A method and system for improving meat quality of jinhua pigs based on data mining

Abstract

This invention provides a method and system for improving the quality of Jinhua pork based on data mining, belonging to the field of smart farming technology. The method includes: acquiring a raw dataset; performing data fusion based on the raw dataset, constructing an entity-relationship network topology to obtain a multimodal association graph; extracting potential influence paths from the multimodal association graph to obtain a set of influence paths; modeling interaction effects based on the set of influence paths to obtain a dynamic action model; tracing the source of meat quality defects based on the dynamic action model to obtain a set of defect driving factors; and generating optimization strategies based on the set of defect driving factors to obtain a meat quality improvement decision scheme. This invention integrates multi-source data from the entire lifecycle of Jinhua pigs to construct a multi-factor relationship network, accurately extracts key transmission paths and dynamic interaction models affecting meat quality formation, and systematically quantifies the nonlinear antagonistic relationship between the tendency for back fat deposition and intramuscular fat accumulation under the unique physiological mechanisms of Jinhua pigs.

Description

Technical Field

[0001] This invention relates to the field of intelligent pig farming technology, and more specifically, to a method and system for improving the quality of Jinhua pork based on data mining. Background Technology

[0002] With the increasing market demand for high-quality meat products, Jinhua pigs, with their unique meat flavor, early maturity and easy fattening, abundant back fat and thin subcutaneous fat, have become an important resource for making Jinhua ham. However, this characteristic makes nutrient allocation control during fattening particularly sensitive: fat tends to be excessively deposited in the viscera (back fat), while the ideal intramuscular fat content is difficult to consistently achieve. Current mainstream meat quality improvement technologies have shortcomings: firstly, most are developed based on foreign lean-type pig breed systems, failing to adapt to the unique fat metabolism pathway and energy allocation mechanism of Jinhua pigs, ignoring the inherent contradiction of "back fat deposition tendency - intramuscular fat competition"; secondly, they often analyze the dynamic effects of genetics, nutritional timing, and environmental factors on directional fat deposition in isolation, lacking the ability to quantify the nonlinear antagonistic relationship between visceral fat accumulation and the formation of high-quality muscle fat. This results in existing strategies being unable to simultaneously control excessive back fat development and precisely improve intramuscular fat deposition, restricting the improvement of the core economic traits of Jinhua pork.

[0003] Based on the shortcomings of the existing technologies, there is an urgent need for a method and system for improving the quality of Jinhua pork based on data mining. Summary of the Invention

[0004] The purpose of this invention is to provide a data mining-based method for improving the quality of Jinhua pork, thereby addressing the aforementioned problems. To achieve this objective, the technical solution adopted by this invention is as follows:

[0005] Firstly, this application provides a method for improving the quality of Jinhua pork based on data mining, including:

[0006] Obtain the original dataset, which includes time-series data of physiological parameters of individual Jinhua pigs during their growth process, genetic marker locus data, environmental sensor data, feeding record event data, and meat quality evaluation sample data;

[0007] Data fusion is performed based on the original dataset, and an entity-relationship network topology is constructed by calculating the strength of multi-dimensional implicit associations between entities to obtain a multimodal association graph.

[0008] Based on the multimodal association map, potential influence paths are extracted to obtain a set of influence paths;

[0009] Based on the set of influence paths, interaction effect modeling is performed, and the synergistic effect of latent variables is quantified by decomposing the nonlinear coupling relationship between paths to obtain a dynamic effect model;

[0010] Based on the dynamic action model, the source of meat quality defects is traced. By analyzing the gradient sensitivity of target parameters to system variables and simulating pseudo-intervention response, a set of defect driving factors is obtained.

[0011] An optimization strategy is generated based on the set of defect driving factors to obtain a meat quality improvement decision scheme.

[0012] Secondly, this application also provides a data mining-based system for improving the quality of Jinhua pork, including:

[0013] The acquisition module is used to acquire the raw dataset, which includes time-series data of physiological parameters of individual Jinhua pigs during their growth process, genetic marker locus data, environmental sensor data, feeding record event data, and meat quality evaluation sample data.

[0014] The fusion module is used to perform data fusion based on the original dataset, construct an entity-relationship network topology by calculating the strength of multi-dimensional implicit associations between entities, and obtain a multimodal association graph.

[0015] The extraction module is used to extract potential influence paths based on the multimodal association map to obtain a set of influence paths;

[0016] The modeling module is used to model the interaction effect based on the set of influence paths, and to quantify the synergistic effect of latent variables by decomposing the nonlinear coupling relationship between paths to obtain a dynamic action model.

[0017] The traceability module is used to trace the source of meat quality defects according to the dynamic action model. By analyzing the gradient sensitivity of the target parameters to the system variables, it simulates the pseudo-intervention response and obtains the set of defect driving factors.

[0018] The generation module is used to generate optimization strategies based on the set of defect driving factors to obtain a meat quality improvement decision scheme.

[0019] The beneficial effects of this invention are as follows:

[0020] This invention constructs a multi-factor relationship network by integrating multi-source data from the entire life cycle of Jinhua pigs, and accurately extracts key transmission pathways and dynamic interaction models affecting meat quality formation. It systematically quantifies the nonlinear antagonistic relationship between visceral fat deposition tendency and intramuscular fat accumulation under the unique physiological mechanism of Jinhua pigs, overcoming the limitations of existing technologies in analyzing the dynamic coupling effect of multiple factors. Furthermore, based on model-driven meat quality defect tracing, it accurately locates core regulatory nodes and generates optimized schemes that directly guide feed formulation and feeding management adjustments. This achieves the goal of effectively controlling excessive accumulation of visceral fat while directionally increasing intramuscular fat content, significantly improving the uniformity of core commercial traits of Jinhua pork and its product competitiveness. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram of the process for improving the quality of Jinhua pork based on data mining, as described in an embodiment of the present invention.

[0023] Figure 2 This is a schematic diagram of the Jinhua pork quality improvement system based on data mining as described in this embodiment of the invention;

[0024] Figure 3 This is a schematic diagram of the structure of the data mining-based Jinhua pork quality improvement equipment described in this embodiment of the invention.

[0025] The diagram is labeled as follows: 800, a data mining-based device for improving the quality of pork in Jinhua; 801, processor; 802, memory; 803, multimedia component; 804, I / O interface; 805, communication component; 901, acquisition module; 902, fusion module; 903, extraction module; 904, modeling module; 905, traceability module; 906, generation module. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of 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, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0027] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0028] Example 1:

[0029] This embodiment provides a method for improving the quality of Jinhua pork based on data mining.

[0030] See Figure 1 The figure shows that the method includes steps S100 to S600.

[0031] Step S100: Obtain the original dataset, which includes time series data of physiological parameters of individual Jinhua pigs during their growth process, genetic marker locus data, environmental sensor data, feeding record event data, and meat quality evaluation sample data.

[0032] It is understandable that meat quality is the product of long-term interaction among multiple factors, including genetics, environment, feeding, physiological dynamics, and post-slaughter outcomes. This step simultaneously collects key data that characterizes its unique growth patterns: physiological change curves (reflecting growth and development stages), genetic markers (affecting fat metabolism), fluctuations in the pig house environment (inducing stress), management events such as feeding / immunization (directly affecting nutrition), and final meat sample evaluation (defect markers). Data acquisition in this step relies on the seamless integration of automated equipment and production processes: wearable sensors worn on pigs automatically generate time series of physiological parameters; environmental data such as temperature, humidity, and ammonia concentration are collected in real time through an IoT sensor network within the pig house; feeding record events are generated by feeders scanning codes in the digital management system or manually entering data to form structured logs; genetic marker locus data are obtained through gene chip or sequencing analysis of pig ear tissue samples in the laboratory; and meat quality evaluation sample data comes from laboratory physicochemical testing and human sensory evaluation of specific parts after slaughter.

[0033] Step S200: Perform data fusion based on the original dataset, construct the entity-relationship network topology by calculating the strength of multi-dimensional implicit associations between entities, and obtain a multimodal association map;

[0034] It's important to note that the breakthrough in data fusion and graph construction lies in integrating previously fragmented data silos using an "entity-relationship network." Traditional analysis often processes genetic data, environmental records, and production data separately, neglecting the complex collaborative or antagonistic relationships between them. This step, through in-depth calculation of the multidimensional implicit correlation strength between gene loci and growth indicators, environmental sensor readings and feeding record events, and management actions and meat quality results, constructs a topological structure. This accurately recreates the real, complex interaction scenarios in the Jinhua pig farming environment, such as "gene expression being affected by temperature" and "stress events interfering with nutrient absorption efficiency," forming a series of dynamic correlation graphs that provide structural support for subsequent causal relationship mining.

[0035] Step S300: Extract potential influence paths based on the multimodal association map to obtain a set of influence paths;

[0036] Understandably, this step of potential impact path extraction focuses on locating key transmission chains affecting meat quality within a complex network graph. The logic is that not all associations are equally important; it's necessary to identify traceable pathways strongly correlated with specific meat quality issues (such as insufficient intramuscular fat deposition and large variations in tenderness). Addressing the core contradiction in Jinhua pigs—a strong visceral fat generation pathway but a weak intramuscular fat deposition pathway—this step specifically designs an analytical mechanism, such as tracing the potentially suppressed or bypassed pathway of "feed absorption efficiency - skeletal muscle development - intramuscular preadipocyte activation." This step directly serves the specific meat quality goals most urgently needed for optimization in the Jinhua pig breed, rather than broadly screening pathways.

[0037] Step S400: Model the interaction effect based on the set of influence paths, quantify the synergistic effect of latent variables by decomposing the nonlinear coupling relationship between paths, and obtain the dynamic effect model.

[0038] Understandably, traditional linear models cannot explain common phenomena in Jinhua pig farming: increased nutrition may simultaneously stimulate visceral fat deposition while inhibiting intramuscular fat accumulation (non-linear coupling). This step specifically decomposes the interactions at the intersection of multiple pathways (for example, at critical junctures in the rapid growth phase, nutrient supply may simultaneously activate genes for visceral fat accumulation and inhibit signaling pathways for muscle fat precursor differentiation), and quantifies the weight of this "internal friction" (latent variable synergy / competition). This accurately captures the unique nutrient allocation conflict dynamics of Jinhua pigs, establishing a computable model that truly reflects their biological reality.

[0039] Step S500: Based on the dynamic action model, trace the source of meat quality defects. By analyzing the gradient sensitivity of the target parameters to the system variables, simulate the pseudo-intervention response to obtain the set of defect driving factors.

[0040] It's important to note that the role of meat quality defect tracing is to conduct virtual experiments on a dynamic model, enabling low-cost and high-precision localization of the cause. Traditional methods (such as conducting control experiments by actually changing feeding parameters) are costly, time-consuming, and prone to interference. This step analyzes the "gradient sensitivity" of meat quality targets (such as intramuscular fat content) to changes in the entire system (such as gene-environment-nutrition), simulating a "pseudo-intervention" specifically designed for the Jinhua pig model. This doesn't involve actually changing the pig's diet or environment, but rather systematically testing changes at key nodes within the model (such as regulating the expression level of a key lipase), observing the degree of improvement and transmission chain of the target defect (insufficient fat). This specifically addresses the difficulty of tracing the multi-factor coupling and experimental reproducibility of meat quality problems in Jinhua pigs, directly serving subsequent precise regulation.

[0041] Step S600: Generate an optimization strategy based on the set of defect driving factors to obtain a meat quality improvement decision scheme.

[0042] Understandably, strategy optimization is the process of precisely translating the findings of source analysis into actionable aquaculture practices. Its core logic is the recognition that single-point modifications (such as solely increasing intramuscular fat) can be counterproductive in complex systems (e.g., overstimulation leading to thicker basal fat). This step emphasizes the multi-objective balance and dynamic fault tolerance of the strategy. The final output decision plan clearly provides a set of directly executable key operational guidelines, including feed formulations (such as fatty acid ratios), environmental control parameters (such as temperature and humidity ranges), and feeding plans (such as feeding strategies during critical growth periods), ensuring stable and predictable improvement effects.

[0043] Further, step S200 includes steps S210 to S230.

[0044] Step S210: Perform cross-domain temporal alignment processing on the original dataset, and unify the time reference of heterogeneous data through a temporal compensation mechanism to obtain a spatiotemporal calibration dataset.

[0045] Step S220: Perform multidimensional semantic association mining based on the spatiotemporal calibration dataset. By calculating the cross-modal coupling strength of cross-domain combinations of genetic loci-physiological parameters, environmental fluctuations-feeding events, and event sequences-meat quality indicators, obtain the dynamic association matrix between entities.

[0046] Step S230: Based on the dynamic correlation matrix, perform network topology evolution, and generate a sparse connection structure of highly correlated entity edges through iteration to obtain a multimodal correlation graph.

[0047] Understandably, the logical essence of steps S210 to S230 is to gradually extract interpretable patterns of Jinhua pig quality formation from the chaotic raw data: First, step S210 uses cross-domain temporal alignment (such as time calibration between manual feeding records and automatic sensor data) to solve the problem of causal analysis distortion caused by asynchronous collection of multi-source data at the breeding site; then, step S220 deeply mines the cross-modal coupling effect of genetic variations (such as specific fat metabolism sites) on physiological indicators (such as daily weight gain curves) on a unified time benchmark, accurately capturing the nonlinear impact of environmental mutations (such as high temperature) and feeding interventions (such as feed adjustment) on meat quality; finally, step S230 constructs a lightweight topological network through dynamic pruning (such as retaining only key association edges with correlation coefficients > 0.8), condensing the scattered high-dimensional data into a computable knowledge graph pointing to the key quality regulation channels of Jinhua pigs (such as "temperature stress - feeding interruption - intramuscular fat synthesis inhibition"), laying a structured foundation for subsequent traceability modeling.

[0048] Further, step S300 includes steps S310 to S330.

[0049] Step S310: Identify biological system feedback loops based on multimodal association maps. By analyzing the positive and negative feedback directionality and intensity of information flow in the loop topology involving the meat tenderness regulation pathway, a set of potential feedback loops is obtained.

[0050] Step S320: Perform cross-level influence transmission analysis based on the potential feedback loop set. By quantifying the indirect influence strength and path dependence of the temperature and humidity fluctuations in the feeding environment through the composition of the gut microbiota to the rate of intramuscular fat deposition, the potential influence transmission chain set is obtained.

[0051] Step S330: Based on the set of potential influence transmission chains, key cascade nodes are identified, and the set of influence paths is obtained by calculating the sensitivity convergence threshold of simulated path interruption on the final meat juiciness score.

[0052] It should be noted that S310, by analyzing the ring-shaped topology in the graph, specifically identifies the core positive / negative feedback mechanisms that cause fluctuations in the tenderness of Jinhua pigs (such as the continuous reinforcement of appetite suppression signals under high temperature conditions, forming a vicious cycle), thereby revealing the self-reinforcing root cause of meat quality defects. S320 further traces cross-level transmission chains (such as "sudden increase in pig house temperature - intestinal flora disorder - reduced short-chain fatty acid synthesis - inhibited differentiation of intramuscular fat precursor cells"), quantifies the strength of the indirect biological effects of environmental fluctuations on meat quality formation (such as calculating the transmission coefficient that causes a 13% decrease in intramuscular fat deposition efficiency for every 5°C temperature fluctuation), highlighting the specific transmission bottleneck of Jinhua pigs' sensitivity to hot and humid environments. S330, based on pathway interruption simulation (such as blocking acetate synthesis nodes in microbial metabolic pathways), calculates the sensitivity threshold of the final meat juiciness score, and screens out the minimum set of interventions that can significantly improve the specific defects of Jinhua pigs (such as uneven distribution of fat between muscle fibers). These three steps refine the complex network into an operable biological intervention roadmap, ensuring that the source tracing conclusions have both mechanistic explanatory power and practical feasibility. Specifically, the formula for identifying key cascaded nodes is as follows:

[0053]

[0054] in, This represents the set of nodes in the potential influence propagation chain. This indicates the cascade node to be judged, i.e., a specific control unit in the transmission chain; Indicates the juiciness score of the meat; Indicates the intensity of node disturbance; Indicates the sensitivity convergence threshold; Node sensitivity refers to; Represents the set of potential transmission chain nodes In the process, sensitivity indices were selected. Reaching the threshold A set of nodes.

[0055] Further, step S400 includes steps S410 to S430.

[0056] Step S410: Identify the path intersection nodes based on the set of influencing paths. By locating node entities that share multiple paths and have the function of regulating the expression of genes for differentiation of intramuscular fat precursor cells, the set of interaction hub nodes is obtained.

[0057] Step S420: Model the dynamic superposition effect based on the set of interaction hub nodes. By quantifying the temporal interference intensity between the activation level of skeletal muscle energy metabolism pathways and the fat deposition rate during the critical nutrient supply stage of the growth period, the multi-path coupling dynamic equation is obtained.

[0058] Step S430: Decouple the cooperative competition relationship according to the multi-path coupling dynamic equation, and obtain the dynamic action model by decomposing the nonlinear cancellation effect between muscle collagen synthesis and degradation pathways and the weight of implicit regulatory factors.

[0059] Understandably, step S410 addresses the low efficiency of intramuscular fat deposition unique to Jinhua pigs by locating multi-pathway intersection nodes, such as simultaneously regulating "lipid synthesis genes". PPARγThe shared nodes of "and" myostatin (MSTN) address the blind spot of traditional single-pathway analysis, which cannot capture nutrient allocation conflicts (skeletal muscle development competing for adipocyte resources). The core breakthrough lies in identifying pivotal entities (such as AMPK signaling nodes) that participate in the activation of energy metabolism pathways and are influenced by both feeding events (such as high-protein diets) and environmental stress, integrating scattered causal chains into a target set that can simultaneously intervene in multiple meat quality defects. Step S420 targets the critical time window of nutrient allocation during the rapid growth period of Jinhua pigs, quantifying the dynamic inhibitory effect of skeletal muscle glycolysis pathway strength on intramuscular fat deposition rate (e.g., for every 1 unit increase in muscle glucose consumption capacity, adipocyte differentiation efficiency decreases by 0.7 units), and establishing a multi-pathway coupling equation. Its innovation lies in the introduction of time lag factors (such as feed administration). The last two hours represent the period of greatest mutual interference, precisely depicting the unique energy metabolism paradox of Jinhua pigs: "high-energy diet - short-term ATP surplus - inhibition of adipocyte differentiation." Step S430 focuses on resolving the core contradiction of the difficulty in achieving both tenderness and juiciness in Jinhua pigs by decoupling the nonlinear cancellation between collagen synthesis (Col1a1 gene expression) and degradation (MMP1 enzyme activity) pathways (e.g., high summer temperatures increase the degradation rate by 3 times, but simultaneously trigger stress-induced synthetic compensation). The key innovation lies in introducing implicit regulatory factor weight analysis (e.g., calculating that oxytocin receptor expression accounts for 62% of collagen metabolism regulation), revealing the cross-system mechanism of environmental interventions (e.g., spray cooling) indirectly regulating meat structure through neuroendocrine pathways, enabling the model to predict the net effect of complex regulatory strategies. Specifically, the multi-pathway coupling kinetic equation is:

[0060] in, Represents the differential symbol; Indicates intramuscular fat concentration; Indicates the current time; Indicates the efficiency coefficient of fat differentiation; express Nuclear acceptor concentration; Represents the Hill coefficient; express Half-maximal activation concentration; Indicates the rate of uptake of branched-chain amino acids; This represents the muscle synthesis competition constant; Indicates the maximum glycogen reserve capacity of skeletal muscle; Indicates environmental response parameters; Indicates the rate at which the metabolic window closes; This indicates the basic inhibitory efficacy of cortisol per unit. This represents the high-temperature nonlinear gain factor. This indicates the real-time temperature of the pigsty. Indicates plasma cortisol concentration; e represents the natural constant; tfeed This indicates the start time of the day's feed feeding operation.

[0061] Further, step S500 includes steps S510 to S530.

[0062] Step S510: Screen meat defect association paths according to the dynamic action model. By calculating the global sensitivity topology coverage of the target defect quantification state relative to the model path, the defect sensitive path set is obtained.

[0063] Step S520: Perform directional perturbation simulation based on the defect-sensitive path set. By applying a preset step size parameter displacement to the H-FABP gene expression level control point in pigs, quantify the observable response amplitude of the displacement transmitted along the biological path to the meat quality index, and obtain the node perturbation-index response mapping spectrum.

[0064] Step S530: Based on the node disturbance-index response mapping spectrum, the dominant factor is determined. By identifying the control unit whose contribution to the change of the target defect index exceeds the preset multiple of its normal fluctuation standard deviation, the set of defect driving factors is obtained.

[0065] In the meat quality defect tracing stage, steps S510 to S530 construct a closed-loop decision chain from system modeling to precise intervention: firstly, through global sensitivity topology analysis (such as calculating the impact of decreased intramuscular fat content on...) PPARγ The study investigates the combined sensitivity of pathways to heat stress and other pathways, screening biological pathways strongly associated with target defects (such as the "H-FABP gene expression - fatty acid transport efficiency - myolipin deposition" chain) at the whole network scale. Then, targeted virtual perturbations are implemented (e.g., adjusting the expression level of the H-FABP_rs329 gene locus specific to Jinhua pigs in the model by ±15%) to precisely quantify the nonlinear response amplitude of changes in this regulatory point along the metabolic pathway to final meat quality indicators (such as the coefficient of variation of adipocyte diameter distribution), generating a multidimensional response spectrum including dose-response relationships. Finally, based on statistical contribution thresholds, minor interference terms (such as the influence of seasonal temperature fluctuations) are eliminated, identifying the core driving units that can directly explain the main causes of meat quality defects (such as the methylation level of the H-FABP gene promoter), providing molecular targets and quantitative operation windows for subsequent gene selection or nutritional regulation programs.

[0066] Further, step S520 includes steps S521 to S523.

[0067] Step S521: Define the biologically feasible perturbation interval based on the defect-sensitive path set. By matching the phenotypic variation range of the rs329 site of the H-FABP gene in Jinhua pigs, generate a displacement vector sequence that conforms to the actual breeding / nutritional intervention capability using an equal-interval discretization strategy, and obtain a standardized perturbation parameter set.

[0068] This step aims to establish an operational intervention range consistent with the genetic characteristics of Jinhua pigs. The core principle is to transform biological constraints into a discretized mathematical perturbation space by analyzing the phenotypic variation characteristics of the H-FABP gene rs329 locus in the Jinhua pig population (the measured distribution follows a normal distribution with μ=100 AU and σ=15 AU). Specifically, an equally spaced discretization strategy is employed to ensure that each perturbation step size covers both the possible range of genetic variation and matches the capability boundaries of realistic breeding (±15%) and nutritional intervention (±30%). This design avoids meaningless simulations exceeding physiological limits and directly serves subsequent precise quantification. The implemented mathematical model is a discrete perturbation vector generation, and the formula for the discrete perturbation vector is:

[0069] in, Represents the set of standardized perturbation parameters; This represents the average expression level of the H-FABP gene at the rs329 site in Jinhua pigs. represents the relative disturbance intensity; sng(l) represents the disturbance direction; Indicates the amount of disturbance displacement; The baseline expression level; L represents the maximum perturbation order; l represents the discrete perturbation index value.

[0070] Step S522: Perform nonlinear response simulation of biological pathway based on the set of perturbation parameters. By constructing a set of coupled differential equations, the fourth-order Runge-Kutta method is used to numerically integrate to steady state, quantify the dose-effect relationship of gene expression displacement being transmitted along the biological pathway to the final meat quality index, and obtain the node perturbation-dynamic response dataset.

[0071] This step quantifies the chain transmission of gene perturbations to final meat quality indicators by constructing a set of differential equations specific to Jinhua pigs. The core innovation is the introduction of a transport efficiency defect parameter to model the decreased muscle fat deposition efficiency caused by restricted fatty acid transport. A fourth-order Runge-Kutta method is used for 48-hour numerical integration (covering the complete postprandial metabolic cycle) to capture the dose-response relationship under steady-state conditions. This method overcomes the limitation of traditional correlation analysis in resolving transport dynamics. The coupled differential equation set is as follows:

[0072]

[0073] in, Indicates the baseline H-FABP expression level; t0 represents the fatty acid transport rate; t0 represents the metabolic time. Indicates maximum transfer capacity; This parameter represents the defective efficiency of pig transportation in Jinhua. Indicates the rate of attenuation during transport; Indicates the efficiency of myofascial deposition; This indicates the rate of fat degradation.

[0074] Step S523: Perform nonlinear feature analysis based on the node perturbation-dynamic response dataset, generate a continuous response surface function through cubic spline interpolation and calculate the gradient distribution, extract the positive and negative perturbation asymmetry coefficients and sensitive window boundaries of Jinhua pigs, and obtain the perturbation-response mapping spectrum containing biological pathway conduction characteristics.

[0075] This step aims to extract the specific response patterns of Jinhua pigs to gene perturbations. Discrete data is transformed into a continuous response function using cubic spline interpolation, and its first-order gradient ∇M is calculated to reveal the nonlinear sensitivity region. A key finding is a significant asymmetry between positive and negative perturbations in Jinhua pigs (downregulation response intensity is 52% higher than upregulation), and the sensitivity window is determined to be [−10%, +11%). The output is a perturbation-response mapping spectrum containing the function graph, gradient distribution, and numerical parameters, directly supporting breeding decisions. These features provide a basis for avoiding ineffective intervention areas. The formulas for constructing the continuous response function and extracting features are as follows:

[0076]

[0077] in, Represents a continuous response function; Represent the B-spline basis functions; Represents the least squares fitting coefficients; Represents the response gradient; Indicates the asymmetry coefficient; represents the sensitive window; z represents the index value of the B-spline basis function.

[0078] Further, step S600 includes steps S610 to S630.

[0079] Step S610: Define a multi-objective constraint space based on the set of defect driving factors. By setting a decision space with increasing intramuscular fat content as the priority objective, maintaining the lean meat percentage of the carcass as the limiting boundary, and the threshold of muscle antioxidant enzyme activity change as the biochemical constraint, the Pareto front search domain is obtained.

[0080] Step S620: Based on the Pareto front search domain, generate a set of candidate strategies that satisfy all constraints by simulating the coordinated changes of genetic and environmental regulatory variables within their physiologically feasible range.

[0081] Step S630: Perform dynamic robustness verification based on the candidate strategy set. By calculating the expected protection of key meat quality indicators under the preset fattening cycle fluctuation range of the candidate strategies, a meat quality improvement decision scheme is obtained. The meat quality improvement decision scheme includes a feed formula adjustment table, aquaculture environment control standards, a phased feeding plan, and meat quality compliance requirements.

[0082] In the meat quality improvement decision generation stage, steps S610 to S630 construct a closed-loop framework for multi-objective optimization and robustness verification: First, a multi-objective constraint space is defined based on the set of defect driving factors (such as H-FABP gene expression level and heat stress response nodes). This is achieved by setting increasing intramuscular fat content (target > 4.2 mg / g) as the priority objective, carcass lean meat percentage (constraint > 58%) as the boundary constraint, and muscle antioxidant enzyme activity (SOD > 135) as the limit. Using U / mgprot as the biochemical threshold, a three-dimensional Pareto front search domain that balances quality and production performance was constructed. Within this search domain, the synergistic changes of genetic regulation (e.g., the intensity of rs329 site selection) and environmental variables (e.g., the summer pig house temperature control range) within their physiologically feasible range were simulated. A non-dominated sorting genetic algorithm was used to generate a set of candidate strategies that simultaneously satisfy meat quality improvement, lean meat maintenance, and oxidative stability. Finally, through dynamic robustness verification, the expected guarantee degree (e.g., >90% probability of achieving the target) of each candidate strategy under fluctuations in the fattening cycle (e.g., ±5% deviation in feed composition, ±3℃ fluctuation in temperature) for key meat quality indicators (intracranial fat, shear force, drip loss) was calculated. An executable decision scheme was output, including a precise feed formulation adjustment table (linoleic acid ratio ≥1.8%), environmental control standards (temperature reduction ≥4℃ during high-temperature periods), a phased feeding plan (nighttime feeding during the 60-90kg weight stage), and meat quality achievement requirements (marbling score ≥3.5 grade), ensuring the stable realization of the improvement effect in the actual breeding scenario of Jinhua pigs.

[0083] Example 2:

[0084] like Figure 2 As shown, this embodiment provides a data mining-based system for improving the quality of Jinhua pork. The system includes:

[0085] The acquisition module 901 is used to acquire the raw dataset, which includes time series data of physiological parameters of individual Jinhua pigs during their growth process, genetic marker locus data, environmental sensor data, feeding record event data, and meat quality evaluation sample data.

[0086] The fusion module 902 is used to perform data fusion based on the original dataset, construct the entity-relationship network topology by calculating the strength of multi-dimensional implicit associations between entities, and obtain a multimodal association graph.

[0087] Extraction module 903 is used to extract potential impact paths based on the multimodal association map to obtain a set of impact paths;

[0088] Modeling module 904 is used to model interaction effects based on the set of influence paths. It quantifies the synergistic effect of latent variables by decomposing the nonlinear coupling relationship between paths, and obtains a dynamic action model.

[0089] The traceability module 905 is used to trace the source of meat quality defects according to the dynamic action model. It simulates the pseudo-intervention response by analyzing the gradient sensitivity of the target parameters to the system variables, and obtains the set of defect driving factors.

[0090] The generation module 906 is used to generate optimization strategies based on the set of defect driving factors to obtain meat quality improvement decision schemes.

[0091] In one specific embodiment disclosed in this application, the fusion module 902 includes:

[0092] The first fusion unit is used to perform cross-domain temporal alignment processing based on the original dataset, and to unify the time reference of heterogeneous data through a temporal compensation mechanism to obtain a spatiotemporal calibration dataset.

[0093] The second fusion unit is used to perform multidimensional semantic association mining based on the spatiotemporal calibration dataset. By calculating the cross-modal coupling strength of cross-domain combinations of genetic loci-physiological parameters, environmental fluctuations-feeding events, and event sequences-meat quality indicators, the dynamic association matrix between entities is obtained.

[0094] The third fusion unit is used to perform network topology evolution based on the dynamic correlation matrix. By iteratively generating a sparse connection structure of highly correlated entity edges, a multimodal correlation graph is obtained.

[0095] In one specific embodiment disclosed in this application, the extraction module 903 includes:

[0096] The first extraction unit is used to identify biological system feedback loops based on multimodal association maps. By analyzing the positive and negative feedback directionality and intensity of information flow in the loop topology involving the meat tenderness regulation pathway, a set of potential feedback loops is obtained.

[0097] The second extraction unit is used to perform cross-level influence transmission analysis based on the set of potential feedback loops. By quantifying the indirect influence strength and path dependence of the fluctuation of temperature and humidity in the feeding environment through the composition of the gut microbiota to the rate of intramuscular fat deposition, the set of potential influence transmission chains is obtained.

[0098] The third extraction unit is used to identify key cascade nodes based on the set of potential influence transmission chains. By calculating the sensitivity convergence threshold of simulated path interruption on the final meat juiciness score, the set of influence paths is obtained.

[0099] In one specific embodiment disclosed in this application, the modeling module 904 includes:

[0100] The first modeling unit is used to identify path intersection nodes based on the set of influencing paths. By locating node entities that share multiple paths and have the function of regulating the expression of genes for differentiation of intramuscular fat precursor cells, the set of interaction hub nodes is obtained.

[0101] The second modeling unit is used to model the dynamic superposition effect based on the set of interaction hub nodes. By quantifying the temporal mutual interference intensity between the activation level of skeletal muscle energy metabolism pathways and the fat deposition rate during the critical nutrient supply stage of the growth period, the multi-path coupling dynamic equation is obtained.

[0102] The third modeling unit is used to decouple the cooperative competition relationship based on the multi-path coupling dynamic equation. By decomposing the nonlinear cancellation effect between muscle collagen synthesis and degradation pathways and the weight of implicit regulatory factors, a dynamic action model is obtained.

[0103] In one specific embodiment disclosed in this application, the traceability module 905 includes:

[0104] The first tracing unit is used to screen the meat quality defect association path according to the dynamic action model. It obtains the defect sensitive path set by calculating the global sensitivity topology coverage of the target defect quantification state relative to the model path.

[0105] The second source tracing unit is used to perform directional perturbation simulation based on the defect-sensitive path set. By applying a preset step size parameter displacement to the H-FABP gene expression level control point in pigs, the observable response amplitude of the displacement transmitted along the biological path to the meat quality index is quantified, and the node perturbation-index response mapping spectrum is obtained.

[0106] The third source tracing unit is used to identify the dominant factors based on the node disturbance-index response mapping spectrum. By identifying the control units whose contribution to the change of the target defect index exceeds a preset multiple of its normal fluctuation standard deviation, the set of defect driving factors is obtained.

[0107] In one specific embodiment disclosed in this application, the generation module 906 includes

[0108] The first generation unit is used to define a multi-objective constraint space based on the set of defect driving factors. By setting a decision space with increasing intramuscular fat content as the priority objective, maintaining the lean meat percentage of the carcass as the limiting boundary, and the threshold of muscle antioxidant enzyme activity change as the biochemical constraint, the Pareto front search domain is obtained.

[0109] The second generation unit is used to generate a set of candidate strategies that satisfy all constraints by simulating the coordinated changes of genetic and environmental regulatory variables within their physiologically feasible range, based on the Pareto front search domain.

[0110] The third generation unit is used to perform dynamic robustness verification based on the candidate strategy set. By calculating the expected protection of key meat quality indicators under the preset fattening cycle fluctuation range of the candidate strategies, a meat quality improvement decision scheme is obtained. The meat quality improvement decision scheme includes a feed formula adjustment table, aquaculture environment control standards, a phased feeding plan, and meat quality compliance requirements.

[0111] Example 3:

[0112] Corresponding to the above method embodiments, this embodiment also provides a Jinhua pork quality improvement device based on data mining. The Jinhua pork quality improvement device based on data mining described below and the Jinhua pork quality improvement method based on data mining described above can be referred to and corresponded to each other.

[0113] Figure 3 This is a block diagram illustrating a data mining-based pork quality improvement device 800 for Jinhua pork, according to an exemplary embodiment. Figure 3 As shown, the Jinhua pork quality improvement device 800 based on data mining may include: a processor 801 and a memory 802. The Jinhua pork quality improvement device 800 based on data mining may also include one or more of the following: a multimedia component 803, an I / O interface 804, and a communication component 805.

[0114] The processor 801 controls the overall operation of the data mining-based Jinhua pork quality improvement device 800 to complete all or part of the steps in the aforementioned data mining-based Jinhua pork quality improvement method. The memory 802 stores various types of data to support the operation of the data mining-based Jinhua pork quality improvement device 800. This data may include, for example, instructions for any application or method operating on the data mining-based Jinhua pork quality improvement device 800, as well as application-related data such as contact data, sent and received messages, images, audio, video, etc. The memory 802 can be implemented using any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touchscreen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signals may be further stored in the memory 802 or transmitted via the communication component 805. The audio component also includes at least one speaker for outputting audio signals. I / O interface 804 provides an interface between processor 801 and other interface modules, such as keyboards, mice, and buttons. These buttons can be virtual or physical. Communication component 805 is used for wired or wireless communication between the data mining-based Jinhua pork quality improvement device 800 and other devices. Wireless communication includes Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, or 4G, or a combination thereof. Therefore, the corresponding communication component 805 may include a Wi-Fi module, a Bluetooth module, and an NFC module.

[0115] In an exemplary embodiment, a data mining-based Jinhua pork quality improvement device 800 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to execute the aforementioned data mining-based Jinhua pork quality improvement method.

[0116] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, which, when executed by a processor, implement the steps of the above-described data mining-based method for improving the quality of Jinhua pork. For example, the computer-readable storage medium may be the memory 802 including the program instructions, which may be executed by a processor 801 of a data mining-based Jinhua pork quality improvement device 800 to complete the above-described data mining-based method for improving the quality of Jinhua pork.

[0117] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for improving the quality of Jinhua pork based on data mining, characterized in that, include: Obtain the original dataset, which includes time-series data of physiological parameters of individual Jinhua pigs during their growth process, genetic marker locus data, environmental sensor data, feeding record event data, and meat quality evaluation sample data; Data fusion is performed based on the original dataset, and an entity-relationship network topology is constructed by calculating the strength of multi-dimensional implicit associations between entities to obtain a multimodal association graph. Based on the multimodal association map, potential influence paths are extracted to obtain a set of influence paths; Based on the set of influence paths, interaction effect modeling is performed, and the synergistic effect of latent variables is quantified by decomposing the nonlinear coupling relationship between paths to obtain a dynamic effect model; Based on the dynamic action model, meat quality defects are traced back to their source. By analyzing the gradient sensitivity of target parameters to system variables, pseudo-intervention responses are simulated to obtain a set of defect driving factors. The pseudo-intervention responses do not actually change the pig's diet or environment, but rather systematically explore changes in key nodes within the model to observe the degree of improvement and transmission chain of the target defects. An optimization strategy is generated based on the set of defect driving factors to obtain a meat quality improvement decision scheme. The extraction of potential influence paths based on the multimodal association map includes: Based on the multimodal association map, biological system feedback loops are identified. By analyzing the positive and negative feedback directionality and intensity of information flow in the loop topology involving the meat tenderness regulation pathway, a set of potential feedback loops is obtained. Based on the set of potential feedback loops, cross-level influence transmission analysis was conducted. By quantifying the indirect influence intensity and path dependence of fluctuations in the feeding environment temperature and humidity through the composition of the gut microbiota to the rate of intramuscular fat deposition, a set of potential influence transmission chains was obtained. The path dependence is a cross-level transmission chain of sudden increase in pig house temperature - gut microbiota disorder - reduced synthesis of short-chain fatty acids - inhibited differentiation of intramuscular fat precursor cells. The cross-level influence transmission analysis based on the set of potential feedback loops includes: Tracing the transmission chain across levels; The intensity of the indirect biological effects of environmental fluctuations on meat formation is quantified, including the calculation of the conduction coefficient by which a 5°C temperature fluctuation leads to a 13% decrease in intramuscular fat deposition efficiency. Highlighting the specific transmission bottleneck of Jinhua pigs' sensitivity to hot and humid environments; Based on the set of potential influence transmission chains, key cascade nodes are identified, and the set of influence paths is obtained by calculating the sensitivity convergence threshold of simulated path interruption on the final meat juiciness score. The interaction effect modeling based on the set of influence paths includes: Based on the set of influence paths, the path intersection nodes are identified, and by locating node entities that share multiple paths and have the function of regulating the expression of genes for differentiation of intramuscular fat precursor cells, the set of interaction hub nodes is obtained. Based on the set of interaction hub nodes, a dynamic superposition effect model is performed. By quantifying the temporal mutual interference intensity between the activation level of skeletal muscle energy metabolism pathways and the fat deposition rate during the critical nutrient supply stage of the growth period, a multi-path coupling dynamic equation is obtained. Based on the multi-path coupling dynamic equation, the cooperative competition relationship is decoupled, and a dynamic action model is obtained by decomposing the nonlinear cancellation effect between muscle collagen synthesis and degradation pathways and the weight of implicit regulatory factors.

2. The method for improving the quality of Jinhua pork based on data mining according to claim 1, characterized in that, Data fusion is performed based on the original dataset, including: Cross-domain temporal alignment processing is performed on the original dataset, and the time reference of heterogeneous data is unified through a temporal compensation mechanism to obtain a spatiotemporal calibration dataset. Multidimensional semantic association mining is performed based on the spatiotemporal calibration dataset. By calculating the cross-modal coupling strength of cross-domain combinations of genetic loci-physiological parameters, environmental fluctuations-feeding events, and event sequences-meat quality indicators, the dynamic association matrix between entities is obtained. Based on the dynamic correlation matrix, network topology evolution is performed, and a sparse connection structure of highly correlated entity edges is generated iteratively to obtain a multimodal correlation graph.

3. The method for improving the quality of Jinhua pork based on data mining according to claim 1, characterized in that, The source tracing of meat quality defects based on the dynamic action model includes: Based on the dynamic action model, the path of meat defect association is screened, and the set of defect sensitive paths is obtained by calculating the global sensitivity topology coverage of the target defect quantification state relative to the model path. Based on the defect-sensitive path set, directional perturbation simulation is performed. By applying a preset step size parameter displacement to the H-FABP gene expression level control point in pigs, the observable response amplitude of the displacement transmitted along the biological path to meat quality indicators is quantified, and the node perturbation-indicator response mapping spectrum is obtained. The baseline expression level of the preset step size parameter is 100 AU, the step size coefficient is 0.05, the displacement direction is distinguished as positive and negative by the sign function sgn, and the maximum perturbation order is 6. Based on the node disturbance-index response mapping spectrum, the dominant factor is determined, and by identifying the control unit whose contribution to the change of the target defect index exceeds a preset multiple of its normal fluctuation standard deviation, the set of defect driving factors is obtained.

4. A data mining-based system for improving the quality of Jinhua pork, characterized in that, include: The acquisition module is used to acquire the raw dataset, which includes time-series data of physiological parameters of individual Jinhua pigs during their growth process, genetic marker locus data, environmental sensor data, feeding record event data, and meat quality evaluation sample data. The fusion module is used to perform data fusion based on the original dataset, construct an entity-relationship network topology by calculating the strength of multi-dimensional implicit associations between entities, and obtain a multimodal association graph. The extraction module is used to extract potential influence paths based on the multimodal association map to obtain a set of influence paths; The modeling module is used to model the interaction effect based on the set of influence paths, and to quantify the synergistic effect of latent variables by decomposing the nonlinear coupling relationship between paths to obtain a dynamic action model. The traceability module is used to trace the source of meat quality defects according to the dynamic action model. By analyzing the gradient sensitivity of the target parameters to the system variables, it simulates the pseudo-intervention response to obtain the set of defect driving factors. The pseudo-intervention response does not actually change the pig's diet or environment, but systematically explores the changes of key nodes within the model to observe the degree of improvement of the target defect and the transmission chain. The generation module is used to generate optimization strategies based on the set of defect driving factors to obtain a meat quality improvement decision scheme. The extraction module includes: The first extraction unit is used to identify biological system feedback loops based on the multimodal association map. By analyzing the positive and negative feedback directionality and intensity of information flow in the loop topology involving the meat tenderness regulation pathway, a set of potential feedback loops is obtained. The second extraction unit is used to perform cross-level influence transmission analysis based on the set of potential feedback loops. By quantifying the indirect influence intensity and path dependence of fluctuations in the feeding environment temperature and humidity through the composition of the intestinal microbial community to the rate of intramuscular fat deposition, a set of potential influence transmission chains is obtained. The path dependence is a cross-level transmission chain of sudden increase in pig house temperature - intestinal microbial community disorder - reduced synthesis of short-chain fatty acids - inhibited differentiation of intramuscular fat precursor cells. The cross-level influence transmission analysis based on the set of potential feedback loops includes: Tracing the transmission chain across levels; The intensity of the indirect biological effects of environmental fluctuations on meat formation is quantified, including the calculation of the conduction coefficient by which a 5°C temperature fluctuation leads to a 13% decrease in intramuscular fat deposition efficiency. Highlighting the specific transmission bottleneck of Jinhua pigs' sensitivity to hot and humid environments; The third extraction unit is used to identify key cascade nodes based on the set of potential influence transmission chains, and to obtain the set of influence paths by calculating the sensitivity convergence threshold of simulated path interruption on the final meat juiciness score. The modeling module includes: The first modeling unit is used to identify path intersection nodes based on the set of influence paths, and obtain the set of interaction hub nodes by locating node entities that share multiple paths and have the function of regulating the expression of genes for differentiation of intramuscular fat precursor cells. The second modeling unit is used to model the dynamic superposition effect based on the set of interaction hub nodes. By quantifying the temporal interference intensity between the activation level of skeletal muscle energy metabolism pathways and the fat deposition rate during the critical nutrient supply stage of the growth period, a multi-path coupling dynamic equation is obtained. The third modeling unit is used to decouple the cooperative competition relationship according to the multi-path coupling dynamic equation, and obtain the dynamic action model by decomposing the nonlinear cancellation effect between muscle collagen synthesis and degradation pathways and the weight of the implicit regulatory factors.

5. The Jinhua pork quality improvement system based on data mining according to claim 4, characterized in that, The fusion module includes: The first fusion unit is used to perform cross-domain temporal alignment processing based on the original dataset, and to unify the time reference of heterogeneous data through a temporal compensation mechanism to obtain a spatiotemporal calibration dataset. The second fusion unit is used to perform multidimensional semantic association mining based on the spatiotemporal calibration dataset. By calculating the cross-modal coupling strength of cross-domain combinations of genetic loci-physiological parameters, environmental fluctuations-feeding events, and event sequences-meat quality indicators, the dynamic association matrix between entities is obtained. The third fusion unit is used to perform network topology evolution based on the dynamic correlation matrix, and obtain a multimodal correlation graph by iteratively generating a sparse connection structure of highly correlated entity edges.

6. The Jinhua pork quality improvement system based on data mining according to claim 4, characterized in that, The tracing module includes: The first tracing unit is used to screen the meat defect association path according to the dynamic action model, and obtain the defect sensitive path set by calculating the global sensitivity topology coverage of the target defect quantification state relative to the model path. The second source tracing unit is used to perform directional perturbation simulation based on the defect-sensitive path set. By applying a preset step size parameter displacement to the H-FABP gene expression level control point in pigs, the observable response amplitude of the displacement transmitted along the biological path to the meat quality index is quantified, and the node perturbation-index response mapping spectrum is obtained. The baseline expression level of the preset step size parameter is 100 AU, the step size coefficient is 0.05, the displacement direction is distinguished by the sign function sgn to be positive and negative, and the maximum perturbation order is 6. The third tracing unit is used to identify the dominant factor based on the node disturbance-index response mapping spectrum. By identifying the control unit whose contribution to the change of the target defect index exceeds a preset multiple of its normal fluctuation standard deviation, a set of defect driving factors is obtained.

Images