A modeling method and system for correlating feces weight changes with giant panda estrus cycles

Abstract

The present invention provides a modeling method and system for associating fecal weight changes with the estrus cycle of giant pandas, relating to the field of giant panda breeding technology. The method comprises the following steps: obtaining basic monitoring data of individual female giant pandas; extracting fiber metabolism characteristics based on daily fecal weight values ​​to obtain a metabolic efficiency sequence for bamboo fiber; constructing a matrix based on the metabolic efficiency sequence and serum progesterone concentration detection values ​​to obtain a progesterone-metabolic coupling matrix; modifying the matrix based on natural seasonal classification labels to obtain an inhibitory factor; identifying the estrus critical state characteristic value based on the progesterone-metabolic coupling matrix and the inhibitory factor; and constructing a model based on the metabolic efficiency sequence, progesterone-metabolic coupling matrix, inhibitory factor, and estrus critical state characteristic value to obtain a dynamic prediction model for the giant panda estrus window. The present invention reduces the miss rate of artificial insemination window identification and significantly improves the breeding efficiency of endangered species.

Description

Technical Field

[0001] The present invention relates to the field of giant panda breeding technology, and in particular to a method and system for modeling the association between feces weight changes and the giant panda's estrus cycle. Background Art

[0002] In the field of endangered species conservation, a key challenge in giant panda captive breeding lies in accurately predicting the panda's estrus window—a physiological phase that lasts only 24-72 hours and exhibits significant individual variability and seasonal fluctuations. Existing technical approaches rely primarily on two types of methods: invasive serum hormone monitoring requires frequent blood sampling to track changes in progesterone concentrations, but this sampling process can easily trigger a stress response, leading to abnormally elevated cortisol levels and disrupting reproductive endocrine homeostasis. Behavioral observation methods manually identify estrus behaviors such as groin rubbing and distinctive calls, but early behavioral signals are subtle and susceptible to interference from captive noise and keeper experience, resulting in a high rate of misjudgment. While the recent emergence of non-invasive fecal monitoring technology mitigates the risk of animal stress, existing systems still face key bottlenecks: a lack of established correlations between fecal weight and metabolic efficiency, a lack of dynamic coupling models for multiple physiological signals, and an inability to capture critical transient features of estrus. These limitations result in insufficient prediction accuracy, and the efficiency of captive breeding is in urgent need of improvement.

[0003] Based on the above shortcomings of the existing technology, there is an urgent need for a modeling method and system for the correlation between feces weight changes and the estrus cycle of giant pandas. Summary of the Invention

[0004] The present invention aims to provide a method for modeling the correlation between fecal weight changes and the estrus cycle of giant pandas to improve the above-mentioned problem. To achieve the above-mentioned object, the technical solution adopted by the present invention is as follows:

[0005] In a first aspect, the present application provides a modeling method for associating feces weight changes with the estrus cycle of giant pandas, comprising:

[0006] Obtain basic monitoring data for individual female giant pandas, including daily fecal weights, serum progesterone concentrations, and natural seasonal classification labels;

[0007] Extracting fiber metabolism characteristics based on the daily feces weighing values ​​to obtain a metabolic efficiency sequence of bamboo fiber;

[0008] Performing matrix construction based on the metabolic efficiency sequence and the serum progesterone concentration detection value to obtain a progesterone-metabolism coupling matrix;

[0009] Corrected according to the natural season classification labels, the inhibition factor was obtained by quantifying the inhibitory effect of seasonal fluctuations of bamboo leaf secondary metabolites on the fermentation rate;

[0010] Identifying the progesterone-metabolism coupling matrix and the inhibitory factor, and obtaining a critical estrus state characteristic value by detecting the derivative jump characteristics and stability characteristics of the modified coupling matrix within a preset time window;

[0011] A model is constructed based on the metabolic efficiency sequence, the progesterone-metabolism coupling matrix, the inhibitory factor and the estrus critical state characteristic value to obtain a giant panda estrus window dynamic prediction model.

[0012] In a second aspect, the present application also provides a modeling system for associating feces weight changes with the estrus cycle of giant pandas, comprising:

[0013] The acquisition module is used to obtain basic monitoring data of individual female giant pandas, including daily fecal weight values, serum progesterone concentration test values, and natural season classification labels;

[0014] An extraction module is used to extract fiber metabolism characteristics based on the daily feces weighing value to obtain a metabolic efficiency sequence of bamboo fiber;

[0015] A construction module, configured to construct a matrix based on the metabolic efficiency sequence and the serum progesterone concentration detection value to obtain a progesterone-metabolism coupling matrix;

[0016] A correction module, configured to perform correction according to the natural season classification label, and obtain an inhibition factor by quantifying the inhibitory effect of seasonal fluctuations of bamboo leaf secondary metabolites on the fermentation rate;

[0017] an identification module, configured to identify the progesterone-metabolism coupling matrix and the inhibitory factor, and obtain a critical estrus state characteristic value by detecting a derivative jump characteristic and a stability characteristic of the modified coupling matrix within a preset time window;

[0018] The output module is used to construct a model based on the metabolic efficiency sequence, the progesterone-metabolism coupling matrix, the inhibitory factor and the estrus critical state characteristic value to obtain a dynamic prediction model for the giant panda estrus window.

[0019] The beneficial effects of the present invention are:

[0020] The present invention establishes a quantitative characterization mechanism for the metabolic efficiency of bamboo fiber, removes the interference of feeding fluctuations, and significantly improves the mapping accuracy of fecal weight changes and core metabolic characteristics; by constructing a progesterone-metabolism coupling matrix and introducing dynamic compensation of seasonal inhibitory factors, it realizes multi-dimensional collaborative modeling of digestive system rhythms and reproductive endocrine signals, and solves the problem of physiological system coupling fracture; based on the dual-feature spatiotemporal detection mechanism of the critical state of estrus, it captures the transient correlation characteristics of the system stability jump and the hormone flux derivative step, and realizes the precise positioning of the estrus critical point; finally, through the time domain alignment and dynamic weight fusion of multi-source physiological signals, a high-confidence estrus window probability prediction model is formed, which reduces the miss rate of artificial insemination window identification and greatly improves the breeding efficiency of endangered species. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0022] Figure 1 A flow chart of a modeling method for associating feces weight changes with the estrous cycle of giant pandas, as described in an embodiment of the present invention;

[0023] Figure 2 This is a schematic diagram of the structure of a modeling system for associating feces weight changes with the estrus cycle of giant pandas, as described in an embodiment of the present invention;

[0024] Figure 3 This is a schematic diagram of the structure of a device for modeling the correlation between feces weight changes and the estrus cycle of giant pandas, as described in an embodiment of the present invention;

[0025] Figure 4 A flow chart for constructing the estrus state distribution model described in an embodiment of the present invention.

[0026] Markings in the figure: 800, a modeling device for the association between feces weight changes and the estrus cycle of giant pandas; 801, processor; 802, memory; 803, multimedia component; 804, I / O interface; 805, communication component; 901, acquisition module; 902, extraction module; 903, construction module; 904, correction module; 905, recognition module; 906, output module. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0028] It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are used only to distinguish the description and should not be understood as indicating or implying relative importance.

[0029] Example 1:

[0030] This embodiment provides a modeling method for associating feces weight changes with the estrus cycle of giant pandas.

[0031] See also Figure 1 , the figure shows that the method includes steps S100 to S600.

[0032] Step S100: Obtain basic monitoring data of individual female giant pandas, including daily fecal weight values, serum progesterone concentration detection values, and natural season classification labels;

[0033] This step collects three key physiological data sets from female giant pandas: daily fecal weights reflect changes in bamboo fiber mass during digestion and serve as a direct indicator of digestive system output; serum progesterone concentrations characterize cyclical fluctuations in reproductive hormones and are a core biomarker of estrus; and natural seasonal classification labels encode environmental factors (spring / summer / autumn / winter) to quantify seasonal differences in bamboo phytochemical composition. This step establishes a foundation for multidimensional physiological monitoring, overcoming the limitations of a single data source.

[0034] Step S200: extracting fiber metabolism characteristics based on daily feces weight values ​​to obtain a metabolic efficiency sequence of bamboo fiber;

[0035] This step separates the behavioral noise of eating from stool weight (such as abnormal weight spikes caused by overeating in winter) to extract the basal metabolic component, which reflects only the digestive efficiency of bamboo fiber. Combined with analysis of colonic rhythmic characteristics, this captures the intensity of cecal microbial fermentation at a biophysical level, ultimately generating a bamboo fiber metabolic efficiency sequence. This step transforms "weight signals" into "metabolic essence," overcoming the bottleneck of traditional stool metrics, which cannot distinguish between physiological processes and behavioral interference.

[0036] Step S300, constructing a matrix based on the metabolic efficiency sequence and the serum progesterone concentration detection value to obtain a progesterone-metabolism coupling matrix;

[0037] Preferably, this step dynamically analyzes the metabolic efficiency sequence and progesterone concentration, revealing a hysteresis pattern of periodic fluctuations in metabolic efficiency after 72 hours of progesterone concentration changes. By constructing a three-dimensional response surface representing the hormone-metabolism dose relationship, where the X-axis represents progesterone concentration, the Y-axis represents time delay, and the Z-axis represents metabolic gain, a progesterone-metabolism coupling matrix is ​​formed, and a quantitative correlation model between the digestive and reproductive systems is established.

[0038] Step S400: Correcting the natural seasonal classification labels and obtaining an inhibition factor by quantifying the inhibitory effect of seasonal fluctuations in bamboo leaf secondary metabolites on the fermentation rate;

[0039] This step analyzes the cyclical fluctuations in bamboo secondary biomass based on natural seasonal classification labels. By quantifying the dose-dependent inhibitory effect of these substances on cecal cellulase activity, the biochemical inhibition is converted into a correction parameter for the coupling matrix—the inhibition factor. The key to this step is to establish a quantitative mapping between environmental seasonal variables and digestive physiological disturbances, addressing model prediction bias caused by seasonal variation in bamboo-derived phytochemical components.

[0040] Step S500: Identify the progesterone-metabolism coupling matrix and the inhibitory factor, and obtain the estrus critical state characteristic value by detecting the derivative jump characteristics and stability characteristics of the modified coupling matrix within a preset time window;

[0041] Preferably, within the seasonally corrected coupling matrix, binary estrous critical state eigenvalues ​​are generated by detecting the spatiotemporal coupling of abrupt first-order derivative jumps in hormone flux (a mathematical representation of hypothalamic GnRH pulse bursts) and abrupt drops in system stability (a Lyapunov exponent < 0.15 for six hours, marking the onset of ovarian feedback) within a 48-hour window. This step enables precise conversion from continuous physiological signals to discrete ovulation events.

[0042] Step S600: constructing a model based on the metabolic efficiency sequence, the progesterone-metabolism coupling matrix, the inhibitory factor and the estrus critical state characteristic value to obtain a giant panda estrus window dynamic prediction model.

[0043] As can be understood, this step eliminates the time difference in multi-source signal acquisition through time axis alignment technology, assigns differentiated contributions to basic metabolic characteristics, hormone response strength, inhibitory factors, and critical signals, and constructs a weighted feature vector. A sliding time window mechanism is then used to continuously output the probability of future estrus, forming a dynamic prediction model for the estrus window. Ultimately, the model uses fecal weight changes as input, analyzes bamboo fiber metabolic efficiency, integrates hormone responses and seasonal interference corrections, and captures critical transient characteristics of reproduction. It ultimately outputs a function mapping system for the high-precision probability distribution of the future estrus window. This achieves a direct mapping from fecal weight changes to estrus cycle probability, providing a quantitative decision-making basis for artificial breeding.

[0044] Furthermore, step S200 includes steps S210 to S230.

[0045] Step S210: Separate the eating frequency fluctuations based on the daily stool weight values, and separate the abnormal eating frequency fluctuation components by moving standard deviation time domain decomposition to obtain a basal metabolic component sequence;

[0046] Specifically, this step uses moving standard deviation time-domain decomposition to remove abnormal weight fluctuations caused by sudden changes in eating frequency (such as overeating during winter), retaining the component sequence reflecting basal metabolic homeostasis. The key goal is to resolve the physiological paradox that "high food intake does not equal high metabolic efficiency" and extract the basal metabolic component sequence synchronized with the digestive cycle.

[0047] Step S220: extract colon rhythm according to the basal metabolic component sequence, extract energy values ​​of characteristic frequency bands of cecal cellulose fermentation by wavelet packet analysis, and obtain a colon rhythm energy sequence;

[0048] Preferably, this step uses the principle of wavelet packet energy focusing to isolate a 2-4 Hz characteristic frequency band (corresponding to the cecal cavity resonance frequency) within the basal metabolic component to quantify cellulose fermentation intensity. By capturing fermentation rhythm biomarkers generated by the giant panda's unique colonic anatomical structure, a colonic rhythm energy sequence is output.

[0049] Step S230 , performing metabolic efficiency integration processing based on the basal metabolic component sequence and the colon rhythm energy sequence, analyzing the synergistic mechanism of colon motility and steady-state metabolism and correcting the influence of biological sample physical property variation, thereby obtaining a bamboo fiber metabolic efficiency sequence.

[0050] This step, understandably, analyzes the biological synergy between colonic mechanical motility (basal metabolic component) and microbial fermentation (rhythmic energy), generating a bamboo fiber metabolic efficiency sequence by correcting for the physical interference of fecal moisture variation on the weight signal. This process deconstructs fecal weight values, which are susceptible to environmental interference, into three biological process characteristics reflecting the giant panda's unique digestive physiology, ultimately yielding a bamboo fiber metabolic efficiency sequence stripped of interference from feeding behavior, seasonal moisture content, and physical properties.

[0051] Furthermore, step S300 includes steps S310 to S330.

[0052] Step S310: quantify the metabolic fluctuation pattern of the digestive tract metabolic state according to the bamboo fiber metabolic efficiency sequence, and obtain the reproductive cycle metabolic characteristic sequence by extracting the metabolic efficiency periodic characteristics and critical point precursor signals;

[0053] It should be noted that this step identifies periodic fluctuation patterns in the bamboo fiber metabolic efficiency sequence, extracting characteristic sharp rise and fall transitions through time window analysis. A sliding peak-valley detection method is used to capture the phase transitions of metabolic efficiency near estrus, identifying the synchronous relationship between digestive metabolic fluctuations and the reproductive cycle. Continuous metabolic fluctuations are quantified into discrete sequences representing the evolution of reproductive stages, establishing a bridge between digestive tract metabolic activity and reproductive physiological status.

[0054] Step S320: Evaluate the hormone lag effect based on the serum progesterone concentration test value, and obtain the progesterone regulation kinetic sequence by identifying the delayed regulation of intestinal function by the progesterone concentration gradient change;

[0055] Based on serum progesterone concentrations, this step conducts a lagged correlation analysis. By defining a fixed delay window to track the relationship between progesterone concentration changes and metabolic responses, a mapping model between the magnitude of hormonal changes and intestinal functional responses was constructed. A saturation function was used to fit the nonlinear effects of progesterone concentration gradients on metabolic regulation, generating a kinetic sequence that quantifies the intensity of hormonal regulation and reveals the delayed regulation of reproductive endocrine function on the digestive system.

[0056] Step S330: Modeling the biological coupling relationship based on the metabolic characteristic sequence of the reproductive cycle and the progesterone regulation dynamics sequence, constructing a hormone-metabolism dose response surface to represent the three-dimensional association, and obtaining a progesterone-metabolism coupling matrix.

[0057] It is understood that this step models the spatiotemporal correlation between the metabolic signature sequence and the hormone dynamics sequence. By establishing a three-dimensional coordinate system and employing surface interpolation techniques to generate a continuous response surface, feature point intensification and encryption are implemented in the progesterone peak region and lag-sensitive regions, ultimately outputting a quantized progesterone-metabolism coupling matrix, achieving the transition from linear to spatially correlated modeling.

[0058] Furthermore, step S400 includes steps S410 to S430.

[0059] Step S410: Dynamically analyze bamboo plant secondary metabolites based on natural seasonal classification labels, and obtain a characteristic sequence of secondary metabolite concentrations by establishing a mapping relationship between seasonal cycles and fluctuations in leaf tannic acid concentrations;

[0060] It is important to note that this step establishes a mapping model based on natural seasonal classification labels and bamboo leaf secondary metabolite concentrations, converting seasonal characteristics into secondary metabolite concentration sequences. By leveraging the physiological patterns of bamboo plants, we quantify the seasonal fluctuations of secondary metabolites (such as tannins) and output a concentration feature sequence that strictly corresponds to the seasonal labels over time. This step avoids the operational complexity of direct chemical testing and enables the quantitative conversion of environmental factors into biochemical interferences.

[0061] Step S420: performing digestion inhibition effect evaluation based on the secondary metabolite concentration characteristic sequence, and obtaining a fermentation inhibition intensity sequence by analyzing the dosage inhibitory effect of tannins on cecal microbial cellulase activity;

[0062] Based on the secondary biomass concentration sequence, this step constructs a dose-inhibition response model. By analyzing the negative correlation between specific secondary biomass concentration and cellulase activity, a nonlinear function is used to characterize the biological principle of "weak inhibition at low concentrations, followed by a sharp increase in inhibition after a high concentration threshold." The output fermentation inhibition intensity sequence directly quantifies the dynamic inhibitory strength of bamboo leaf chemical components on cecal microbial fermentation efficiency. Each value in this sequence represents the theoretical inhibition rate of the secondary biomass on digestive efficiency on that day.

[0063] Step S430: Perform bioregulatory factor conversion processing according to the fermentation inhibition intensity sequence, and obtain a seasonal inhibition factor by converting the inhibition intensity into a digestive system function correction coefficient.

[0064] Preferably, this step converts the fermentation inhibition intensity sequence into correction parameters suitable for the reproductive-metabolism coupling model. Through piecewise linear transformation, the theoretical inhibition rate is mapped to a multiplicative correction coefficient in the range of 0.2-0.9 (inhibition factor = 1 indicates no inhibition at all, 0.2 indicates 80% inhibition). This step establishes a complete link of "vegetation chemical inhibition-digestive function attenuation-model correction coefficient", making seasonal effect compensation both biologically adaptable and computationally feasible. This design converts the biochemical interaction patterns of plants and animals in ecosystems into computable model parameters, significantly improving the robustness of the prediction model in cross-seasonal scenarios.

[0065] Furthermore, step S500 includes steps S510 to S530.

[0066] Step S510: performing matrix compensation based on the progesterone-metabolism coupling matrix and the seasonal inhibition factor, and performing a weighted product operation on each element of the coupling matrix and the inhibition factor to achieve seasonal effect attenuation correction, thereby obtaining a seasonal correction coupling matrix;

[0067] Specifically, this step applies seasonal interference correction to the progesterone-metabolism coupling matrix, incorporating inhibition factors into each matrix element through a weighted product operation to achieve seasonal effect attenuation correction. By establishing a scalar multiplication rule between inhibition factors and matrix elements, the interference of bamboo secondary biomass on the metabolism-hormone relationship is eliminated. This step converts the chemical inhibition effect of vegetation into linear compensation parameters in the model calculation space, and the output seasonally corrected coupling matrix significantly improves cross-seasonal stability.

[0068] Step S520: Dynamic feature extraction is performed based on the seasonally corrected coupling matrix. First-order derivative sequence calculation and standard deviation analysis are performed within a sliding time window to identify the persistent jump interval. The falling position interval of the stability index is also detected to obtain the reproductive characteristic marker sequence.

[0069] It should be noted that this step analyzes the seasonally corrected coupling matrix within a preset time window (adapted to the estrous cycle) and performs dual-channel feature capture: First-order derivative sequences are used to identify sustained spikes exceeding baseline fluctuations (indicating abrupt changes in hormone flux); and simultaneously detect periods of abrupt drops in system stability indicators (reflecting transitions in reproductive system status). By defining thresholds for spike persistence and stability cliff depth, a sequence of reproductive signatures with both temporal location and feature strength is generated.

[0070] Step S530: perform critical state determination based on the reproductive characteristic marker sequence, generate a binary state marker sequence through a dual-feature spatiotemporal overlap detection mechanism, and obtain an estrus critical state characteristic value.

[0071] Specifically, this step applies a dual-feature spatiotemporal coupling analysis to the reproductive signature sequence: when the duration of the jump completely encompasses the stability cliff time point, the binary state flag is activated; otherwise, it is set to zero. Preferably, a spatiotemporal overlap determinator is constructed to output a characteristic value containing only 0 / 1 estrus critical states. This design accurately captures the coordinated triggering events of hormone pulses and system phase transitions, achieving high-precision event labeling.

[0072] Furthermore, step S600 includes steps S610 to S630.

[0073] Step S610: performing multi-source feature integration processing based on the bamboo fiber metabolic efficiency sequence, progesterone-metabolism coupling matrix, seasonal inhibitory factor, and estrus critical state characteristic value, unifying the heterogeneous data to the same time reference through a synchronization mechanism that eliminates time mismatches, and obtaining a time-referenced feature set;

[0074] This step uses a unified time baseline mechanism to perform temporal alignment on four heterogeneous data sets: bamboo fiber metabolic efficiency series, progesterone-metabolic coupling matrix, inhibitory factors, and critical eigenvalues. The core technology employs resampling and interpolation to eliminate differences in data acquisition frequency and construct a feature cube (time point × feature type × feature depth) with strict temporal synchronization. This operation addresses the modeling distortion caused by the "temporal mismatch" of traditional multi-source physiological data, forming a standardized time-based feature set and laying the foundation for collaborative analysis.

[0075] Step S620: Perform dynamic weight configuration processing based on the time-based feature set, construct collaborative prediction features through the correlation between metabolic base signals, hormone response intensity, biological inhibitors and critical signals, and obtain a weighted prediction feature vector;

[0076] As can be understood, this step establishes a differentiated feature contribution allocation mechanism based on the time-based feature set: assigning a core weight to metabolic efficiency (reflecting the baseline state of the digestive system), a moderate weight to hormonal response (characterizing the dynamics of reproductive regulation), a modified weight to inhibitory factors (eliminating seasonal interference), and a weight to critical signal triggers (event-driven factors). Principal component contribution analysis quantifies this weight allocation, generating a weighted predictive feature vector that integrates the four-dimensional features. This transforms the static feature stack into a dynamic, collaborative architecture consistent with physiological logic, enabling adaptive weight matching between the metabolic foundation and reproductive events.

[0077] Step S630: Window period probability sequence modeling is performed based on the weighted prediction feature vector, and an estrus state distribution model in a continuous time dimension is formed through periodic state evolution law extraction and probability mapping mechanism to obtain a giant panda estrus window dynamic prediction model.

[0078] It should be noted that this step uses weighted feature vectors as input and extracts periodic state transition patterns using a sliding time window (7-day width). The estrus probability density is dynamically calculated for features within the window, mapped to probability values ​​in the [0, 1] interval using a sigmoid function, and continuously outputs a probability distribution curve for estrus status over the next 168 hours. The model automatically integrates new data and updates it every 24 hours, generating a dynamic prediction map of the estrus window (horizontal axis = future time, vertical axis = estrus probability) that combines temporal continuity and confidence. This represents a breakthrough in the transition from discrete physiological parameters to continuous reproductive event prediction.

[0079] Specifically, if Figure 4As shown, the model employs a dual-channel hybrid architecture. The temporal convolutional network (TCN) pathway employs eight layers of dilated causal convolutions, with expansion coefficients designed to increase as powers of two. Residual connections prevent vanishing gradients in the deep network, specifically capturing long-range trends with a period of approximately seven days, such as bamboo fiber metabolic efficiency and inhibitory factors. The gated recurrent unit (GRU) pathway employs a three-layer bidirectional structure to learn the short-term temporal dependencies between progesterone responses and critical signals (adapting to the propagation of hormone lag effects within 72 hours). The output features of the two pathways are concatenated and then subjected to dimensionality reduction using a fully connected layer, resulting in a one-dimensional probability output. This architecture incorporates anatomical and physiological adaptation mechanisms: the depth of the TCN pathway precisely matches the giant panda's cellulose metabolic cycle, and the bidirectional flow characteristics of the gated recurrent unit effectively simulate the bidirectional regulation of progesterone's metabolic promotion and post-estrus inhibition. The residual structure and layered dropout techniques enhance robustness to sudden disturbances in the captive environment.

[0080] Training data is constructed using a weighted feature vector as input, centered around the true ovulation time. Probability labels are generated using a Gaussian distribution with a standard deviation of twelve hours, simulating the characteristic biological pattern of the estrus window: a slow rise before ovulation and a steep drop after ovulation. The loss function incorporates three components: a mean squared error term as the basic prediction error metric, a KL divergence term to constrain the output distribution to conform to reproductive physiology, and a peak penalty term to specifically suppress spurious spikes. Training utilizes a cosine annealing learning rate strategy, with early stopping triggered when the validation loss decreases by less than one thousandth over ten consecutive training epochs. A transfer learning paradigm (pre-training on historical data from the Chengdu base followed by fine-tuning on the target individual) is also introduced to effectively address the challenge of small sample sizes.

[0081] The Gaussian distribution label generation formula is:

[0082] ,in, P (t) represents the probability of estrus at time t; t represents the time point at which the probability is to be calculated; Indicates the actual ovulation moment; e represents the natural base; σ represents the standard deviation, which is 12 hours.

[0083] The loss function formula is: ;in, L Represents the loss function; α represents the weight of the mean square error term; MSE represents the mean square error between the predicted value and the label; β represents the weight of the KL divergence term; KL represents the divergence between the predicted distribution and the biological prior; γ represents the weight of the peak penalty term; PeakPenalty represents the pseudo-peak suppression loss term.

[0084] Finally, the verification system establishes three core indicators: peak time error, curve steepness (maximum probability change rate), and confidence (area under the curve three hours before and after ovulation). The clinical deployment requirements are no more than 6 hours, no less than 0.15 per hour, and no less than 0.9, respectively. The deployment mechanism is updated every 24 hours: receiving new physiological data - executing feature updates - sliding window predictions for the next 168 hours. The output decision includes three elements: the optimal operating time window, real-time confidence, and alert level. Through lightweight compression, a single prediction can be completed within three seconds, providing a feasible edge computing solution for wild protected areas. Adversarial training injects ±20% fecal weight noise samples to improve robustness, and designs an attention gating mechanism (the gating signal is an S-type function of the weight matrix, the gated recurrent unit hidden state, and the critical feature splicing vector). When the critical feature is activated, the hormone pathway weight is automatically increased by 37%. The attention gating formula is expressed as:

[0085] Among them, Attention represents dynamic enhanced gating; W c represents the learnable weight matrix; h GRU represents the hidden state vector of the gated recurrent unit path; F critical represents the critical feature activation vector; σ1(·) represents the S-type activation function.

[0086] Example 2:

[0087] like Figure 2 As shown, this embodiment provides a modeling system for associating feces weight changes with the estrus cycle of giant pandas, the system comprising:

[0088] Acquisition module 901 is used to obtain basic monitoring data of individual female giant pandas, including daily fecal weight values, serum progesterone concentration detection values, and natural season classification labels;

[0089] Extraction module 902, for extracting fiber metabolism characteristics based on daily feces weighing values ​​to obtain a metabolic efficiency sequence of bamboo fiber;

[0090] A construction module 903 is used to construct a matrix based on the metabolic efficiency sequence and the serum progesterone concentration detection value to obtain a progesterone-metabolism coupling matrix;

[0091] A correction module 904 is used to perform correction based on the natural seasonal classification labels, and obtain an inhibition factor by quantifying the inhibitory effect of seasonal fluctuations of bamboo leaf secondary metabolites on the fermentation rate;

[0092] Identification module 905 is used to identify the progesterone-metabolism coupling matrix and the inhibitory factor, and obtain the estrus critical state characteristic value by detecting the derivative jump characteristics and stability characteristics of the modified coupling matrix within a preset time window;

[0093] Output module 906 is used to construct a model based on the metabolic efficiency sequence, progesterone-metabolism coupling matrix, inhibitory factors and estrus critical state characteristic values ​​to obtain a dynamic prediction model for the giant panda estrus window.

[0094] In a specific embodiment disclosed in this application, the extraction module 902 includes:

[0095] The first extraction unit is used to separate the fluctuation of eating frequency based on the daily fecal weight value, and separate the abnormal fluctuation component of eating frequency by moving standard deviation time domain decomposition to obtain the basal metabolic component sequence;

[0096] The second extraction unit is used to extract colon rhythm according to the basal metabolic component sequence, extract the energy value of the characteristic frequency band of cecal cellulose fermentation by wavelet packet analysis, and obtain the colon rhythm energy sequence;

[0097] The third extraction unit is used to integrate metabolic efficiency according to the basal metabolic component sequence and the colon rhythm energy sequence. By analyzing the synergistic mechanism of colon motility and steady-state metabolism and correcting the influence of physical property variation of biological samples, the bamboo fiber metabolic efficiency sequence is obtained.

[0098] In a specific embodiment disclosed in this application, the construction module 903 includes:

[0099] The first construction unit is used to quantify the metabolic fluctuation pattern of the digestive tract metabolic state based on the bamboo fiber metabolic efficiency sequence, and obtain the metabolic characteristic sequence of the reproductive cycle by extracting the periodic characteristics of metabolic efficiency and the critical point precursor signal;

[0100] The second building block is used to evaluate the hormone lag effect based on the serum progesterone concentration test value, and to obtain the progesterone regulation kinetic sequence by identifying the delayed regulation of intestinal function by the progesterone concentration gradient change;

[0101] The third construction unit is used to model the biological coupling relationship based on the metabolic characteristic sequence of the reproductive cycle and the progesterone regulation dynamic sequence, and to obtain the progesterone-metabolism coupling matrix by constructing a hormone-metabolism dose-response surface to represent the three-dimensional association.

[0102] In a specific embodiment disclosed in this application, the correction module 904 includes:

[0103] The first correction unit is used to dynamically analyze bamboo plant secondary metabolites based on natural seasonal classification labels. By establishing a mapping relationship between seasonal cycles and fluctuations in leaf tannin concentrations, a characteristic sequence of secondary metabolite concentrations is obtained.

[0104] The second correction unit is used to evaluate the digestion inhibition effect based on the characteristic sequence of secondary metabolite concentrations. By analyzing the dosage inhibitory effect of tannins on the cellulase activity of cecal microorganisms, a fermentation inhibition intensity sequence is obtained.

[0105] The third correction unit is used to perform biological regulation factor conversion processing according to the fermentation inhibition intensity sequence, and obtain the seasonal inhibition factor by converting the inhibition intensity into a digestive system function correction coefficient.

[0106] In a specific embodiment disclosed in this application, the identification module 905 includes:

[0107] The first identification unit is used to perform matrix compensation based on the progesterone-metabolism coupling matrix and the seasonal inhibition factor, and to achieve seasonal effect attenuation correction by performing a weighted product operation on each element of the coupling matrix and the inhibition factor to obtain a seasonal correction coupling matrix;

[0108] The second recognition unit is used to extract dynamic features based on the seasonally corrected coupling matrix, identify the persistent jump interval by performing first-order derivative sequence calculation and standard deviation analysis within the sliding time window, and simultaneously detect the falling position interval of the stability index to obtain the reproductive characteristic marker sequence;

[0109] The third recognition unit is used to determine the critical state according to the reproductive characteristic marker sequence, generate a binary state marker sequence through a dual-feature spatiotemporal overlapping detection mechanism, and obtain the estrus critical state characteristic value.

[0110] In a specific embodiment disclosed in this application, the output module 906 includes:

[0111] The first output unit is used to integrate multi-source features based on the bamboo fiber metabolic efficiency sequence, progesterone-metabolism coupling matrix, seasonal inhibitory factor, and estrus critical state eigenvalues. The heterogeneous data are unified to the same time base through a synchronization mechanism that eliminates time mismatches, thereby obtaining a time-based feature set.

[0112] The second output unit is used to perform dynamic weight configuration processing based on the time-based feature set, construct collaborative prediction features through the correlation between metabolic base signals, hormone response intensity, biological inhibitory factors and critical signals, and obtain a weighted prediction feature vector;

[0113] The third output unit is used to perform window period probability sequence modeling based on the weighted prediction feature vector, and to form an estrus state distribution model in a continuous time dimension through the extraction of periodic state evolution laws and the probability mapping mechanism, thereby obtaining a dynamic prediction model for the giant panda estrus window.

[0114] Example 3:

[0115] Corresponding to the above method embodiment, this embodiment also discloses a modeling device for associating feces weight changes with the estrus cycle of giant pandas. The modeling device for associating feces weight changes with the estrus cycle of giant pandas described below and the modeling method for associating feces weight changes with the estrus cycle of giant pandas described above can be referenced to each other.

[0116] Figure 3 FIG. 8 is a block diagram of a device 800 for modeling the association between feces weight changes and the estrus cycle of a giant panda according to an exemplary embodiment. Figure 3 As shown, the device 800 for modeling the association between feces weight changes and giant panda estrus cycles may include: a processor 801 and a memory 802. The device 800 may also include one or more of a multimedia component 803, an I / O interface 804, and a communication component 805.

[0117] Processor 801 is used to control the overall operation of device 800 for modeling the association between feces weight changes and the panda's estrus cycle, thereby completing all or part of the steps in the aforementioned method for modeling the association between feces weight changes and the panda's estrus cycle. Memory 802 is used to store various types of data to support the operation of device 800. This data may include, for example, instructions for any application or method operating on device 800, as well as application-related data such as contact information, sent and received messages, images, audio, and video. The memory 802 can be implemented by 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 memory, 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 touch screen, 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 signal 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. The I / O interface 804 provides an interface between the processor 801 and other interface modules, and the above-mentioned other interface modules can be a keyboard, a mouse, buttons, etc. These buttons can be virtual buttons or physical buttons. The communication component 805 is used for wired or wireless communication between the feces weight change and giant panda estrus cycle correlation modeling device 800 and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G or 4G, or a combination of one or more of them, so the corresponding communication component 805 can include: a Wi-Fi module, a Bluetooth module, an NFC module.

[0118] In an exemplary embodiment, a modeling device 800 for associating feces weight changes with the estrus cycle of giant pandas can 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 above-mentioned modeling method for associating feces weight changes with the estrus cycle of giant pandas.

[0119] In another exemplary embodiment, a computer-readable storage medium containing program instructions is also provided. When executed by a processor, the program instructions implement the steps of the aforementioned method for modeling the association between feces weight changes and the giant panda's estrus cycle. For example, the computer-readable storage medium may be the aforementioned memory 802 containing the program instructions. The program instructions may be executed by the processor 801 of the device 800 for modeling the association between feces weight changes and the giant panda's estrus cycle to implement the aforementioned method for modeling the association between feces weight changes and the giant panda's estrus cycle.

[0120] The above description is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed by the present invention, which should be covered by the scope of protection of the present invention.

Claims

1. A modeling method for associating feces weight changes with the estrus cycle of giant pandas, characterized in that: include: Obtain basic monitoring data for individual female giant pandas, including daily fecal weights, serum progesterone concentrations, and natural seasonal classification labels; Extracting fiber metabolism characteristics based on the daily feces weighing values ​​to obtain a metabolic efficiency sequence of bamboo fiber; Performing matrix construction based on the metabolic efficiency sequence and the serum progesterone concentration detection value to obtain a progesterone-metabolism coupling matrix; Corrected according to the natural season classification labels, the inhibition factor was obtained by quantifying the inhibitory effect of seasonal fluctuations of bamboo leaf secondary metabolites on the fermentation rate; Identifying the progesterone-metabolism coupling matrix and the inhibitory factor, and obtaining a critical estrus state characteristic value by detecting the derivative jump characteristics and stability characteristics of the modified coupling matrix within a preset time window; A model is constructed based on the metabolic efficiency sequence, the progesterone-metabolism coupling matrix, the inhibitory factor and the estrus critical state characteristic value to obtain a giant panda estrus window dynamic prediction model.

2. The method for modeling the association between feces weight changes and the estrus cycle of giant pandas according to claim 1, characterized in that: Fiber metabolism characteristics are extracted based on the daily feces weighing values ​​to obtain a bamboo fiber metabolic efficiency sequence, including: Separating the eating fluctuations based on the daily stool weight values, separating the abnormal eating frequency fluctuation components by moving standard deviation time domain decomposition, and obtaining a basal metabolic component sequence; Extracting colon rhythm according to the basal metabolic component sequence, extracting energy values ​​of characteristic frequency bands of cecal cellulose fermentation by wavelet packet analysis, and obtaining a colon rhythm energy sequence; Metabolic efficiency integration processing is performed based on the basal metabolic component sequence and the colon rhythm energy sequence, and the bamboo fiber metabolic efficiency sequence is obtained by analyzing the synergistic mechanism of colon motility and steady-state metabolism and correcting the influence of physical property variation of biological samples.

3. The method for modeling the association between feces weight changes and the estrus cycle of giant pandas according to claim 1, characterized in that: A matrix is ​​constructed based on the metabolic efficiency sequence and the serum progesterone concentration detection value to obtain a progesterone-metabolism coupling matrix, including: The metabolic fluctuation pattern of the digestive tract metabolic state is quantified according to the bamboo fiber metabolic efficiency sequence, and the metabolic characteristic sequence of the reproductive cycle is obtained by extracting the periodic characteristics of the metabolic efficiency and the critical point precursor signal; The hormone lag effect is evaluated based on the serum progesterone concentration test value, and the progesterone regulation kinetic sequence is obtained by identifying the delayed regulation of intestinal function by the progesterone concentration gradient change; A biological coupling relationship model is performed based on the reproductive cycle metabolic characteristic sequence and the progesterone regulation kinetic sequence, and a progesterone-metabolism coupling matrix is ​​obtained by constructing a hormone-metabolism dose response surface to represent the three-dimensional association.

4. The method for modeling the association between feces weight changes and the estrus cycle of giant pandas according to claim 1, characterized in that: According to the natural seasonal classification labels, the inhibitory factors were obtained by quantifying the inhibitory effect of seasonal fluctuations of bamboo leaf secondary metabolites on the fermentation rate, including: Dynamic analysis of bamboo plant secondary metabolites is performed based on the natural seasonal classification labels, and a characteristic sequence of secondary metabolite concentrations is obtained by establishing a mapping relationship between seasonal cycles and fluctuations in leaf tannic acid concentrations; The digestion inhibition effect is evaluated based on the characteristic sequence of secondary metabolite concentrations, and a fermentation inhibition intensity sequence is obtained by analyzing the dosage inhibition effect of tannins on the cellulase activity of cecal microorganisms; The bioregulatory factor conversion process is performed according to the fermentation inhibition intensity sequence, and the seasonal inhibition factor is obtained by converting the inhibition intensity into a digestive system function correction coefficient.

5. The method for modeling the association between feces weight changes and the estrus cycle of giant pandas according to claim 1, characterized in that: Identifying the progesterone-metabolism coupling matrix and the inhibitory factor, and detecting the derivative jump characteristics and stability characteristics of the modified coupling matrix within a preset time window to obtain the estrus critical state characteristic value, including: Performing matrix compensation based on the progesterone-metabolism coupling matrix and the seasonal inhibition factor, and performing a weighted product operation on each element of the coupling matrix and the inhibition factor to achieve seasonal effect attenuation correction, thereby obtaining a seasonal correction coupling matrix; Dynamic feature extraction is performed based on the seasonally corrected coupling matrix, and the persistent jump interval is identified by performing first-order derivative sequence calculation and standard deviation analysis within a sliding time window, while the falling position interval of the stability index is detected to obtain a reproductive feature marker sequence; The critical state is determined according to the reproductive characteristic marker sequence, a binary state marker sequence is generated through a dual-feature spatiotemporal overlapping detection mechanism, and an estrus critical state characteristic value is obtained.

6. A modeling system for associating feces weight changes with the estrus cycle of giant pandas, characterized in that: include: The acquisition module is used to obtain basic monitoring data of individual female giant pandas, including daily fecal weight values, serum progesterone concentration test values, and natural season classification labels; An extraction module is used to extract fiber metabolism characteristics based on the daily feces weighing value to obtain a metabolic efficiency sequence of bamboo fiber; A construction module, configured to construct a matrix based on the metabolic efficiency sequence and the serum progesterone concentration detection value to obtain a progesterone-metabolism coupling matrix; A correction module, configured to perform correction according to the natural season classification label, and obtain an inhibition factor by quantifying the inhibitory effect of seasonal fluctuations of bamboo leaf secondary metabolites on the fermentation rate; an identification module, configured to identify the progesterone-metabolism coupling matrix and the inhibitory factor, and obtain a critical estrus state characteristic value by detecting a derivative jump characteristic and a stability characteristic of the modified coupling matrix within a preset time window; The output module is used to construct a model based on the metabolic efficiency sequence, the progesterone-metabolism coupling matrix, the inhibitory factor and the estrus critical state characteristic value to obtain a dynamic prediction model for the giant panda estrus window.

7. The feces weight change and giant panda estrus cycle correlation modeling system according to claim 6, characterized in that: The extraction module includes: a first extraction unit, configured to separate the eating frequency fluctuations based on the daily stool weight values, and separate the abnormal eating frequency fluctuation components by moving standard deviation time domain decomposition to obtain a basal metabolic component sequence; A second extraction unit is configured to extract colon rhythm according to the basal metabolic component sequence, extract energy values ​​of characteristic frequency bands of cecal cellulose fermentation by wavelet packet analysis, and obtain a colon rhythm energy sequence; The third extraction unit is used to perform metabolic efficiency integration processing based on the basal metabolic component sequence and the colon rhythm energy sequence, and obtain the bamboo fiber metabolic efficiency sequence by analyzing the synergistic mechanism of colon motility and steady-state metabolism and correcting the influence of physical property variation of biological samples.

8. The feces weight change and giant panda estrus cycle correlation modeling system according to claim 6, characterized in that: The building blocks include: The first construction unit is used to quantify the metabolic fluctuation pattern of the digestive tract metabolic state according to the bamboo fiber metabolic efficiency sequence, and obtain the reproductive cycle metabolic characteristic sequence by extracting the metabolic efficiency periodic characteristics and critical point precursor signals; The second construction unit is used to evaluate the hormone lag effect based on the serum progesterone concentration detection value, and obtain the progesterone regulation kinetic sequence by identifying the delayed regulation law of the progesterone concentration gradient change on intestinal function; The third construction unit is used to model the biological coupling relationship based on the reproductive cycle metabolic characteristic sequence and the progesterone regulation dynamics sequence, and to obtain a progesterone-metabolism coupling matrix by constructing a hormone-metabolism dose response surface to represent the three-dimensional association.

9. The feces weight change and giant panda estrus cycle correlation modeling system according to claim 6, characterized in that: The correction module includes: A first correction unit is used to dynamically analyze bamboo plant secondary metabolites based on the natural season classification labels, and obtain a characteristic sequence of secondary metabolite concentrations by establishing a mapping relationship between seasonal cycles and fluctuations in leaf tannic acid concentrations; The second correction unit is used to perform digestion inhibition effect evaluation processing based on the secondary metabolite concentration characteristic sequence, and obtain a fermentation inhibition intensity sequence by analyzing the dosage inhibition effect of tannins on the cellulase activity of cecal microorganisms; The third correction unit is used to perform biological regulation factor conversion processing according to the fermentation inhibition intensity sequence, and obtain a seasonal inhibition factor by converting the inhibition intensity into a digestive system function correction coefficient.

10. The feces weight change and giant panda estrus cycle correlation modeling system according to claim 6, characterized in that: The identification module includes: a first identification unit, configured to perform matrix compensation based on the progesterone-metabolism coupling matrix and the seasonal inhibition factor, and to achieve seasonal effect attenuation correction by performing a weighted product operation on each element of the coupling matrix and the inhibition factor, thereby obtaining a seasonal correction coupling matrix; A second identification unit is configured to extract dynamic features based on the seasonally corrected coupling matrix, identify persistent jump intervals by performing first-order derivative sequence calculation and standard deviation analysis within a sliding time window, and simultaneously detect the falling position interval of the stability index to obtain a reproductive characteristic marker sequence; The third recognition unit is used to perform critical state judgment based on the reproductive characteristic marker sequence, generate a binary state marker sequence through a dual-feature spatiotemporal overlapping detection mechanism, and obtain an estrus critical state characteristic value.

Images