a schematic diagram of the system timing interaction of the present invention; Figure 2 This is a schematic diagram of the method flow of the present invention; Figure 3 This is a schematic diagram comparing the discrete-time dynamic responses of the multi-processor groups according to the present invention; Figure 4 This is a schematic diagram of the data tensor calculation and objective weighting process of the feature index of the present invention; Figure 5 This is a schematic diagram of the extraction and control curve generation process of the target bacteria with tolerance bottleneck in this invention; Figure 6 This is a schematic diagram showing the changes in the logarithmic values of specific putrefactive bacteria under different treatment groups according to the present invention. Figure 7 This is a schematic diagram illustrating the changes in the amount of biofilm formed by specific putrefactive bacteria under different treatment groups according to the present invention. Detailed Implementation
[0030] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0031] See attached document Figure 1 This invention provides a differential antibacterial evaluation system for photocatalytic coatings against specific spoilage bacteria in chilled fresh meat. The system includes: a sample acquisition module, a data detection module, a tensor calculation module, and a process output module.
[0032] The sample acquisition module constructs a physical field with set basic illumination parameters and constant temperature and humidity parameters, and configures an independent number of parallel test matrix units for various specific spoilage bacteria of chilled fresh meat. The sample acquisition module extracts a fixed number of parallel test matrix units at preset discrete time series nodes to terminate the photocatalytic reaction and separates them into discrete destructive detection samples.
[0033] The data detection module connects to an external biological detection terminal to receive destructive detection data corresponding to each discrete time point. The data detection module acquires the macroscopic log decrease value of viable bacteria, the microscopic increase rate of cell membrane permeability, the biofilm formation amount, the relative content of malondialdehyde, and the residual activity rate of key enzymes in the tricarboxylic acid cycle of the test matrix unit. The data detection module merges the multidimensional discrete test data into an initial discrete data matrix according to the time dimension, strain dimension, detection index dimension, and treatment group dimension. The treatment group dimension includes at least a blank control group, a light-only group, a coating-only group, and a light-and-coating synergistic group, which is used to distinguish the destructive detection data under different treatment conditions.
[0034] The tensor calculus module is connected to the data detection module and is used to receive the initial discrete data matrix. The tensor calculus module calls the Weibull distribution model and the Logistic growth model to fit the discrete data into a continuous biological response decay function and a cumulative function. It also extracts virtual continuous data points according to equal step time intervals and reconstructs and generates a third-order continuous temporal response tensor. The tensor calculus module substitutes the biofilm formation quantity function in the third-order continuous temporal response tensor into the light transmission attenuation model and generates corresponding interface shading attenuation coefficients for different specific spoilage bacteria of chilled fresh meat. The tensor calculus module performs range normalization processing on the third-order continuous temporal response tensor and uses the information entropy algorithm to calculate the information entropy of the feature index to output the feature index weight vector of each detection dimension.
[0035] The process output module is connected to the tensor calculation module to receive the feature index weight vector and the third-order continuous temporal response tensor. At the end of the set evaluation period, the process output module calculates the comprehensive antibacterial response index of different specific spoilage bacteria in fresh meat and calls the one-dimensional clustering algorithm to perform sensitivity clustering to extract the target bacteria of the tolerance bottleneck. The process output module backtracks the dominant weighted index response trajectory of the target bacteria of the tolerance bottleneck. The dominant weighted index response trajectory is obtained by multiplying the feature index weight of the corresponding detection index with the normalized tensor response value, and its resistance mechanism is mapped to physical shielding defense type or endogenous metabolic dormancy type. For physical shielding defense types, the process output module uses the interface shielding attenuation coefficient to generate a stepped dynamic lighting control curve with superimposed pulse compensation waveform. For endogenous metabolic dormancy types, the process output module combines the residual safety threshold boundary conditions to solve and generate a constant light intensity control curve. The generated stepped dynamic lighting control curve or constant light intensity control curve is used as the evaluation result, and the lighting control curve is output offline to the external hardware lighting controller.
[0036] See attached document Figure 2 This invention provides a method for evaluating the differential antibacterial activity of a photocatalytic coating against specific spoilage bacteria in chilled fresh meat, the method comprising the following steps: S1, the sample acquisition module constructs a parallel sample array containing independent test matrix units. At the set discrete time series nodes, a fixed number of test matrix units are extracted to terminate the photocatalytic reaction. The data detection module obtains the macroscopic logarithmic decrease value of viable bacteria, the microscopic cell membrane permeability increase rate, the biofilm formation amount, the relative content of malondialdehyde, and the residual activity rate of key enzymes in the tricarboxylic acid cycle for the corresponding test matrix units. The above multidimensional discrete test data are fused according to the time dimension, strain dimension, detection index dimension, and treatment group dimension to construct an initial discrete data matrix. Among them, the initial discrete data matrix contains the destructive detection discrete feature data corresponding to the time series nodes. The treatment group dimension is used to identify the detection source of the blank control group, the light-only group, the coating-only group, and the light and coating synergistic group.
[0037] S2, the tensor calculus module extracts the initial discrete data matrix, calls the Weibull distribution model and the Logistic growth model to fit the multidimensional discrete test data into the biological response decay function and the cumulative function, respectively, and extracts virtual continuous data points in the continuous time domain according to the set step size, and reconstructs and generates a third-order continuous time series response tensor.
[0038] S3, the tensor calculus module extracts the biofilm formation quantity function from the third-order continuous time-series response tensor, substitutes the biofilm formation quantity function into the preset light transmission attenuation model, calculates and generates the interface shading attenuation coefficient corresponding to each specific spoilage strain of chilled fresh meat, and establishes a quantitative mapping structure between the bioaccumulation characteristics of the target strain and the photocatalytic physical shading effect.
[0039] S4, the tensor calculus module performs range normalization on the third-order continuous temporal response tensor to eliminate dimensional differences, uses the information entropy algorithm to calculate the information entropy of each detection index in the entire temporal and spatial domain, and generates and outputs the feature index weight vector of each detection dimension based on the distribution result of the information entropy.
[0040] S5, the process output module extracts the feature index weight vector and the third-order continuous time-series response tensor. Combining the weight data and the effective photocatalytic transmission factor converted from the interface shading attenuation coefficient, it calculates the comprehensive antibacterial response index of different specific spoilage bacteria of chilled fresh meat at the end of the evaluation period. It calls the one-dimensional clustering algorithm to perform sensitivity clustering based on the comprehensive antibacterial response index, extracts and locks the tolerance bottleneck target bacteria belonging to the low sensitivity cluster. The larger the comprehensive antibacterial response index, the more sensitive the corresponding strain is to the photocatalytic coating. The smaller the comprehensive antibacterial response index, the stronger the tolerance of the corresponding strain.
[0041] S6, the process output module 400 reads the time-dimensional index data corresponding to the target bacteria that tolerate the bottleneck and traces back the dominant weighted index response trajectory of the target bacteria in the tensor data. The dominant weighted index response trajectory is formed by multiplying the feature index weight with the corresponding normalized tensor value, and determines whether the target bacteria that tolerate the bottleneck execute a physical shielding defense mechanism or an endogenous metabolic dormancy mechanism. For the physical shielding defense mechanism, the process output module 400 extracts the interface shielding attenuation coefficient to generate a stepped dynamic illumination control curve with superimposed pulse compensation waveform. For the endogenous metabolic dormancy mechanism, the process output module 400 extracts the residual safety threshold as a boundary condition, calculates and solves to generate a constant light intensity control curve, thereby generating a stepped dynamic illumination control curve or a constant light intensity control curve as the evaluation result. The process output module 400 merges and encapsulates the calculated illumination control curve parameters and outputs them offline to an external hardware lighting controller to execute control commands.
[0042] See attached document Figure 3 In specific implementations, step S1 provided by the present invention may include the following steps: S11, the sample acquisition module starts the environmental control cabinet in the laboratory environment and constructs a physical field with set basic illumination parameters and constant temperature and humidity parameters. The sample acquisition module maintains the internal environment of the physical field within the operating range that simulates the actual cold chain processing and storage of chilled fresh meat through the internal refrigeration compressor and humidifier. The physical field is configured to maintain a constant temperature range of 0°C to 4°C and a relative humidity range of 80% to 95%, or it can be preset according to the actual storage conditions of chilled fresh meat, and is equipped with a visible light LED light source with a peak wavelength of 420 nanometers.
[0043] The base illuminance, distance between the light source and the test substrate unit, and illumination angle of the visible light LED light source should be kept consistent within the same evaluation batch. The base illuminance is preferably set to 1 mW / cm². 2 Up to 50mW / cm 2The distance between the light source and the liquid surface of the test matrix unit is preferably set to 5cm to 30cm. The sample collection module is configured with an independent number of parallel test matrix units for each of the various specific spoilage bacteria of chilled fresh meat. The target evaluation object is selected from the typical spoilage dominant strains isolated and identified from the chilled fresh meat source and inoculated into the corresponding number of parallel test matrix units. The extraction and culture amplification of specific spoilage strains can be carried out according to the existing microbial isolation and purification technology. The parallel test matrix units were uniformly filled with a set volume of simulated meat gravy liquid matrix, specifically 10 mL to 50 mL. The initial inoculation concentration of the simulated meat gravy liquid matrix was set at 10 mg / mL. 6 A suspension of specific putrefactive bacteria from colony-forming units.
[0044] Each parallel test matrix unit includes a transparent reaction well, a photocatalytic coating slide fixedly disposed at the bottom or sidewall of the transparent reaction well, and a simulated meat juice liquid matrix covering the photocatalytic coating slide. The photocatalytic coating slide has a consistent effective reaction area, coating thickness, and light-receiving orientation. The biofilm formation amount is measured as a quantitative value of the biofilm formed on the light-receiving surface of the photocatalytic coating slide or at the interface adjacent to the light-receiving surface of the photocatalytic coating slide.
[0045] The sample acquisition module deploys isolated blank control groups, individual light groups, individual coating groups, and light and coating synergy groups within the physical field. Based on the preset number of discrete time series nodes within the evaluation period and the matrix consumption limit that must be met for a single destructive test, the sample acquisition module generates the total scale value of the parallel sample array. The number of discrete time series nodes is set to 10 to 15, the matrix consumption limit is set to 2 mL to 5 mL, the discrete time series nodes are set with a fixed step size of two hours, and the total evaluation period is set to 24 hours. The sample acquisition module places the corresponding number of parallel test matrix units on a porous reaction plate and arrays them in the corresponding photocatalytic test station within the physical field.
[0046] S12, the sample acquisition module starts its internal timer counter during the physical field operation according to the set evaluation cycle. When the sample acquisition module reaches the set discrete time sequence node, it removes a fixed number of parallel test matrix units from the porous reaction plate according to the random sampling principle. The sample acquisition module applies an opaque cover to the removed parallel test matrix units to perform physical shielding and light avoidance operation, and injects a chemical quencher into the matrix liquid surface of the parallel test matrix units.
[0047] The chemical quencher is selected from a quenching system that can remove reactive oxygen species generated by the photocatalytic reaction and does not significantly change the detection results of the target putrefactive bacteria activity at a set concentration. The set concentration ranges from 10 mmol / L to 50 mmol / L. The chemical quencher includes one or more of sodium thiosulfate, catalase, superoxide dismutase, mannitol or isopropanol. The specific type is determined according to the main type of reactive oxygen species generated by the photocatalytic coating and is corrected by a blank quenching control.
[0048] The physical shielding and chemical quencher injection operations immediately terminate the photocatalytic reaction inside the corresponding parallel test matrix unit and block the dark reaction continuation caused by the extraction time difference. The sample acquisition module separates and encapsulates the parallel test matrix unit after the reaction has been terminated and constructs it into discrete destructive detection samples. The data detection module connects to the external biological detection terminal through the data bus and acquires the destructive detection data of each discrete time series node.
[0049] The data detection module acquires the macroscopic decrease in the number of viable bacteria, the microscopic increase in cell membrane permeability, the biofilm formation amount, the relative content of malondialdehyde, and the residual activity rate of key enzymes in the tricarboxylic acid cycle for the corresponding parallel test matrix units. The quantitative detection method for multidimensional characteristic response indicators can be implemented based on existing biological analysis techniques.
[0050] Specifically, the data detection module obtains the macroscopic viable bacteria log decrease value, which characterizes the macroscopic bactericidal kinetics, by calling the external plate counter interface, and then calculates the residual viable bacteria log value. The data detection module obtains the microscopic cell membrane permeability increase rate by calling the external flow cytometer interface. The data detection module obtains the biofilm formation amount, which characterizes the degree of biofilm accumulation, by calling the external ELISA reader interface. The data detection module obtains the relative malondialdehyde content, which characterizes the degree of lipid peroxidation damage of the target strain, and the residual rate of tricarboxylic acid cycle key enzyme activity, which characterizes the degree of energy metabolism damage, by calling the colorimetric analyzer. The residual rate of tricarboxylic acid cycle key enzyme activity is the percentage of the selected key enzyme activity relative to the untreated control group. The key enzyme includes at least one of citrate synthase, succinate dehydrogenase, or isocitrate dehydrogenase.
[0051] S13, after acquiring the multidimensional response index data generated by multiple sets of destructive test samples at all discrete time series nodes, the data detection module activates its internal cache. The data detection module performs outlier removal and arithmetic averaging on the parallel sample data extracted from the same time series node. The outlier removal and arithmetic averaging operations smooth the random data fluctuation characteristics of a single test sample collection. The data detection module establishes a four-dimensional mapping space coordinate axis in the system's running memory. The data detection module merges and encapsulates the processed multidimensional discrete test data into an initial discrete data matrix according to the time dimension, the specific putrefactive strain dimension, the multidimensional stress detection index dimension, and the treatment group dimension. The initial discrete data matrix constitutes a multidimensional feature vector space containing the destructive test data corresponding to each discrete time series node.
[0052] Among them, the treatment group dimension is used to distinguish between the blank control group, the light-only group, the coating-only group, and the light-and-coating synergistic group. Data from different treatment groups under the same strain, the same time point, and the same detection index are stored independently and used for subsequent calculation of the correction response value under the synergistic effect of light and coating. The data detection module will persistently transfer the generated initial discrete data matrix to the system's local relational database or physical cache through the data bus.
[0053] The data detection module exposes the memory addressing pointers corresponding to the initial discrete data matrix to subsequent logic processing nodes. The specific basic data structure of the initial discrete data matrix is as follows: ; In the formula, This is the initial discrete data matrix; These are the specific discrete test feature values measured at each positive integer index mapping position; For each node in a discrete time series, there is a positive integer index. A positive integer index for a specific type of putrefactive bacteria; A positive integer index item for multidimensional stress detection index type; This is for processing positive integer index items in the group dimension.
[0054] The initial discrete data matrix resides in the system memory architecture as an independent data object, and the actual stress performance parameters of the underlying photocatalytic coating are transformed into a structured continuous computational basis through the direct mapping between physical environmental characteristics and biological response characteristics.
[0055] See attached document Figure 4 In specific implementations, step S2 provided by the present invention may include the following steps: S21, the tensor calculus module calls the memory addressing pointer through the system's internal data bus and extracts the initial discrete data matrix. The tensor calculus module performs attribute classification on the multidimensional stress detection indicators in the initial discrete data matrix according to the biological change trend. The tensor calculus module divides the residual rate of the key enzyme activity of the tricarboxylic acid cycle and the logarithmic value of residual viable bacteria obtained by converting macroscopic viable bacteria count into the attenuation index set.
[0056] The logarithmic decline value of macroscopic viable bacteria is classified into a set of positive efficacy accumulation indicators. Under photocatalytic physical field stress, specific spoilage bacteria in chilled meat exhibit a nonlinear survival curve characteristic of slow decay due to initial resistance and rapid extinction due to later defense collapse. Based on the nonlinear survival curve characteristics, the tensor calculation module calls the Weibull distribution model to perform nonlinear regression fitting calculation on the multidimensional discrete test data in the decay index set. The tensor calculation module solves the model parameters by iterative approximation based on the existing least squares method principle.
[0057] The tensor calculus module fits and maps discrete decay test data into a continuous biological response decay function. This biological response decay function constitutes a continuous mathematical expression of the continuous decline in the survival ability of specific spoilage bacteria in chilled meat under photocatalytic stress. The specific calculation formula for the biological response decay function is as follows: ; In the formula, For continuous time nodes The response values of the decay-type indicators; For physical quantities of time in the continuous time domain; These are the initial index values before photocatalytic treatment; A scale parameter to characterize the basic resistance of specific spoilage bacteria in chilled fresh meat to photocatalytic stress; The shape parameter is used to characterize the nonlinear concave-convex features of the decay rate as a function of time.
[0058] In S22, the tensor calculus module classifies the macroscopic decrease in the logarithm of viable bacteria, the increase rate of microscopic cell membrane permeability, the measured value of biofilm formation, and the relative content of malondialdehyde in the initial discrete data matrix into a set of cumulative indicators. Among them, the macroscopic decrease in the logarithm of viable bacteria, the increase rate of microscopic cell membrane permeability, and the relative content of malondialdehyde belong to positive efficacy cumulative indicators, while the measured value of biofilm formation belongs to defense cumulative indicators. The physical process of biofilm defense substance accumulation and the chemical process of oxidative damage substance accumulation inside cells conform to the typical population environmental capacity-limited growth kinetics.
[0059] The Tensor Calculus module, based on the growth dynamics of population environmental capacity constraints, calls the Logistic growth model to perform nonlinear regression fitting calculations on multidimensional discrete test data in the cumulative index set. The Tensor Calculus module relies on the existing least squares method to perform regression fitting operations on model parameters, and maps the discrete cumulative test data to a continuous cumulative function.
[0060] The cumulative function constitutes a continuous mathematical expression describing the degree of damage to specific spoilage bacteria in chilled meat under photocatalytic stress or the nonlinear accumulation process of defensive substances over time. The specific formula for calculating the cumulative function is as follows: ; In the formula, For continuous time nodes The cumulative response value of the following indicators; For physical quantities of time in the continuous time domain; This represents the upper limit of the theoretical physical extreme value that cumulative indicators can reach within the evaluation period. This is a characteristic cumulative rate constant that characterizes the maximum growth rate of the index response value; The consecutive time points corresponding to the point where half of the maximum accumulation rate is reached.
[0061] S23. After completing the continuous function fitting of decay and accumulation indicators, the tensor calculus module performs virtual data resampling in the continuous time domain. The tensor calculus module divides the total evaluation period into a continuous set of virtual time-series nodes according to the set time step. The set time step is fixed to 0.1 hours to 0.5 hours based on the Nyquist sampling theorem and the floating-point computing power boundary of the underlying computer processor. The fixed time step avoids the risk of distortion of high-frequency biological response characteristics in the tensor mapping space. The tensor calculus module substitutes all virtual time-series nodes into the biological response decay function and accumulation function and calculates and generates virtual continuous data points for each detection indicator under each virtual time-series node.
[0062] The tensor calculus module pre-calculates the correction response value under the synergistic effect of light and coating based on the data from the blank control group, the single illumination group, the single coating group, and the light and coating synergistic group in the processing group dimension. The correction response value is the net synergistic gain obtained by subtracting the discrete test feature values of the single illumination group and the single coating group from the discrete test feature values of the light and coating synergistic group, and then adding the discrete test feature values of the blank control group. All virtual continuous data points are spatially combined according to the continuous time dimension, the specific putrefactive strain dimension, and the multidimensional stress detection index dimension.
[0063] The tensor calculus module reconstructs aligned virtual continuous data points in three-dimensional space to generate a third-order continuous temporal response tensor. The specific mathematical structure of the third-order continuous temporal response tensor is as follows: ; In the formula, The generated third-order continuous temporal response tensor; It is a piecewise function mapping operator composed of the biological response decay function and the accumulation function; These are the positive integer indexes of the consecutive virtual time-series nodes after resampling; This is a fixed physical time step constant. For the first This test identifies the characteristic markers of specific spoilage bacteria in chilled fresh meat. A positive integer index for a specific type of putrefactive bacteria; For the first Characteristic identifiers of multidimensional stress detection indicators; A positive integer index item for the multidimensional stress detection index type.
[0064] In specific implementations, step S3 provided by this invention may include the following steps: S31, the tensor calculus module calculates and generates the final mathematical expression of the interface occlusion attenuation coefficient, which dynamically changes with the continuous time dimension, as follows: ; In the formula, The attenuation coefficient of the photocatalytic coating excitation efficiency caused by the physical shielding effect of the biofilm over a continuous time dimension; For physical quantities of time in the continuous time domain; Extinction coefficient for a specific wavelength; This represents the continuous response values of biofilm formation quantity extracted from the third-order continuous-time response tensor, which vary with the continuous time dimension.
[0065] The initial biofilm structure equivalent logic and basic parameter configuration state of the interface shading attenuation coefficient have been verified and established by the tensor calculus module. The tensor calculus module extracts continuous time-series slice data from the third-order continuous time-series response tensor and physically and macroscopically maps the continuous time-series slice data to the thickness reference of the photon absorption medium. Among them, the biofilm formation amount measurement value is converted into equivalent shading thickness or equivalent absorbance through a preset conversion factor. The specific value range of the preset conversion factor is configured to be 0.1 to 0.5. The preset conversion factor is determined through transmittance calibration experiments of blank coated slides in the same batch. The tensor calculus module directly completes the fusion operation of the previous correlation variables based on the existing Beer-Lambert law.
[0066] S32, the tensor calculation module skips the linear transition derivation process and directly performs the objective function transition calculation. The specific range of the extinction coefficient at a specific wavelength is fixed and configured as 0.05 to 0.15 absorbance units per unit of biofilm formation. The tensor calculation module directly merges the specific wavelength extinction coefficient with the continuous response value of biofilm formation and substitutes it into the exponential decay space to output the final continuous time domain interface shading attenuation coefficient matrix. The interface shading attenuation coefficient matrix is encapsulated by the tensor calculation module and written into the system's local relational database as the pre-absolute constraint boundary for external nonlinear process control curve calculation.
[0067] In specific implementations, step S4 provided by this invention may include the following steps: S41, the tensor calculus module extracts multidimensional stress detection index data covering all specific spoilage strains of chilled fresh meat from the third-order continuous temporal response tensor through the system's internal data bus. The multidimensional stress detection index includes parameters of different physical dimensions, such as the macroscopic decrease in the logarithm of viable bacteria and the microscopic increase in cell membrane permeability. Based on the positive and negative mapping relationship between each detection index and the final photocatalytic sterilization efficiency, the tensor calculus module divides the multidimensional stress detection index into a set of positive characteristic indicators and a set of negative characteristic indicators. Among them, the macroscopic decrease in the logarithm of viable bacteria, the microscopic increase in cell membrane permeability, and the relative content of malondialdehyde are divided into a set of positive characteristic indicators, while the biofilm formation measurement value and the residual activity rate of key enzymes in the tricarboxylic acid cycle are divided into a set of negative characteristic indicators.
[0068] The tensor calculus module extracts the system smoothing constant from the underlying database as an arithmetic compensation term. The specific value of the system smoothing constant is configured to be 10. -6 The system smoothing constant is used to avoid overflow anomalies caused by division by zero due to constant detection index values. The tensor calculus module calls the range normalization process with the system smoothing constant to perform dimensionless mapping operations for the positive feature index set and the negative feature index set, respectively.
[0069] Range normalization scales the original discrete physical unit measurements to a specific range of pure numbers. It eliminates the arithmetic distortion caused by cross-dimensional physical quantities and outputs normalized tensor data. This normalized tensor data forms the dimensionless algebraic basis for subsequent information content evaluation. The range normalization mapping formula between the positive and negative feature index sets is as follows: ; ; In the formula, The normalized tensor values of the positive feature index set after dimensionless processing; The normalized tensor values of the negative feature index set after dimensionless processing; This represents the single original continuous-time response value extracted from the third-order continuous-time response tensor; This refers to the maximum original numerical extreme value of the same detection index within a continuous time domain; The minimum original numerical extreme value corresponding to the same detection index within the continuous time domain; This is a system smoothing constant set to prevent division by zero.
[0070] S42, after acquiring normalized tensor data, the tensor calculus module starts its internal information entropy analysis engine. The system information content evaluation principle is based on the inverse proportional mapping relationship between data variation dispersion and objective information carrying capacity. The tensor calculus module calculates the feature weight of each detection index at a specific spatiotemporal coordinate point relative to the sum of the same index in the entire spatiotemporal domain. Based on the feature weight values, the tensor calculus module injects a minimum positive offset to construct a compensation weight set. The minimum positive offset is set to the negative 10th power of the mathematical natural constant. The minimum positive offset fills the algebraic meaningless dead zone caused by substituting absolute zero values into the logarithmic function.
[0071] The tensor calculus module uses the Shannon information entropy theory model based on the compensation weight set to calculate and generate the feature index information entropy of the multidimensional stress detection index. The feature index information entropy quantifies the degree of change of the corresponding detection index with time and strain dimension within the evaluation period. The specific arithmetic solution process for the compensation weight set and feature index information entropy can be executed according to the existing information entropy algorithm. The tensor calculus module constrains the value range of the feature index information entropy to a closed interval between 0 and 1.
[0072] S43, the tensor calculus module constructs feature difference coefficients based on the calculated feature index information entropy. The feature difference coefficient is calculated by subtracting the corresponding feature index information entropy from the unit constant 1. The tensor calculus module divides the feature difference coefficient of a single detection index by the sum of the feature difference coefficients of all detection indices to output the feature index weight vector. The specific calculation formula for the feature index weight vector is as follows: ; In the formula, For the first The weight values of the characteristic indicators of each detection indicator; A positive integer index item for multidimensional stress detection index type; For the first The feature index information entropy corresponding to each detection indicator; is a positive integer constant greater than 1 for the total number of multidimensional stress detection indicators.
[0073] The feature index weight vector is encapsulated by the tensor calculus module in the form of a data dictionary and registered in the system's global variable sharing pool. The global variable sharing pool grants unidirectional access to the feature index weight vector to subsequent nonlinear performance evaluation nodes. The cross-dimensional detection index system completes the objective weighting process based on the discrete evolution characteristics of the data itself in the time and space coordinate system. The objective weighting process eliminates the static solidification intervention of subjective prior experience on the evaluation system.
[0074] See attached document Figure 5 In specific implementations, step S5 provided by the present invention may include the following steps: S51, the process output module synchronously calls the feature index weight vector and the third-order continuous time series response tensor generated by the tensor calculation module through the internal data bus of the system. The process output module extracts the static slice data corresponding to the end time of the set evaluation period from the third-order continuous time series response tensor. The static slice data represents the final response state of the multi-dimensional stress detection index at the end of the evaluation period. The process output module performs inner product summation operation on the static slice data and feature index weight vector of a single chilled fresh meat specific spoilage strain. The inner product summation operation compresses the multi-dimensional feature information to a one-dimensional scalar space and outputs the static basic response value.
[0075] The process output module synchronously extracts the interface shading attenuation coefficient corresponding to the end of the evaluation cycle and converts the interface shading attenuation coefficient into an effective photocatalytic transmission factor. The effective photocatalytic transmission factor is 1 minus the interface shading attenuation coefficient. The process output module performs a product operation on the effective photocatalytic transmission factor and the static basic response value. The product operation completes the arithmetic coupling operation of physiological resistance and photon deprivation effect. The process output module generates the comprehensive antibacterial response index of the corresponding specific spoilage strain of chilled fresh meat at the end of the evaluation cycle through the product operation. The larger the comprehensive antibacterial response index, the more sensitive the corresponding strain is to the photocatalytic coating. The smaller the comprehensive antibacterial response index, the stronger the tolerance of the corresponding strain.
[0076] S52, the specific algebraic logic for calculating the comprehensive antibacterial response index in the process output module is as follows: ; In the formula, For the corresponding number The item identifies the comprehensive antimicrobial response index of specific spoilage strains in chilled fresh meat; For the first This test identifies the characteristic markers of specific spoilage bacteria in chilled fresh meat. A positive integer index for a specific type of putrefactive bacteria; The attenuation coefficient of the photocatalytic coating excitation efficiency is calculated at the end of the evaluation period. The positive integer index of the time sequence node corresponding to the end of the evaluation period; This is a fixed physical time step constant. A positive integer index item for multidimensional stress detection index type; The total number of positive integer constants greater than 1 for the multidimensional stress detection index; For the first The weight values of the characteristic indicators of each detection indicator; To be at the end of the evaluation period strain and the first Normalized tensor data obtained from the index mapping calculation.
[0077] The process output module combines the comprehensive antibacterial response index of all specific spoilage strains of chilled fresh meat into a one-dimensional feature vector. The one-dimensional feature vector is written into the underlying computing memory by the process output module and grants one-way memory read permissions to subsequent clustering nodes.
[0078] S53, the process output module starts the built-in one-dimensional clustering algorithm engine for the one-dimensional feature vector. The process output module configures the one-dimensional clustering algorithm as the standard K-means clustering model. When the number of specific spoilage strains of chilled fresh meat participating in the evaluation is greater than or equal to 3, the K value is set to 3; when the number of specific spoilage strains of chilled fresh meat participating in the evaluation is less than 3, the K value is set to the number of strains participating in the evaluation. The process output module extracts the extreme values and mean values of the one-dimensional feature vector. The process output module establishes the minimum value, mean value, and maximum value as the initial geometric centers of the low-sensitivity cluster, the medium-sensitivity cluster, and the high-sensitivity cluster.
[0079] When the difference between the maximum and minimum values of the comprehensive antimicrobial response index is less than the system tolerance constant, the process output module determines that all strains are at the same sensitivity level. Based on the minimum value of the comprehensive antimicrobial response index and the lexicographical order of the strain numbers, it directly locks onto the target strain that is the bottleneck of tolerance. A fixed initial geometric center avoids the algorithm dead zone caused by random initialization falling into local optima. The process output module calculates the Euclidean distance between the comprehensive antimicrobial response index and the initial geometric center. Based on the principle of minimizing the Euclidean distance, the process output module assigns the strains to the corresponding sensitivity clusters. The process output module repeats the cluster center update and assignment operation until the cluster center coordinate offset is lower than the system tolerance constant. The process output module then configures the system tolerance constant to 10. -5 Tolerance numerical constraints on the infinite loop boundary of floating-point arithmetic.
[0080] S54, after clustering convergence, the process output module extracts a set of specific spoilage strains of chilled fresh meat belonging to the low-sensitivity cluster. The process output module performs a minimum extreme value search operation on candidate strains within the low-sensitivity cluster and extracts the specific candidate strain with the lowest comprehensive antibacterial response index.
[0081] When multiple candidate strains have the same minimum comprehensive antimicrobial response index, the process output module retrieves the internal allocation number dictionary of the test strain system and extracts a unique specific candidate strain according to the principle of minimum lexicographical order. The lexicographical order comparison logic avoids algorithm selection deadlock when extreme values are equal. The process output module locks the specific candidate strain and marks it as the target strain tolerate the bottleneck. The target strain tolerate the bottleneck represents the process optimization limit target point of the test matrix under photocatalytic field conditions. The process output module pushes the data packet containing the identification number and comprehensive antimicrobial response index of the target strain tolerate the bottleneck to the system's process traceability cache pool. The process traceability cache pool connects to execute the task of generating the illumination control curve.
[0082] In specific implementations, step S6 provided by the present invention may include the following steps: S61, the process output module reads the time dimension index data corresponding to the target bacteria with tolerance bottleneck from the system shared pool and backtracks to extract the response trajectory of the dominant weighted index. Based on the response trajectory of the dominant weighted index, the process output module assigns the weighted response trajectory corresponding to the biofilm formation amount measurement value to the external physical barrier feature group, and assigns the weighted response trajectory corresponding to the residual activity rate of the key enzyme in the tricarboxylic acid cycle to the internal physiological dormancy feature group.
[0083] The process output module calculates the integral ratio of the external physical barrier feature group and the internal physiological dormancy feature group in the continuous time domain. The integral value of the external physical barrier feature group is the integral value of the product of the normalized response value of biofilm formation and its feature index weight in the continuous time domain. The integral value of the internal physiological dormancy feature group is the integral value of the product of the normalized response value of the residual activity rate of the key enzyme in the tricarboxylic acid cycle and its feature index weight in the continuous time domain.
[0084] The process output module embeds the value 10 in the denominator of the integral ratio calculation. -6 The system smoothing constant avoids the arithmetic overflow dead zone when the division by zero at the underlying level. Since the characteristic index adopts negative range normalization, the tolerance strength is inversely proportional to the normalized integral value. Therefore, when the integral ratio is less than the decision constant 1.0, the process output module determines that the target bacteria of the tolerance bottleneck will execute the physical shielding defense mechanism. When the integral ratio is greater than or equal to the decision constant 1.0, the process output module determines that the target bacteria of the tolerance bottleneck will execute the endogenous metabolic dormancy mechanism.
[0085] S62, the process output module extracts the interface shielding attenuation coefficient for specific target points implementing the physical shielding defense mechanism. The process output module uses this interface shielding attenuation coefficient to calculate and generate a stepped dynamic illumination control curve with superimposed pulse compensation waveforms. The increasing interface shielding attenuation coefficient over time characterizes the physical intensification of the photon blocking effect. The process output module superimposes a pulsed high-energy beam, positively correlated with the interface shielding attenuation coefficient, onto a base constant light intensity. This high-energy beam penetrates the thickened biofilm within an extremely short duty cycle and maintains the lethal photon absorption of specific putrefactive bacteria at the bottom layer. The specific calculation formula for the stepped dynamic illumination control curve is as follows: ; In the formula, For continuous time nodes The dynamic illumination output intensity calculated below; For physical quantities of time in the continuous time domain; The set baseline constant light intensity; The pulse compensation gain coefficient is a positive number, and its specific value ranges from 0.1 to 0.5. The attenuation coefficient of the photocatalytic coating excitation efficiency caused by the physical shielding effect of the biofilm over a continuous time dimension; This is a normalized high-frequency square wave pulse oscillation generation function.
[0086] The process output module discretizes and samples the dynamic illumination output intensity in the continuous time domain and writes it into the output buffer queue.
[0087] S63, the process output module extracts the residual safety threshold as an algebraic boundary condition for the specific target point of the endogenous metabolic dormancy mechanism. The residual safety threshold represents the maximum logarithmic value of specific spoilage bacteria that can survive when chilled meat reaches the end of its commercial shelf life. It is preset according to the microbial limit requirements of chilled meat shelf life, the enterprise's preset preservation control standards or experimental evaluation standards. The residual safety threshold is specifically set to 2log CFU / mL to 4log CFU / mL.
[0088] The process output module converts the residual safety threshold into the total photon demand dose based on a first-order kinetic lethality model. The conversion logic for the total photon demand dose is as follows: the total photon demand dose equals the difference between the initial logarithmic value of the inoculated viable bacteria and the residual safety threshold, multiplied by the decimal dose constant of the specific putrefactive bacteria. The decimal dose constant represents the known specific photon dose required to reduce the number of bacteria by one logarithmic cycle, and its specific value range is 1 J / cm². 2 Up to 10 J / cm 2 .
[0089] The process output module divides the total photon requirement dose by the set total cold chain storage time to generate a constant light intensity control curve, and limits the obtained constant light intensity within a preset safe light intensity upper limit, which is set to 100mW / cm². 2 When the light intensity exceeds the preset safe light intensity limit, the process output module extends the irradiation time, reduces the target residual bacterial count requirement, or outputs an over-limit warning parameter. The continuous, constant light field input without mutation suppresses the potential risk of the dynamic waveform inducing putrefactive bacteria to start deep dormancy stress expression. The static light output intensity corresponding to the constant light intensity control curve is temporarily stored in the process cache node as a low-level execution parameter by the process output module.
[0090] S64, the process output module merges and encapsulates the execution parameters of the calculated step-type dynamic illumination control curve or constant light intensity control curve into a standard industrial control protocol data packet. The process output module embeds a cyclic redundancy check code at the end of the industrial control protocol data packet. The cyclic redundancy check code verifies the data integrity of the long-distance physical bus transmission. The process output module outputs the industrial control protocol data packet offline to the external hardware lighting controller through a standard serial interface. The external hardware lighting controller parses the received execution parameters and directly drives the photocatalytic light-emitting array in the cold fresh meat storage environment. Based on the differential antibacterial evaluation system of the underlying data evolution characteristics, the customized physical field environment parameters for high-resistance target strains are output in a closed loop. That is, the generated step-type dynamic illumination control curve or constant light intensity control curve is output as the final evaluation result.
[0091] To aid in understanding the technical solution of this invention, a specific application embodiment and test example are provided below.
[0092] Application Examples: The differential antimicrobial evaluation system provided by this invention is deployed in a cold chain environment simulation chamber with a set temperature of 4°C and a relative humidity of 90%. The target test objects include three specific spoilage bacteria isolated from chilled fresh meat: *Pseudomonas berriesii*, *Thermophyton floccosum*, and *Lactobacillus sakei*. The sample collection module is configured with parallel test matrix units for each of the three specific spoilage bacteria, each containing a simulated meat juice matrix and a photocatalytically coated slide. The basic constant light intensity is set at 20 mW / cm². 2 .
[0093] The system operation evaluation cycle was set to 24 hours. The sample collection module extracted a fixed number of test matrix units at the set discrete time series nodes, injected sodium thiosulfate at a concentration of 20 mmol / L to terminate the photocatalytic reaction, and the data detection module obtained the viable number of each strain by plate counting method and calculated the antibacterial rate based on the viable number of the treatment group and the corresponding blank control group of the same strain at the same time node. At the same time, the cell membrane permeability increase rate, biofilm formation amount, residual activity rate of key enzymes in the tricarboxylic acid cycle and relative content of malondialdehyde were obtained by flow cytometry and enzyme-linked immunosorbent assay (ELISA).
[0094] The tensor calculus module receives the aforementioned discrete data matrix. Based on the plate counting results, it constructs time-series data on the viable bacterial survival rate and performs continuous mapping using the Weibull distribution model to analyze the decay trend of the viable bacterial survival rate. It also performs continuous mapping using the Logistic growth model to analyze the biofilm accumulation trend. The tensor calculus module reconstructs and generates a third-order continuous time-series response tensor based on a set step size. Finally, it performs range normalization on the third-order continuous time-series response tensor and outputs a feature index weight vector using the information entropy algorithm.
[0095] The process output module calculates the comprehensive antimicrobial response index of three specific putrefactive bacteria at the end of the evaluation period by combining the feature index weight vector and the interface shading attenuation coefficient. The process output module calls a one-dimensional clustering algorithm to perform sensitivity clustering. After system calculation, the comprehensive antimicrobial response indices of *Pseudomonas berryis* and *Lactobacillus sakei* are similar in spatial distance and are merged into the susceptible response cluster by the algorithm. *Thermophyton floccosum*, due to its low response index, is independently separated into the tolerance feature cluster by the algorithm. The process output module thus quantifies and extracts *Thermophyton floccosum* and locks it as the target bacteria for tolerance bottleneck. In response to the physical shading defense mechanism, the process output module uses the interface shading attenuation coefficient to generate a stepped dynamic lighting control curve with superimposed pulse compensation waveform and outputs it to the external hardware lighting controller.
[0096] Test example: This test case sets up three comparative experimental groups: a blank control group without photocatalytic coating and light source, a control group with photocatalytic coating and constant input of 20mW / cm², and a control group with photocatalytic coating and constant input of 20mW / cm². 2 A static illumination group for light intensity, and a dynamic control group configured with a photocatalytic coating and driven by a stepped dynamic illumination control curve.
[0097] The cell membrane permeability increase rate was calculated based on the cell membrane permeability of the initial sample at 0 hours for each strain. For ease of comparison, the table lists four representative response indicators: inhibition rate, biofilm formation amount, cell membrane permeability increase rate, and relative malondialdehyde content.
[0098] The test data is summarized in the table below.
[0099] Table of test data for multidimensional response indexes in the differential antibacterial evaluation system: According to the table above and appendix Figure 6 and attached Figure 7 The data shows that in the static light group, the inhibition rates of *Pseudomonas berriesii* and *Lactobacillus sakei* were 94.2% and 93.6%, respectively, while the inhibition rate of *Thermophyton floccosum* was only 78.4%, demonstrating strong environmental tolerance. The tensor calculus module captured the biofilm OD of *Thermophyton floccosum* under static light conditions. 590 The measured value was 0.91, which was significantly higher than that of the other two strains, indicating that the biofilm formed by it produced an interfacial shielding effect, reduced the effective light intensity reaching the photocatalytic coating, and weakened the transfer of photocatalytic active species to the bacterial interface.
[0100] This invention identifies and targets *Thermophyton floccosum* as a bottleneck-tolerant target bacterium through objective weighting and clustering. Based on its interface occlusion characteristics, it generates a stepped dynamic illumination control curve with superimposed pulse-compensated waveforms. In the dynamic control group, the biofilm OD of *Thermophyton floccosum*... 590 The measured value decreased to 0.31, the cell membrane permeability increase rate increased to 96.3%, the relative content of malondialdehyde increased to 2.16 μmol / mg prot, and the antibacterial rate increased to 97.6%; the antibacterial rates of Pseudomonas berries and Lactobacillus sakei also increased to 99.3% and 99.1%, respectively. The above results indicate that the present invention can achieve closed-loop regulation of resistant bacteria recognition, dynamic light parameter output, and targeted inhibition.
[0101] The "-" in the column corresponding to the antibacterial rate of the blank control group in the table indicates that this group is used as the baseline group for calculating the antibacterial rate, and its own antibacterial rate is not calculated. The antibacterial rates of the static light group and the dynamic control group are calculated based on the ratio of their viable bacteria count to the viable bacteria count of the corresponding blank control group of the same strain at the same time point. Therefore, the blank control group does not have an independent antibacterial rate value relative to itself. The use of "-" as a placeholder for the corresponding position does not mean that viable bacteria count testing was not performed or that relevant data was not recorded.
[0102] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A method for evaluating the differential bacteriostatic effect of a photocatalytic coating on specific spoilage bacteria of chilled fresh meat, characterized in that, The method includes the following steps: Test matrix units of discrete time series nodes are extracted to obtain multidimensional discrete test data, and the multidimensional discrete test data is used to construct an initial discrete data matrix; The Weibull distribution model and the Logistic growth model are used to fit the multidimensional discrete test data in the initial discrete data matrix into continuous biological response decay functions and cumulative functions, respectively. The virtual continuous data points in the continuous time domain are extracted and recombined into a third-order continuous time series response tensor. Extract the continuous response values of biofilm formation from the third-order continuous time-series response tensor and calculate the generation interface shading attenuation coefficient. The range normalization process is performed on the third-order continuous time-series response tensor to obtain normalized tensor data. The information entropy is calculated using the information entropy algorithm and a feature index weight vector is generated. Extract the static slice data corresponding to the end time of the set evaluation period from the third-order continuous time-series response tensor. Combine the feature index weight vector, static slice data and effective photocatalytic transmission factor converted from the interface shading attenuation coefficient to calculate the comprehensive antibacterial response index. Call the one-dimensional clustering algorithm to lock and mark the target bacteria that are tolerant to the bottleneck. Read the time-dimensional index data corresponding to the target bacteria of tolerance bottleneck and backtrack to extract the response trajectory of the dominant weighted index. Determine whether the target bacteria of tolerance bottleneck execute a physical shielding defense mechanism or an endogenous metabolic dormancy mechanism, and generate a step-like dynamic light control curve or a constant light intensity control curve as the evaluation result.
2. The method for evaluating the difference in bacteriostatic effect of the photocatalytic coating on specific spoilage bacteria of chilled fresh meat according to claim 1, characterized in that, The step of extracting test matrix units from discrete time series nodes to obtain multidimensional discrete test data, and constructing an initial discrete data matrix from the multidimensional discrete test data, includes: Multidimensional discrete test data were extracted, including macroscopic decrease in the number of viable bacteria, microscopic increase in cell membrane permeability, biofilm formation, relative malondialdehyde content, and residual activity of key enzymes in the tricarboxylic acid cycle. The multidimensional discrete test data were used to construct an initial discrete data matrix according to the time dimension, strain dimension, detection index dimension, and treatment group dimension. The treatment group dimension included at least the blank control group, the light-only group, the coating-only group, and the light-and-coating synergistic group.
3. The method for evaluating the difference in bacteriostatic effect of the photocatalytic coating on specific spoilage bacteria of chilled fresh meat according to claim 2, characterized in that, The steps of using the Weibull distribution model and the Logistic growth model to fit the multidimensional discrete test data in the initial discrete data matrix into continuous biological response decay functions and accumulation functions, respectively, and extracting virtual continuous data points in the continuous time domain to reconstruct a third-order continuous time-series response tensor include: The residual activity rate of key enzymes in the tricarboxylic acid cycle and the logarithmic value of residual viable bacteria obtained by converting macroscopic viable bacteria counts were classified into a set of attenuation indicators. The Weibull distribution model was called to perform nonlinear regression fitting calculation, and the discrete attenuation test data were fitted and mapped into a continuous biological response attenuation function. The macroscopic decrease in the logarithm of viable bacteria, the increase in microscopic cell membrane permeability, the measured value of biofilm formation, and the relative content of malondialdehyde were classified into a cumulative index set. The Logistic growth model was called to perform nonlinear regression fitting calculation, and the discrete cumulative test data were fitted and mapped to a continuous cumulative function.
4. The method for evaluating the differential antibacterial activity of the photocatalytic coating against specific spoilage bacteria in chilled fresh meat according to claim 1, characterized in that, The step of extracting the continuous response values of biofilm formation in the third-order continuous temporal response tensor and calculating the interface shading attenuation coefficient includes: The extinction coefficient at a specific wavelength is multiplied with the continuous response value of biofilm formation, which is extracted from the third-order continuous time-series response tensor. Take the negative value of the product operation result, and use the negative product operation result as the exponential term of the exponential function with the natural constant as the base to obtain the exponential decay value; By subtracting the exponential decay value from the numerical value, the excitation efficiency decay coefficient of the photocatalytic coating caused by the physical shielding effect of the biofilm is calculated and generated in the continuous time dimension. The excitation efficiency decay coefficient of the photocatalytic coating is then output as the interface shielding decay coefficient.
5. The method for evaluating the differential antibacterial activity of the photocatalytic coating against specific spoilage bacteria in chilled fresh meat according to claim 1, characterized in that, The steps of performing range normalization on the third-order continuous temporal response tensor to obtain normalized tensor data, calculating information entropy using the information entropy algorithm, and generating feature index weight vectors include: Based on the positive and negative mapping relationship between various detection indicators and the final photocatalytic sterilization efficiency, the multidimensional stress detection indicators are divided into a set of positive characteristic indicators and a set of negative characteristic indicators. Extract the system smoothing constant from the underlying database as an arithmetic compensation term, and call the range normalization process with the system smoothing constant to perform dimensionless mapping operation for the positive feature index set and the negative feature index set respectively, and output normalized tensor data. The feature weight of each detection index at a specific spatiotemporal coordinate point is calculated relative to the sum of the same index in the entire spatiotemporal domain. A compensation weight set is constructed by injecting a minimal positive offset into the feature weight value. Based on the compensation weight set, the feature index information entropy of the multidimensional stress detection index is generated by calling the Shannon information entropy theory model.
6. The method for evaluating the differential antibacterial activity of the photocatalytic coating against specific spoilage bacteria in chilled fresh meat according to claim 1, characterized in that, The steps of extracting static slice data corresponding to the end time of the set evaluation period from the third-order continuous time-series response tensor, and calculating the comprehensive antibacterial response index by combining the feature index weight vector, the static slice data, and the effective photocatalytic transmission factor converted from the interface shading attenuation coefficient, include: Extract the static slice data corresponding to the end time of the set evaluation period from the third-order continuous time-series response tensor, and perform inner product summation operation on the static slice data of a single spoilage strain of fresh meat and the feature index weight vector to output the static basic response value. Extract the interface shading attenuation coefficient corresponding to the end time of the evaluation period, and convert the interface shading attenuation coefficient into an effective photocatalytic transmission factor. The effective photocatalytic transmission factor is multiplied with the static basic response value to generate the comprehensive antibacterial response index of the specific spoilage strain of chilled fresh meat at the end of the evaluation period.
7. The method for evaluating the differential antibacterial activity of the photocatalytic coating against specific spoilage bacteria in chilled fresh meat according to claim 1, characterized in that, The step of calling a one-dimensional clustering algorithm to locate and label target bacteria resistant to bottlenecks includes: The one-dimensional clustering algorithm was configured as a standard K-means clustering model, and the minimum, mean, and maximum values of the comprehensive antibacterial response index were extracted and established as the initial geometric centers of the low-sensitivity cluster, the medium-sensitivity cluster, and the high-sensitivity cluster, respectively. Based on the principle of minimizing Euclidean distance, the strains are assigned to the corresponding sensitivity clusters. The cluster center update and assignment operations are repeated until the cluster center coordinate offset is lower than the system tolerance constant. A set of specific spoilage strains belonging to the low-sensitivity cluster of chilled fresh meat is extracted. Within the low-sensitivity cluster, a minimum value search operation is performed on candidate strains to extract specific candidate strains with the lowest comprehensive antimicrobial response index and lock them as the target bacteria for tolerance bottleneck.
8. The method for evaluating the differential antibacterial activity of the photocatalytic coating against specific spoilage bacteria in chilled fresh meat according to claim 1, characterized in that, The steps of reading the time-dimensional index data corresponding to the target bacteria of the tolerance bottleneck and backtracking to extract the response trajectory of the dominant weighted index, and determining whether the target bacteria of the tolerance bottleneck are executing a physical shielding defense mechanism or an endogenous metabolic dormancy mechanism, include: The weighted response trajectory corresponding to the biofilm formation measurement value is assigned to the external physical barrier feature group. The integral value of the external physical barrier feature group is the integral value of the product of the normalized response value of biofilm formation and the feature index weights of the biofilm formation measurement value in the continuous time domain. The weighted response trajectory corresponding to the residual activity rate of key enzymes in the tricarboxylic acid cycle is assigned to the internal physiological dormancy feature group. The integral value of the internal physiological dormancy feature group is the integral value of the normalized response value of the residual activity rate of key enzymes in the tricarboxylic acid cycle and the weight of the feature index of the residual activity rate of key enzymes in the tricarboxylic acid cycle in the continuous time domain. The integral ratio of the external physical barrier feature group and the internal physiological dormancy feature group in the continuous time domain is calculated. When the integral ratio is less than the decision constant 1.0, the target bacteria that tolerate the bottleneck are determined to execute the physical shielding defense mechanism. When the integral ratio is greater than or equal to the decision constant 1.0, the target bacteria that tolerate the bottleneck are determined to execute the endogenous metabolic dormancy mechanism.
9. The method for evaluating the differential antibacterial activity of the photocatalytic coating against specific spoilage bacteria in chilled fresh meat according to claim 8, characterized in that, In the step of generating a stepped dynamic illumination control curve or a constant light intensity control curve as an evaluation result, when it is determined that the target bacteria resistant to the bottleneck are implementing a physical shielding defense mechanism, the specific process of generating the stepped dynamic illumination control curve includes: Extract the interface shading attenuation coefficient and superimpose a pulsed high-energy beam that is positively correlated with the interface shading attenuation coefficient onto a base constant light intensity; The algorithm calculates and generates a stepped dynamic illumination control curve with superimposed pulse compensation waveform, and discretizes the dynamic illumination output intensity in the continuous time domain.
10. The method for evaluating the differential antibacterial activity of the photocatalytic coating against specific spoilage bacteria in chilled fresh meat according to claim 8, characterized in that, In the step of generating a stepped dynamic illumination control curve or a constant light intensity control curve as an evaluation result, when it is determined that the target bacteria resistant to the bottleneck are performing an endogenous metabolic dormancy mechanism, the specific process of generating a constant light intensity control curve includes: The residual safety threshold is extracted as an algebraic boundary condition. Based on the first-order kinetic lethality model, the residual safety threshold is converted into the total photon demand dose. The total photon demand dose is equal to the difference between the initial inoculated live bacteria logarithm and the residual safety threshold multiplied by the tenth-order dose constant of the specific putrefactive bacteria. The constant light intensity control curve is generated by dividing the total photon requirement dose by the set total cold chain storage time, and the constant light intensity obtained is limited to within the preset safe light intensity upper limit.
