Environment state analysis method and system based on smart agriculture

Abstract

The invention relates to the technical field of intelligent agriculture, and discloses an environment state analysis method and system based on intelligent agriculture, and the method comprises the steps: constructing a line belt hydraulic network reflecting the ridge and ditch space connectivity based on field multi-point monitoring data, and generating a topology operator; performing characteristic decomposition on the operator to obtain a hydraulic fluctuation mode, and dividing the mode into a row-direction dominant mode set and a transverse dominant mode set; the time response frequency of each mode is extracted, and the multi-time scale hydraulic conduction wave velocity is calculated; respectively calculating row-direction and transverse conduction dispersion degrees based on statistical differences of wave velocities in different frequency bands; constructing an asymmetric feature by using the ratio of the two; and further calculating an environmental structure deviation index of the asymmetric feature relative to the health reference, and comparing the environmental structure deviation index with a preset threshold value to output an environmental state judgment result. According to the method, the multi-dimensional fluctuation fingerprint features are utilized, and accurate recognition and anomaly diagnosis of the structural integrity of the ridge and furrow film mulching field row belt hydraulic channel are achieved.

Description

Technical Field

[0001] This invention relates to the field of smart agriculture technology, and more specifically, to a method and system for environmental status analysis based on smart agriculture. Background Technology

[0002] Modern smart agriculture widely adopts row and strip cultivation models such as furrow mulching to optimize water and fertilizer use efficiency. The accompanying Internet of Things (IoT) environmental monitoring systems typically use multiple sensor devices deployed in the field according to certain rules to collect real-time spatiotemporal environmental data such as soil moisture and electrical conductivity. This has become a fundamental technical means to guide precision irrigation decisions and agricultural management. Existing multi-point environmental data analysis methods mainly focus on the statistical analysis of absolute values. For example, they identify drought or waterlogging by using threshold determination of single-point time series data, or use spatial interpolation algorithms to generate moisture content distribution maps to observe regional differences in dryness and wetness. These methods primarily attribute fluctuations in monitoring data to external hydrological and meteorological factors such as rainfall, irrigation volume, or evaporation.

[0003] However, this traditional method, which only focuses on the level of moisture, lacks the ability to perceive the integrity of the hydraulic channel structure within the field. It cannot distinguish from the overall fluctuations in environmental data whether they are non-structural fluctuations caused by natural changes in dryness and wetness, or structural anomalies caused by the degradation of the row-guided flow structure, such as ridge collapse, ditch blockage, and large-scale damage to the mulch film. As a result, when the function of agricultural infrastructure is damaged, the system often can only give a superficial report of moisture anomalies, but cannot identify deep structural faults, making it difficult to meet the needs of high-standard farmland for automated diagnosis of the health status of facilities. Summary of the Invention

[0004] This invention provides an environmental status analysis method and system based on smart agriculture, which solves the technical problems mentioned in the background.

[0005] The first aspect is the environmental status analysis method based on smart agriculture, including:

[0006] Based on multi-point field monitoring data, a row-zone hydraulic network reflecting the spatial connectivity of furrows is constructed, and corresponding row-zone hydraulic topology operators are generated. The row-zone hydraulic topology operators are eigenvalued to obtain hydraulic fluctuation modes describing water distribution patterns. Based on the spatial projection direction, these hydraulic fluctuation modes are divided into a row-direction dominant mode set propagating along the planting row and a transverse dominant mode set propagating across the planting row. The time response frequencies of each hydraulic fluctuation mode are extracted, and multi-timescale hydraulic transmission velocities are calculated in conjunction with the corresponding spatial frequencies. Based on the statistical differences of the multi-timescale hydraulic transmission velocities in different frequency bands, row-direction transmission dispersion and transverse transmission dispersion are calculated respectively. The asymmetric characteristics of the transverse transmission dispersion and the row-direction transmission dispersion are determined. The difference between the asymmetric characteristics and the preset health benchmark characteristics is calculated to obtain the environmental structure deviation index. The environmental structure deviation index is compared with a preset structural anomaly threshold to obtain the environmental state analysis results.

[0007] Secondly, an environmental status analysis system based on smart agriculture, applied to any of the aforementioned environmental status analysis methods based on smart agriculture, includes:

[0008] The topology operator generation module is used to construct a row hydraulic network reflecting the spatial connectivity of furrows based on multi-point field monitoring data, and to generate the corresponding row hydraulic topology operators.

[0009] The mode decomposition and classification module is used to perform feature decomposition on the row hydraulic topology operator to obtain hydraulic wave modes describing the water distribution pattern, and to divide the hydraulic wave modes into a row-direction dominant mode set that propagates along the planting row and a lateral dominant mode set that propagates across the planting row based on the spatial projection direction; to extract the time response frequency of each hydraulic wave mode, and to calculate the multi-timescale hydraulic propagation wave velocity in combination with the corresponding spatial frequency;

[0010] The dispersion calculation module is used to calculate the row-direction dispersion and the lateral-direction dispersion respectively based on the statistical differences of the hydraulically conducted wave velocity at the multi-time scale in different frequency bands.

[0011] An asymmetric feature extraction module is used to determine the asymmetric features of the transverse conduction dispersion and the row conduction dispersion.

[0012] The index calculation module is used to calculate the difference between the asymmetric feature and the preset health benchmark feature to obtain the environmental structure deviation index.

[0013] The state determination module is used to compare the environmental structure deviation index with a preset structural anomaly threshold to obtain the environmental state analysis result.

[0014] The beneficial effects of this invention are as follows: This invention effectively solves the technical challenge of existing agricultural monitoring technologies in distinguishing between unstructured overall wet-dry fluctuations and structural degradation of row hydraulic channels. By exploring the dynamic response differences in water conduction within a field in different directions, this invention can detect hidden structural damage (such as collapse, blockage, or breakage) in the furrow system (e.g., ridges, ditches, mulch). This significantly improves the accuracy of smart agriculture systems in diagnosing the health status of farmland infrastructure, avoids false alarms caused by normal rainfall or irrigation, and provides a reliable decision-making basis for the precise maintenance of anisotropic flow guidance functions in fields. Attached Figure Description

[0015] Figure 1 This is a flowchart of the environmental status analysis method based on smart agriculture of the present invention;

[0016] Figure 2 This is a block diagram of the environmental status analysis system based on smart agriculture of the present invention. Detailed Implementation

[0017] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0018] Example 1: As Figure 1 As shown, the environmental status analysis method based on smart agriculture includes:

[0019] Based on multi-point field monitoring data, a row-zone hydraulic network reflecting the spatial connectivity of furrows is constructed, and corresponding row-zone hydraulic topology operators are generated. The row-zone hydraulic topology operators are eigenvalued to obtain hydraulic fluctuation modes describing water distribution patterns. Based on the spatial projection direction, these hydraulic fluctuation modes are divided into a row-direction dominant mode set propagating along the planting row and a transverse dominant mode set propagating across the planting row. The time response frequencies of each hydraulic fluctuation mode are extracted, and multi-timescale hydraulic transmission velocities are calculated in conjunction with the corresponding spatial frequencies. Based on the statistical differences of the multi-timescale hydraulic transmission velocities in different frequency bands, row-direction transmission dispersion and transverse transmission dispersion are calculated respectively. The asymmetric characteristics of the transverse transmission dispersion and the row-direction transmission dispersion are determined. The difference between the asymmetric characteristics and the preset health benchmark characteristics is calculated to obtain the environmental structure deviation index. The environmental structure deviation index is compared with a preset structural anomaly threshold to obtain the environmental state analysis results.

[0020] It is important to note that the process of constructing a row-zone hydraulic network reflecting the spatial connectivity of furrows and generating corresponding row-zone hydraulic topology operators is primarily based on a deep understanding of the characteristics and hydraulic response mechanisms of field row structures. Specifically, mulched furrow fields are not homogeneous media; they exhibit significant structural anisotropy. Along the planting row direction, they form hydraulically dominant channels similar to waveguides, while across rows, they exhibit highly damped diffuse media. Distance alone cannot fully characterize this complex connectivity because adjacent but hydraulically blocked structural degradations (such as collapses or blockages) may exist within the furrows. Therefore, a dual weighting mechanism of geometric distance and statistical response is employed to construct the graph structure. Geometric distance constrains basic proximity, while statistical correlation activates or suppresses connection weights through feedback from actual data, ensuring that the generated Laplace matrix accurately reflects the effective hydraulic topology within the field at the current moment.

[0021] In a preferred embodiment, the equipment deployed in the field is first obtained. The physical coordinates of the monitoring node are denoted as follows: The coordinate vector of each node is ,in Represents the coordinates along the direction of furrow extension (row direction). This represents the coordinates perpendicular to the furrow direction (row crossing direction). Subsequently, to capture the dynamic response under specific irrigation or rainfall events, the system selects multiple test data points according to a preset time window. Let the... The time window for each event is The window contains At the sampling time, the first The time series data of each node within this window is denoted as follows: ,in Based on this, the system calculates any two nodes. and Pearson correlation coefficient within the event window The calculation formula is as follows: ,in and They are nodes and The mean value within that time window. The correlation coefficient. The degree of synchronization between the two nodes in terms of moisture fluctuation trends was quantified; if the hydraulic channel between them is unobstructed, then... It approaches 1; if the channel is blocked, even if the physical distance is very close, It will also decrease significantly.

[0022] Furthermore, a fully connected weighted adjacency matrix is ​​constructed. To simultaneously reflect the nearest neighbor effect and functional connectivity, matrix elements... Defined as the product of the spatial distance attenuation term and the correlation enhancement term, the specific calculation formula is as follows: When hour, ;when hour, In the formula, Represents a node and The Euclidean distance between them; The preset spatial scale parameter is used to control the range of influence of distance. It is usually set to 1.5 to 2 times the average distance between adjacent sensors to ensure the smoothness of local connections. The relevance weight parameter is used to adjust the strength of statistical similarity correction on graph structure. The smaller the value, the more sensitive the system is to correlation, and the better it highlights the actual connected hydraulic channels. Finally, based on this adjacency matrix... Calculate the degree matrix It is a diagonal matrix, with diagonal elements This leads to the generation of the row-band hydraulic topology operator, i.e., the combined Laplace matrix. Its calculation formula is .

[0023] In some possible embodiments, considering that sensors are distributed in a regular grid in certain large-scale field deployment scenarios, the distance metric can be replaced by Manhattan distance instead of Euclidean distance. This more directly reflects the orthogonal path characteristics along and across ridges. Simultaneously, when calculating statistical correlation, to capture the nonlinear dependence of soil moisture changes, mutual information or Spearman's rank correlation coefficient can be used instead of the Pearson correlation coefficient. In this case, an adjacency matrix is ​​constructed. The weighting function can be adjusted to a threshold-truncation type, for example: only weights that are less than a threshold are considered. And the correlation index is greater than the threshold. hour, ,otherwise The simplified binary graph structure (UnweightedGraph) achieves higher execution efficiency on computationally limited embedded edge computing devices while still retaining the topological features of the row strip structure.

[0024] In some possible embodiments, to address the uneven distribution density of sensor nodes in the field, and to eliminate the influence of differences in node degree distribution on spectral analysis, a normalized Laplacian matrix is ​​used as the row-zone hydraulic topology operator. Specifically, this involves calculating the degree matrix... and adjacency matrix Then, construct the symmetric normalized Laplace matrix. ,in It is the identity matrix. This is the inverse square root of the degree matrix. Alternatively, the Laplace matrix of the random walk can also be used. The eigenvalues ​​of the normalized Laplace matrix are strictly constrained to... Within the range, this characteristic enables the asymmetric features extracted by sensor networks of different fields and sizes to have better comparability and generalization ability, making it particularly suitable for scenarios that require horizontal comparative analysis of fields in multiple different regions.

[0025] It should be noted that the process of performing eigenvalue decomposition on the row-band hydraulic topology operator to obtain hydraulic fluctuation modes describing the water distribution pattern is a key step in the entire environmental state analysis process to achieve the conversion from the spatial domain to the spectral domain. Its core design principle lies in the fact that, similar to the Fourier transform in traditional signal processing, which decomposes a time-domain signal into sine waves of different frequencies, Spectral Graph Theory allows us to decompose water data defined on irregular sensor networks into a series of orthogonal graph modes. The eigenvalues ​​of the Laplace matrix measure the degree of variation or roughness of the corresponding eigenvector (mode) on the graph structure. Smaller eigenvalues ​​correspond to a global distribution pattern that changes gently on the row-band network (such as an overall wetting trend), while larger eigenvalues ​​correspond to local patterns that fluctuate sharply between adjacent nodes (such as localized waterlogging or noise). By decomposing the complex spatiotemporal distribution of field water into a series of independent standing wave components, a mathematical basis is provided for subsequent identification of differences in the dispersive structure between rows and the lateral direction.

[0026] In a preferred embodiment, the system invokes a standard eigenvalue decomposition algorithm (such as the QR algorithm or divide-and-conquer algorithm) from the numerical linear algebra library to decompose the matrix. Perform full diagonalization and solve the characteristic equation. In this formula, For the first 1 eigenvalue, This corresponds to the feature vector. After calculation, the system will obtain... eigenvalues ​​and 1 eigenvector. Because of It is a positive semi-definite matrix, and the system sorts all eigenvalues ​​in non-decreasing order according to their numerical values, that is... Based on this sorting, define a mapping relationship: the sorted feature values... Defined as the spatial frequency, which characterizes the degree of drastic spatial variation in field water resources. A higher value indicates a higher frequency and more drastic variation in the corresponding water distribution pattern in space. The corresponding feature vector... This is defined as the Hydraulic Fluctuation Mode. At this point... Each component This represents the mode in the 1st... The relative amplitudes at each physical node. In matrix form, this process can be represented as... ,in It is an eigenvalue diagonal matrix. It is a transformation matrix composed of orthogonal eigenvectors (i.e., a graph Fourier transform basis), so that each mode is mutually orthogonal.

[0027] In some possible embodiments, for large-scale field sensor networks (e.g., number of nodes) In scenarios with thousands or even more nodes, performing full feature decomposition (with a complexity of approximately...) is necessary. This approach consumes enormous computational resources and time, and in practical environmental analysis, extremely high-frequency modes often correspond to sensor noise rather than effective hydraulic structures. Therefore, a truncated eigenvalue decomposition strategy is adopted. Specifically, the Lanczos iterative algorithm or the Arnoldi method is used to compute only the first... The smallest non-zero eigenvalues ​​and their corresponding eigenvectors (e.g., taking...) For the total number of nodes At this point, the decomposition process no longer aims to recover the complete matrix. Instead of directly capturing row connectivity features, it acquires principal component modes that include the main hydraulic structure information of the field (low-frequency and mid-frequency components). This significantly reduces the computational load while ensuring the capture of row connectivity features.

[0028] In some possible implementations, the input operator is not a standard combinatorial Laplacian matrix, but a symmetric normalized Laplacian matrix. In this case, the meaning of eigenvalue decomposition will be adjusted. Specifically, it involves solving the generalized eigenvalue problem systematically. The equivalent form. The eigenvalues ​​obtained at this point. Distributed in Within the interval. In this case, the eigenvalues This not only represents absolute smoothness but also the normalized variability relative to the degree. This processing method can eliminate spatial frequency estimation bias caused by uneven density of field sensor deployment (i.e., large differences in degree distribution), ensuring better consistency in modal shape and frequency characteristics when comparing hydraulic fluctuation modes in areas with different deployment densities, thereby improving the applicability of the scheme under non-uniform network conditions.

[0029] It should be noted that, based on the spatial projection direction, the hydraulic wave modes are divided into a row-oriented dominant mode set propagating along the planting row and a lateral dominant mode set propagating across the planting row, in order to solve the technical problem of unclear directionality of the graph Laplacian eigenvectors. Specifically, although the eigenvectors This describes a standing wave structure, but it doesn't directly carry the label of advancing along or across furrows. Two orthogonal spatial reference signals (i.e., a simple linear row gradient and a lateral gradient) are constructed using the coordinates of field nodes. The dominant vibration direction of a mode is determined by calculating the similarity (projection) between the mode and these two reference signals. If a mode's correlation with the row coordinate gradient is significantly stronger than its correlation with the lateral gradient, it indicates that the mode primarily captures water fluctuation energy along the furrow direction; conversely, it reflects lateral seepage or blockage effects across furrows.

[0030] In a preferred embodiment, the system first performs... The coordinates of the monitoring nodes are preprocessed. Let the coordinates of the first monitoring node be... The original coordinates of each node are To eliminate the DC component deviation caused by the selection of the coordinate origin in the projection calculation, the system first performs a centering operation: calculating the coordinate mean. and This leads to the mean-free centered coordinates. and Based on this, the system constructs a row-direction coordinate basis vector that reflects the average positional deviation of the entire field. With the horizontal coordinate basis vector These two vectors represent the ideal pure linear variation in the direction of the field and the pure linear variation in the lateral direction, respectively.

[0031] In a preferred embodiment, the system is configured for each hydraulic fluctuation mode. (in Typically, the first mode representing the DC component is skipped, and its normalized inner product (i.e., cosine similarity) with the spatial basis vectors is calculated separately. The formula for calculating the row projection coefficient is: The formula for calculating the lateral projection coefficient is: In the formula, The Euclidean norm of a vector ( norm), Represents the vector dot product. Parameters The absolute value quantifies the mode. The alignment of the peak and trough distribution with the row coordinate axis. Finally, the system executes the classification decision logic: comparing the absolute values ​​of the two projection coefficients. If... This indicates that the modality has a stronger structural character in the row direction, and its indexing... Add to row-dominant mode set Conversely, if Then add it to the lateral dominant mode set. .

[0032] In some possible implementations, this applies to fields that are not oriented due north or south, or have non-rectangular boundaries. Specifically, the coordinate system used to deploy the sensors (such as plane coordinates converted from GPS latitude and longitude) may have an angle with the actual furrow orientation. To address this issue, a principal axis rotation correction is added before constructing the basis vectors. The system utilizes principal component analysis (PCA) or manually input ridge orientation angles to construct the rotation matrix. . The original centered coordinates Transformed into corrected ridge-direction coordinates Then, basis vectors are constructed using the transformed coordinates. and This makes the projection coefficient... The calculations are always relative to the actual agricultural furrow structure, rather than relying on the coordinate system definition of the measuring equipment, thus avoiding modal classification errors caused by misaligned coordinate systems.

[0033] In some possible embodiments, to address the classification ambiguity problem of diagonal modes or mixed modes, specifically, in the early stages of certain complex line band destruction, there may be a large number of mixed modes propagating along the 45-degree direction. and Very close. To improve the confidence of the classification, a significance threshold factor is introduced. (For example The classification logic has been adjusted to: only when... Only then will the modality be included. Only when Only then will the modality be included. For modes in between (i.e.) ), and label them as isotropic or mixed modes.

[0034] It should be noted that the time response frequencies of each hydraulic wave mode are extracted and combined with the corresponding spatial frequencies to calculate the hydraulic propagation wave velocities at multiple time scales, in order to establish a dispersion relation. Specifically, in wave physics, the propagation characteristics of waves are typically determined by frequency. With wavenumber The relationship between them determines (i.e., phase velocity) In a linear hydraulic network, the square root of the Laplace eigenvalue. It is considered as the generalized wavenumber (spatial frequency), while the oscillation frequency of the mode on the time axis This reflects the dynamic response rate of water under this spatial pattern. Because soil is a porous medium, its water conduction depends not only on the instantaneous pressure head but also on slow variables such as soil bulk density and root water uptake. Therefore, multi-timescale analysis must be introduced. By observing the frequency response under different smoothing windows, the rapid gravity infiltration and slow capillary redistribution can be distinguished. By calculating the normalized effective phase velocity, the water conductivity of the field under different spatial patterns can be detected.

[0035] In a preferred embodiment, the system first receives the result calculated in the preceding steps. One hydraulic fluctuation mode and the Multipoint test data matrix within a rainfall or irrigation event window The system performs a modal projection operation, decoupling the spatiotemporal mixed signal into independent modal time coefficient sequences. The calculation formula is as follows: ,in The coefficient Represents the moment Field water distribution patterns and the first The similarity of the modal shapes (i.e., the instantaneous activation intensity of the modality). Subsequently, to introduce multi-timescale characteristics, the system pre-sets a set of timescales. (For example (This can correspond to the number of sampling points for 1 hour, 6 hours, and 12 hours, respectively). For each scale... The system processes the original coefficient sequence. Perform a moving average process to obtain a smoothed modal coefficient sequence. This filters out items smaller than the time scale. The high-frequency perturbations retain the main trends at this scale.

[0036] In a preferred embodiment, after obtaining the smoothed sequence, the system processes the sequence at each scale. Perform a Discrete Fourier Transform (DFT) to calculate its spectrum. The system has a spectral amplitude. Find the frequency corresponding to the maximum value in the modal response and define it as the modal principal response frequency. At this point, the system combines the spatial frequency corresponding to this mode. The effective phase velocity for a single event is calculated using the following formula: In the formula, Analogous to the wave number in the physical wave equation, It is a very small positive number (e.g.) ), used to prevent when eigenvalues The formula becomes unstable when the denominator is zero, especially when the denominator is close to zero (e.g., the DC component). Quantified at a specific spatial structural scale and specific time scales of observation Below, the propagation rate of moisture fluctuations. Finally, to eliminate the randomness of a single irrigation event (such as the influence of irrigation volume and initial moisture content), the system performs a multi-stage analysis on all selected... The effective phase velocities calculated under each historical event window are averaged to obtain the final multi-timescale hydraulic propagation wave velocity used for dispersive analysis. .

[0037] In some possible implementations, the dominant frequency extraction relies on a global Fourier transform, based on the assumption that fluctuations within the event window are stationary. However, in some rapidly changing rainstorm events, the moisture response exhibits strong transient characteristics. Therefore, wavelet transform or Hilbert-Huang transform (HHT) is used instead of Fourier transform. Specifically, the smoothed mode coefficients are processed... Empirical Mode Decomposition (EMD) is performed to extract its first intrinsic mode functions (IMF1), and its instantaneous frequencies are calculated using Hilbert transform. The system calculates the weighted average or median of the instantaneous frequency within the event window, which is then used as the primary response frequency for that event. This allows for the capture of the impulse response characteristics during the initial stage of water infiltration, resulting in a more accurate calculation of the phase velocity that better reflects the dynamic water conductivity of the soil. This method is particularly suitable for soil environments with extremely rapid infiltration, such as sandy soils.

[0038] In some possible embodiments, in waveguide theory, the speed of energy propagation is often determined by the group velocity. Decision. In calculating the dispersion curve (i.e. about After determining the scatter distribution, instead of directly calculating the ratio, a local difference method is used. Specifically, for two adjacent modes on the eigenvalue spectrum... and Calculate its group velocity estimate The group velocity was then averaged using events. The group velocity is more sensitive to structural changes in the medium, as disruption of the trench structure (such as blockage) can significantly alter the group delay characteristics of the wave packet.

[0039] It should be noted that the row-direction propagation dispersion and transverse propagation dispersion are calculated separately based on the statistical differences in hydraulic propagation wave velocities across different frequency bands at multiple time scales to quantify the integrity of the row band structure. Specifically, dispersion usually refers to the phenomenon of wave velocity varying with frequency. In a row band hydraulic network, if the furrow structure is intact (i.e., a regular row-direction waveguide is formed), then the various row-direction modes within the same frequency band should exhibit convergent propagation characteristics, and their effective phase velocities should be similar. The distribution of these frequencies should be relatively concentrated, i.e., with low dispersion. Conversely, if the zonal structure collapses, becomes blocked, or breaks, disrupting the physical consistency of the hydraulic channels, then even modes with similar frequencies will experience drastic fluctuations in propagation speed due to the randomness of local obstructions, manifesting as... The distribution of wave velocity is extremely discrete. For lateral propagation, due to the obstruction of ridges and heterogeneity, its wave velocity distribution should inherently be divergent. Therefore, quantifying this degree of dispersion by calculating statistical variance is actually measuring the structural entropy or consistency level of row orientation and lateral channels within the field.

[0040] In a preferred embodiment, to examine structural features at different scales, the system first sets the numerical range of the entire spatial frequency range. Divided into A preset spatial frequency band ,in (For example, divided into low frequency bands) Mid-frequency band and high frequency band The boundaries of the division can be based on quantiles of the eigenvalue distribution to ensure that each frequency band contains approximately the same number of modes. Then, for each preset time scale... and each spatial frequency band The system combines the generated row-oriented dominant mode set and lateral dominant mode set Perform the screening.

[0041] In a preferred embodiment, for the calculation of row-directed propagation dispersion, the system filters out all conditions that satisfy the following criteria. and The modal indexes constitute a set. Let the set contain the number of modes. Then the directional propagation dispersion Defined as the sample variance of the effective phase velocities within this set, the specific calculation formula is as follows: .in, The mean of the effective phase velocities within this set, i.e. Similarly, for transverse propagation dispersion, the system filters out all conditions that satisfy the criteria. and The modalities constitute a set And calculate its variance. These two indicators and It statistically accurately characterizes the degree of disorder in the longitudinal and lateral propagation of water at specific spatiotemporal scales.

[0042] In some possible embodiments, under certain irrigation conditions, the overall flow velocity propagating in the row direction may be much faster than that in the transverse seepage (i.e., In this scenario, the absolute variance might be influenced by the magnitude of the mean (i.e., the faster the velocity, the greater the potential for absolute fluctuation), thus masking structural differences. Therefore, the square of the coefficient of variation (CV) is used instead of the variance. The specific calculation formula is adjusted as follows: ,in It is the standard deviation of the effective phase velocity. By introducing mean normalization, the interference caused by the difference in basic flow velocity is eliminated, so that the dispersion index purely reflects the relative dispersion of wave velocity distribution. It is particularly suitable for scenarios that need to compare the field structure under different irrigation intensities (such as drip irrigation and flood irrigation).

[0043] In some possible embodiments, due to individual faults or reading jumps in the field sensors, the calculated effective phase velocity of certain modes may be affected. Extreme outliers can significantly inflate the calculated ordinary variance, leading to misjudgments. Therefore, Median Absolute Deviation (MAD) is used as a robust estimator of dispersion. The specific procedure is as follows: first, calculate the set... Median of all wave velocities Then, the absolute difference between each wave velocity and the median is calculated, and finally, the median of these absolute differences is taken as the dispersion. This quantile-based statistical method is insensitive to extreme values ​​and can effectively filter out false high-dispersion features caused by single-point sensor failures, ensuring the stability and reliability of environmental condition analysis results.

[0044] It should be noted that determining the asymmetric characteristics of lateral and row-direction dispersion transforms statistical dispersion into a dimensionless structural fingerprint. Specifically, for a structurally sound and functionally normal furrow-mulched field, its hydraulic environment should exhibit extremely strong anisotropy. This is manifested in the smooth and uniform propagation of water along the row direction (furrow direction) (i.e., row-direction dispersion). The spread along the lateral direction (across the ridge) should exhibit a high degree of disorder and divergence (i.e., lateral dispersion) due to the barrier of the ridge body and the heterogeneous soil matrix. (relatively large). Therefore, and The ratio between them corresponds to the signal-to-noise ratio or structural integrity of the row belt structure in that field. When the ratio is significantly greater than 1, it indicates that the row belt channels are clear; when the ratio decreases and approaches 1, it indicates that the row channels may be blocked or collapsed, making the propagation along the ridges as disordered as that across the ridges, that is, the system tends to a degenerate state of isotropicity.

[0045] In a preferred embodiment, the system constructs the line band time-vertex dispersion asymmetry index (RTV-DAI) as the asymmetric feature, and its specific calculation formula is as follows: In the formula, For a very small positive regularization parameter (e.g., taking a value of...), Its main function is numerical stability, preventing the denominator from deteriorating under extreme conditions (such as ideal laminar flow). The value approaches zero, causing computational overflow. The calculated value... This constitutes a multidimensional feature tensor, where each element represents the ratio of lateral disorder to row disorder at a specific spatiotemporal scale. A higher asymmetric feature indicates a healthier and more design-compliant row hydraulic structure.

[0046] In some possible embodiments, due to soil heterogeneity, the dispersion may vary across multiple orders of magnitude, resulting in an extremely uneven distribution of direct ratio values ​​(i.e., excessively skewed), which is detrimental to subsequent threshold determination. Therefore, a logarithmic domain asymmetric feature calculation method is adopted. Specifically, the system calculation formula is adjusted as follows: By introducing the natural logarithm transformation, this feature transforms the originally non-linear multiple relationship into a linear difference relationship, compressing the dynamic range of the values ​​and making the calculation results more consistent with the characteristics of a normal distribution.

[0047] In some possible implementations, to map asymmetric features to a bounded normalized interval to facilitate the standardization of the output or input to a machine learning model, a Normalized Dispersion Difference Index (NDDI) is constructed to characterize asymmetry, referencing the design principles of normalized difference indices (such as NDVI) in the remote sensing field. The specific calculation formula is as follows: The range of values ​​for this indicator is limited to... Between. When When it approaches 1, it means That is, structural health; when When it approaches 0, it means This refers to a structure that is severely degraded or completely isotropic. Compared to the potential infinity of the direct quotient, the normalized form is safer and more controllable in engineering implementation, and naturally possesses the ability to adaptively compensate for drifts in environmental benchmarks such as light and temperature. It is particularly suitable for smart agriculture IoT systems that require long-term continuous monitoring and experience drastic changes in the environmental background.

[0048] It should be noted that the environmental structure deviation index is obtained by calculating the difference between asymmetric features and preset health baseline features. This index is then compared with a preset structural anomaly threshold to obtain the environmental state analysis results. Multi-scale RTV-DAI is considered as the structural fingerprint of the field at the current moment. Since the row hydraulic structure of the field (such as ditch depth and membrane integrity) determines the dispersion pattern of water fluctuations, this fingerprint should be relatively stable and have a specific morphology (i.e., baseline features) under structural health conditions. When structural degradation occurs (such as collapse or blockage), the current fingerprint will be distorted at specific frequency bands or time scales. Specifically, the weighted Euclidean distance between the current state point and the ideal health point is calculated in the high-dimensional feature space. By introducing a logarithmic transformation, the asymmetry of the ratio index is eliminated (i.e., it can equally measure excessive and insufficient deviations), and through a weighted summation mechanism, the spatiotemporal scales most sensitive to structural changes are focused on, thereby achieving high-precision and automated judgment of the environmental state.

[0049] In a preferred embodiment, the system first retrieves pre-calibrated preset health benchmark features from the memory. This baseline characteristic is typically obtained by averaging several irrigation events during the early stages of crop planting, immediately after the furrow structure has been constructed and manually verified to be in good condition. Subsequently, the system targets each preset time scale... and each spatial frequency band Calculate the features at the current time. The ratio to the reference feature, and its natural logarithm to obtain the logarithmic deviation value, is calculated using the following formula: The natural logarithm is used because RTV-DAI is a ratioistic indicator, and the logarithmic transformation converts multiplicative bias into additive bias, ensuring that doubling the eigenvalue results in an equal deviation from halving it (i.e.,...). This ensures the fairness of the judgment logic. Next, the system calculates the Environmental Structure Deviation Index (ESI), using the formula: In this formula, Energy representing deviation; These are preset weighting coefficients, and they satisfy... The weighting coefficients are set based on prior knowledge: typically low-frequency bands ( Longer timescales are more sensitive to overall structural damage (such as large-area collapse), and therefore are given greater weight; while higher frequency bands may be more affected by random noise, and are therefore given less weight. Finally, the system calculates the scalar... Compared with the preset structural anomaly threshold Compare. If The system determines the environmental status label. (Abnormality) triggers an alarm or maintenance suggestion; if Then determine (normal).

[0050] In some possible implementations, in microcontroller (MCU) scenarios with extremely limited computing resources, squaring operations and floating-point multiplication can incur additional overhead. Therefore, Manhattan distance is used... The norm is used to construct the deviation index. Specifically, the system directly calculates the weighted sum of absolute deviations, and the formula is adjusted as follows: This linear accumulation method is not only faster to calculate, but also more robust to extreme outliers caused by sensor malfunctions, because it does not over-amplify large deviations in a single dimension as much as squaring, thus reducing the false alarm rate.

[0051] In some possible embodiments, as crops grow, root development may slowly alter the soil's pore structure, causing the healthy baseline itself to drift. Therefore, this auxiliary embodiment introduces a dynamic sliding window baseline update mechanism. Specifically, the system maintains a baseline containing the most recent... The feature queue of events that are subsequently determined to be normal. Each time a decision is made, the mean of the features in this queue is used as the current reference benchmark. At the same time, to prevent abnormal samples from contaminating the benchmark, only when the calculated... Less than a more stringent safety threshold (in Only when the current feature value is updated into the queue is the adaptive update strategy used to distinguish between slow structural evolution caused by crop growth and sudden structural degradation caused by furrow damage. This significantly extends the effective lifespan of the monitoring system and reduces the frequency of manual recalibration.

[0052] Example 2: Figure 2 As shown, the environmental status analysis system based on smart agriculture, applied to any of the environmental status analysis methods based on smart agriculture, includes:

[0053] The topology operator generation module is used to construct a row hydraulic network reflecting the spatial connectivity of furrows based on multi-point field monitoring data, and to generate the corresponding row hydraulic topology operators.

[0054] The mode decomposition and classification module is used to perform feature decomposition on the row hydraulic topology operator to obtain hydraulic wave modes describing the water distribution pattern, and to divide the hydraulic wave modes into a row-direction dominant mode set that propagates along the planting row and a lateral dominant mode set that propagates across the planting row based on the spatial projection direction; to extract the time response frequency of each hydraulic wave mode, and to calculate the multi-timescale hydraulic propagation wave velocity in combination with the corresponding spatial frequency;

[0055] The dispersion calculation module is used to calculate the row-direction dispersion and the lateral-direction dispersion respectively based on the statistical differences of the hydraulically conducted wave velocity at the multi-time scale in different frequency bands.

[0056] An asymmetric feature extraction module is used to determine the asymmetric features of the transverse conduction dispersion and the row conduction dispersion.

[0057] The index calculation module is used to calculate the difference between the asymmetric feature and the preset health benchmark feature to obtain the environmental structure deviation index.

[0058] The state determination module is used to compare the environmental structure deviation index with a preset structural anomaly threshold to obtain the environmental state analysis result.

[0059] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.

Claims

1. An environmental status analysis method based on smart agriculture, characterized in that, include: Based on multi-point field monitoring data, a row hydraulic network reflecting the spatial connectivity of furrows and ridges is constructed, and corresponding row hydraulic topology operators are generated. The row-band hydraulic topology operator is subjected to eigenvalue decomposition to obtain hydraulic wave modes describing the water distribution pattern. Based on the spatial projection direction, the hydraulic wave modes are divided into a row-oriented dominant mode set that propagates along the planting row and a lateral dominant mode set that propagates across the planting row. The time response frequency of each hydraulic wave mode is extracted, and the multi-timescale hydraulic transmission wave velocity is calculated in combination with the corresponding spatial frequency. Based on the statistical differences of the multi-timescale hydraulic transmission wave velocity in different frequency bands, the row-oriented transmission dispersion and the lateral transmission dispersion are calculated respectively. Determine the asymmetric characteristics of the transverse conductive dispersion and the row-directed conductive dispersion; The difference between the asymmetric feature and the preset health benchmark feature is calculated to obtain the environmental structure deviation index; The environmental structure deviation index is compared with a preset structural anomaly threshold to obtain the environmental state analysis results.

2. The environmental status analysis method based on smart agriculture according to claim 1, characterized in that, Based on multi-point field monitoring data, a row hydraulic network reflecting the spatial connectivity of furrows and rows is constructed, and corresponding row hydraulic topology operators are generated, including: The coordinates of each monitoring node deployed in the field are obtained, including row coordinates along the planting row direction and lateral coordinates across the planting row direction; based on the multi-point test data within a preset rainfall or irrigation event window, the statistical correlation coefficient between monitoring node pairs is calculated; a weighted adjacency matrix is ​​constructed, wherein the values ​​of the off-diagonal elements of the weighted adjacency matrix are determined by the product of a negative exponential decay term based on the Euclidean distance between monitoring node pairs and a positive exponential weight term based on the statistical correlation coefficient, and the values ​​of the diagonal elements are set to zero; the sum of the elements in each row of the weighted adjacency matrix is ​​calculated to form a diagonal degree matrix; the weighted adjacency matrix is ​​subtracted from the degree matrix to obtain the row band hydraulic topology operator.

3. The environmental status analysis method based on smart agriculture according to claim 2, characterized in that, Eigenvalue decomposition of the line band hydraulic topology operator yields hydraulic fluctuation modes describing water distribution patterns, including: Eigenvalue decomposition is performed on the row-band hydraulic topology operator; the decomposed eigenvalues ​​are extracted and sorted according to their numerical values, and the eigenvalues ​​are used as spatial frequencies characterizing the intensity of spatial changes in field water; the eigenvectors corresponding to the decomposed eigenvalues ​​are extracted, and the eigenvectors are used as hydraulic fluctuation modes characterizing the spatial distribution structure of field water.

4. The environmental status analysis method based on smart agriculture according to claim 3, characterized in that, Based on the spatial projection direction, the hydraulic wave modes are divided into a row-oriented dominant mode set propagating along the planting row and a lateral dominant mode set propagating across the planting row, including: The directional and lateral coordinates of each monitoring node are centered to construct directional and lateral coordinate basis vectors reflecting the average positional deviation of the entire field. The normalized inner product of each hydraulic fluctuation mode with the directional coordinate basis vector is calculated as the directional projection coefficient, and the normalized inner product of each hydraulic fluctuation mode with the lateral coordinate basis vector is calculated as the lateral projection coefficient. The absolute values ​​of the directional and lateral projection coefficients are compared. If the absolute value of the directional projection coefficient is greater than the absolute value of the lateral projection coefficient, the hydraulic fluctuation mode is assigned to the directional dominant mode set. If the absolute value of the lateral projection coefficient is greater than or equal to the absolute value of the directional projection coefficient, the hydraulic fluctuation mode is assigned to the lateral dominant mode set.

5. The environmental status analysis method based on smart agriculture according to claim 4, characterized in that, Extracting the time response frequencies of each hydraulic wave mode and combining them with the corresponding spatial frequencies to calculate the multi-timescale hydraulic propagation wave velocity, including: Multi-point test data within the rainfall or irrigation event window are projected onto each hydraulic wave mode to obtain a time-varying modal time coefficient sequence. For multiple preset time scales, the modal time coefficient sequences are subjected to moving average processing to obtain smoothed modal coefficient sequences at different time scales. A discrete Fourier transform is performed on the smoothed modal coefficient sequences, and the frequency corresponding to the maximum spectral amplitude after the transform is extracted as the modal principal response frequency. The ratio of the modal principal response frequency to the square root of the corresponding spatial frequency is calculated to obtain the effective phase velocity under a single event. The arithmetic mean of the effective phase velocities calculated under all preset rainfall or irrigation event windows under a single event is taken to obtain the multi-time-scale hydraulic wave velocity.

6. The environmental status analysis method based on smart agriculture according to claim 5, characterized in that, Based on the statistical differences in hydraulically conducted wave velocities across different frequency bands at multiple time scales, the longitudinal and transverse conduction dispersions are calculated, including: The spatial frequency range is divided into several preset spatial frequency bands. For each preset time scale and each spatial frequency band, hydraulic wave modes belonging to the lateral dominant mode set and whose corresponding spatial frequencies fall within the spatial frequency band are selected. The statistical variance of the corresponding multi-time scale hydraulic transmission wave velocity is calculated, and the obtained statistical variance is used as the lateral transmission dispersion. For each preset time scale and each spatial frequency band, hydraulic wave modes belonging to the lateral dominant mode set and whose corresponding spatial frequencies fall within the spatial frequency band are selected. The statistical variance of the corresponding multi-time scale hydraulic transmission wave velocity is calculated, and the obtained statistical variance is used as the lateral transmission dispersion.

7. The environmental status analysis method based on smart agriculture according to claim 6, characterized in that, Determining the asymmetric characteristics of the transverse conduction dispersion and the row conduction dispersion includes: For each preset time scale and each spatial frequency band, the quotient obtained by dividing the transverse conduction dispersion by the longitudinal conduction dispersion is calculated; the quotient is used as the asymmetric feature under that time scale and that spatial frequency band.

8. The environmental status analysis method based on smart agriculture according to claim 7, characterized in that, The difference between the asymmetric feature and the preset health benchmark feature is calculated to obtain the environmental structure deviation index. The environmental structure deviation index is compared with the preset structural anomaly threshold to obtain the environmental state analysis results, including: Obtain the preset health benchmark features pre-calibrated under the field structure health status; for each preset time scale and each preset spatial frequency band, calculate the natural logarithm of the quotient obtained by dividing the asymmetric feature by the preset health benchmark features, as the logarithmic deviation value; calculate the square of the logarithmic deviation value and multiply it by preset weighting coefficients corresponding to the time scale and spatial frequency band to obtain the weighted square deviation; sum the weighted square deviations under all preset time scales and all preset spatial frequency bands to obtain the environmental structure deviation index; if the environmental structure deviation index is greater than or equal to the structural anomaly threshold, the environmental state is determined to be abnormal; if the environmental structure deviation index is less than the structural anomaly threshold, the environmental state is determined to be normal.

9. An environmental status analysis system based on smart agriculture, applied in the environmental status analysis method based on smart agriculture as described in any one of claims 1-8, characterized in that, include: The topology operator generation module is used to construct a row hydraulic network reflecting the spatial connectivity of furrows based on multi-point field monitoring data, and to generate the corresponding row hydraulic topology operators. The mode decomposition and classification module is used to perform feature decomposition on the row hydraulic topology operator to obtain hydraulic wave modes describing the water distribution pattern, and to divide the hydraulic wave modes into a row-direction dominant mode set that propagates along the planting row and a lateral dominant mode set that propagates across the planting row based on the spatial projection direction; to extract the time response frequency of each hydraulic wave mode, and to calculate the multi-timescale hydraulic propagation wave velocity in combination with the corresponding spatial frequency; The dispersion calculation module is used to calculate the horizontal and lateral dispersions based on the statistical differences in the hydraulic wave velocity at different time scales in different frequency bands. An asymmetric feature extraction module is used to determine the asymmetric features of the transverse conductive dispersion and the row-directed conductive dispersion; The index calculation module is used to calculate the difference between the asymmetric feature and the preset health benchmark feature to obtain the environmental structure deviation index. The state determination module is used to compare the environmental structure deviation index with a preset structural anomaly threshold to obtain the environmental state analysis result.

Images