Tea whole-process digital early warning and block chain responsibility tracing method

Abstract

The invention relates to the technical field of multi-source remote sensing data processing, and discloses a tea whole-process digital early warning and block chain responsibility tracing method, which comprises the following steps: firstly, constructing a terrace planting micro-unit distribution map, and analyzing terrace landform repeated spacing; then acquiring multi-source heterogeneous sensing data, determining a comprehensive spatial resolution and a comprehensive time acquisition interval, and calculating a space-time sampling aliasing index according to the comprehensive spatial resolution and the comprehensive time acquisition interval; and generating a monitoring reliability attenuation factor by using the index, and recovering the anti-aliasing signal by using the monitoring reliability attenuation factor as a regular constraint to obtain a reliability correction risk value. The early warning event is generated according to the corrected risk value, the responsibility traceability allocation weight is calculated by using the monitoring reliability attenuation factor, and finally the early warning event, the responsibility weight and the recovery parameter are stored in the block chain system. According to the method, the problem of observation distortion of the mountain tea garden caused by cloud and mist shielding and terrain breakage is effectively solved, and the accuracy and credibility of risk early warning and responsibility determination are ensured.

Description

Technical Field

[0001] This invention relates to the field of multi-source remote sensing data processing technology, and more specifically, to a method for digital early warning and blockchain-based responsibility traceability for the entire tea production process. Background Technology

[0002] In the current digitalization process of the tea industry, the collaborative monitoring of tea gardens using satellite remote sensing, drone imagery, and ground-based IoT sensors, combined with blockchain technology for product traceability, has become an important means to ensure tea quality and safety. Existing technologies typically treat the monitoring area as a continuous, smooth surface or a regular grid, directly collecting environmental information such as spectral data, temperature, and humidity parameters. This data is then used as the basis for determining production risks (such as pests, diseases, and frost), and subsequently, the relevant data is immutably recorded through blockchain to clarify production responsibility.

[0003] However, in actual mountainous and hilly tea-growing areas, the physical morphology and production environment of tea gardens have extremely unique characteristics, leading to significant systematic biases in existing monitoring and traceability systems:

[0004] First, mountain tea gardens are not continuous natural slopes, but rather terraced composite structures constructed through engineering methods. This structure consists of a series of nearly horizontal terraces alternating with steep embankments (or terrace walls) supporting the terraces, forming micro-topographic units that repeat periodically along the slope. Different micro-topographic units (such as terraces, terrace walls, and rows) exhibit significant meter-level differences in soil physicochemical properties, microbial community distribution, and disease incidence rates.

[0005] Secondly, existing monitoring methods often fall short of matching the inherent structure of the mountainous environment in both spatial and temporal dimensions. Spatially, the spatial resolution of mainstream monitoring methods such as satellite remote sensing (typically around 10 meters) is far lower than the repeatability scale of the terraced micro-topographic structure (typically around 1-3 meters). Temporally, mountain tea gardens are often covered by clouds and fog, resulting in frequent gaps in optical remote sensing data; at the same time, the microclimate changes caused by the nighttime cold air infiltration and cold pool effect in mountainous areas have rapid and transient dynamic characteristics.

[0006] When the sampling capability (spatial resolution or temporal acquisition frequency) of a monitoring system is lower than the structural frequency or dynamic change frequency of the monitored object, physical aliasing or folding phenomena will occur. In this case, high-frequency risk signals in the micro-topography of terraced fields (such as localized damage to the terrace walls or instantaneous cold air frost) will be incorrectly folded into low-frequency spurious fluctuations or background drift.

[0007] Existing multi-source data fusion and blockchain traceability technologies have not adequately considered the observation distortion caused by the failure of the sampling theorem. Systems often directly record and upload aliased erroneous signals as actual physical quantities. This not only leads to spatiotemporal misalignment of early warning positioning, but also solidifies the attribution of faults (such as attributing false anomalies caused by observation folding to tea farmers' mismanagement) into immutable evidence, thereby undermining the fairness and credibility of the traceability system. Summary of the Invention

[0008] This invention provides a digital early warning and blockchain-based responsibility traceability method for the entire tea production process, solving the technical problems mentioned in the background section.

[0009] This invention provides a method for digital early warning and blockchain-based traceability of the entire tea production process, including: Construct a distribution map of terraced planting micro-units and analyze the terraced landform repeating spacing along the slope of each terraced planting micro-unit; Acquire multi-source heterogeneous sensing data for the terraced planting micro-units, set a predetermined time window, calculate and determine the comprehensive spatial resolution and comprehensive temporal acquisition interval within the predetermined time window, and obtain the duration of risk processes for the terraced planting micro-units; Based on the ratio of the comprehensive spatial resolution to the terraced landform repetition spacing, and the ratio of the comprehensive temporal acquisition interval to the duration of the risk process, the spatiotemporal sampling aliasing index is calculated. The monitoring reliability attenuation factor is generated using the spatiotemporal sampling aliasing index, and the monitoring reliability attenuation factor is used as a constraint condition to perform anti-aliasing signal restoration processing on the multi-source heterogeneous sensing data to obtain restoration model parameters and reliability correction risk values. An early warning event is generated based on the reliability correction risk value, and the corresponding responsibility traceability allocation weight for the tea batch is calculated based on the monitoring reliability decay factor. The early warning event, the responsibility traceability allocation weight, and the restoration model parameters are stored in the blockchain system for traceability.

[0010] The beneficial effects of this invention are as follows: By constructing a distribution map of terraced planting micro-units and quantifying the repeating spacing of terraced landforms, the structural information of mountain tea gardens is introduced into the monitoring system, effectively solving the signal aliasing problem caused by insufficient sampling capacity in complex terrain. By calculating the spatiotemporal sampling aliasing index and the monitoring reliability attenuation factor, this invention introduces physical constraints in the signal restoration stage, which can distinguish between real risk fluctuations and false signals caused by observation folding, thereby avoiding drift in early warning positioning. At the same time, this invention incorporates monitoring reliability into the responsibility attribution calculation, automatically reducing the responsibility weight in areas with poor observation conditions, preventing erroneous attributions from being solidified by the blockchain, and achieving a synergistic improvement in the accuracy of early warnings and the fairness of responsibility traceability in mountainous scenarios. Attached Figure Description

[0011] Figure 1 This is a flowchart of the tea industry's whole-process digital early warning and blockchain responsibility traceability method. Detailed Implementation

[0012] 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.

[0013] like Figure 1 As shown, the tea industry's end-to-end digital early warning and blockchain-based responsibility traceability method includes: Construct a distribution map of terraced planting micro-units and analyze the terraced landform repeating spacing along the slope of each terraced planting micro-unit; Acquire multi-source heterogeneous sensing data for the terraced planting micro-units, set a predetermined time window, calculate and determine the comprehensive spatial resolution and comprehensive temporal acquisition interval within the predetermined time window, and obtain the duration of risk processes for the terraced planting micro-units; Based on the ratio of the comprehensive spatial resolution to the terraced landform repetition spacing, and the ratio of the comprehensive temporal acquisition interval to the duration of the risk process, the spatiotemporal sampling aliasing index is calculated. The monitoring reliability attenuation factor is generated using the spatiotemporal sampling aliasing index, and the monitoring reliability attenuation factor is used as a constraint condition to perform anti-aliasing signal restoration processing on the multi-source heterogeneous sensing data to obtain restoration model parameters and reliability correction risk values. An early warning event is generated based on the reliability correction risk value, and the corresponding responsibility traceability allocation weight for the tea batch is calculated based on the monitoring reliability decay factor. The early warning event, the responsibility traceability allocation weight, and the restoration model parameters are stored in the blockchain system for traceability.

[0014] In a preferred embodiment, the construction of the terraced planting micro-unit distribution map specifically includes: Collect digital elevation models and high-resolution orthophotos of the tea plantation area to identify the dominant slope direction vectors. ; For any spatial location within the tea garden area Calculate its aspect projection coordinates on the dominant direction vector of the slope. The calculation formula is as follows:

[0015] in, This represents the vector dot product operation; Set a fixed micro-unit segmentation length Divide along the axis of the dominant direction vector of the slope, the first... Each terraced planting micro-unit Defined as a set of points that satisfy the following conditions:

[0016] in, This is the index of the segment number.

[0017] When constructing a distribution map of terraced planting micro-units, based on the periodic distribution characteristics of terraced structures along the slope aspect in mountain tea gardens, precise spatial division is required to effectively represent the micro-topographic units of the terraces. First, a digital elevation model (DEM) and high-resolution orthophotos of the tea garden area are collected. The DEM provides comprehensive topographic elevation information, supporting the identification of the dominant slope direction, while the high-resolution orthophotos help verify detailed topographic features and avoid errors that may arise from a single data source. Through statistical analysis of the slope aspect information in the DEM data, the dominant mode of slope aspect distribution is extracted; the direction corresponding to this dominant mode is the dominant slope direction vector. Based on the construction principles of terraced field projects, namely that terraces typically extend along the dominant direction of the slope, the dominant modal direction best matches the actual distribution trend of the terraces. Next, for any spatial point within the tea garden area... The slope aspect projection coordinates are calculated by vector inner product operation. By projecting two-dimensional spatial coordinates onto the dominant one-dimensional slope direction, the planar terraced fields are transformed into an ordered one-dimensional sequence. The vector inner product operation accurately reflects the relative positional relationship of points along the dominant slope direction, ensuring the continuity and monotonicity of the projected coordinates and avoiding positional misalignment. Finally, a fixed micro-unit segmentation length is set. The tea garden area is divided into equal-interval sections along the axis of the dominant slope direction vector, generating multiple discrete sets of terraced planting micro-units. microunit Defined as satisfying The set of points, with a fixed cutting length To ensure uniform scale across all micro-units, equidistant division guarantees complete coverage of the entire tea garden area, leaving no terraced fragments untouched. The sequence index... Each micro-unit can be uniquely identified.

[0018] In a preferred embodiment, the analysis of the terraced landform repeating interval along the slope of each terraced planting micro-unit is described. Specifically, it includes: Sampling intervals were fixed along the slope profile of the terraced planting micro-units. Get high program column Calculate the slope curvature sequence using second-order difference :

[0019] Calculate the curvature autocorrelation function of the slope curvature sequence. :

[0020] Within the preset search range Finding peak hysteresis And calculate the repeating interval of the terraced landform. :

[0021]

[0022] in, For sampling point index, This represents the autocorrelation lag step.

[0023] When analyzing the repeating intervals of terraced landforms along the slope of each terraced planting micro-unit, based on the alternating distribution pattern of the terraced structure along the slope, it is necessary to extract curvature features through elevation changes and then locate the repeating period using autocorrelation analysis. First, a fixed sampling interval is used along the slope profile of each terraced planting micro-unit. Get high program column Fixed sampling interval This ensures a uniform distribution of elevation data along the aspect dimension, avoiding feature distortion caused by uneven sampling density, and high-precision sequencing. It directly reflects the elevation undulations formed by the alternating terrace surfaces and walls. Next, based on the elevation sequence... Calculate the slope curvature sequence using second-order difference The calculation formula is: Second-order difference operations can effectively amplify the abrupt changes in elevation. The elevation of terraced fields changes rapidly at the terrace walls, corresponding to the curvature sequence. The extreme points in the slope are located at the top of the terrace, while the elevation is relatively gentle and the curvature value is small, thus forming a periodic curvature change corresponding to the terrace structure. Subsequently, the slope curvature sequence was analyzed. Perform autocorrelation analysis to calculate the curvature autocorrelation function. The autocorrelation function is used to measure the curvature sequence at different lag steps. Similarity under the following conditions, when the number of lag steps When the distance equals the terrace repetition spacing, the curvature sequences exhibit significant similarity, and the autocorrelation function reaches its peak. Finally, within the preset search range... Identify the first significant non-zero peak of the internal curvature autocorrelation function and determine the hysteresis step number corresponding to the peak. and through The calculated terrace repeating intervals were obtained. The first significant non-zero peak was selected because it corresponds to the most basic structural period of the terraces, thus eliminating interference from higher-order harmonics or noise. (Lag step count) With sampling interval The product is directly converted into actual spatial distance, quantifying the repetition interval of the terraces along the slope.

[0024] In a preferred embodiment, the step of acquiring multi-source heterogeneous sensing data for the terraced planting micro-units and calculating the comprehensive spatial resolution is described. With integrated time acquisition interval Specifically, it includes: within the scheduled time window Inside, obtain Data sources of different types, each with a single source spatial resolution. and fixed source weight coefficients The comprehensive spatial resolution is calculated using the weighted sum of squares formula. :

[0025] Obtain the predetermined time window Internal valid observation timestamp sequence sorted by time The median of the differences between adjacent timestamps is calculated as the integrated time acquisition interval. :

[0026] in, For the number of valid observations, For timestamp index, It is the median function, used to calculate the median of a data sequence.

[0027] When calculating the integrated spatial resolution and integrated temporal acquisition interval, given the differences in sampling characteristics among multi-source heterogeneous data, a standardized fusion method is needed to transform the spatial sampling capabilities and temporal observation effectiveness of different data sources into unified parameters. First, a predetermined time window needs to be set. Within this scope, data on micro-units for terraced planting were collected. Data sources of different types, each corresponding to a single source spatial resolution. With fixed source weight coefficient The source weighting coefficients are allocated based on the observation accuracy and reliability of the data sources. For example, ground sensor data, because it directly contacts the monitored object, has a higher weighting coefficient than satellite remote sensing data, thus reflecting the different contributions of different data sources to spatial sampling. To integrate the spatial sampling capabilities of multi-source data, a weighted summation method using the inverse square is used to derive the comprehensive spatial resolution. The calculation logic is as follows: the reciprocal of the spatial resolution directly reflects the sampling density. Squaring this value amplifies the differences in sampling density between different data sources. Then, weighted summation using source weight coefficients highlights the influence of highly reliable data sources. Finally, the reciprocal of the weighted sum is taken and the square root is obtained. This formula can transform the spatial resolution of multi-source heterogeneous systems into a single, comparable comprehensive index. Next, the sampling parameters in the time dimension are processed to obtain a predetermined time window. Valid observation timestamp sequence arranged in chronological order , To ensure the validity of observations, factors such as cloud cover and fog can cause some timestamps to be missing, thus affecting the continuity of time sampling. Therefore, the set of differences between two adjacent valid timestamps is calculated. The median of this set is selected as the comprehensive time acquisition interval. ,Right now The median is more resistant to the interference of extreme differences than the mean, and can accurately reflect the actual interval characteristics of time sampling, avoiding distortion of time sampling parameters due to accidental missing measurements.

[0028] In a preferred embodiment, the basis of the integrated spatial resolution Repeating spacing with the terraced landform The ratio relationship, and the comprehensive time acquisition interval. With the continuous timeliness of risk processes The ratio relationship is used to calculate the spatiotemporal sampling aliasing index. Specifically, it includes: For temperature series average value Calculate its normalized autocorrelation function. :

[0029] Find the autocorrelation decays to the reciprocal of the natural constant. The corresponding minimum lag steps Calculate the duration of the risk process. :

[0030] Calculate the first ratio The ratio with the second value The maximum value is used to obtain the spatiotemporal sampling aliasing index. :

[0031] in, This represents the number of lagging steps.

[0032] The calculation of the spatiotemporal sampling aliasing index revolves around the matching degree between spatial sampling and terraced structure, and the fit between temporal sampling and risk processes. By quantifying these two types of matching deviations and taking extreme values, a unified index reflecting the overall degree of aliasing is obtained. First, the temperature sequence of the terraced planting micro-units is acquired. Temperature series directly reflect microclimate changes related to risks such as cold air convergence and frost, and are a key carrier for characterizing risk processes; mean-removed processing yields... This eliminates interference from slow-changing trends such as seasonal fluctuations and overall daytime warming, retaining only the volatility components directly related to risk. The normalized autocorrelation function is calculated based on the mean-removed series. Molecular characterization sequences are delayed The similarity at each step is normalized so that the autocorrelation result is limited to the [0,1] interval, facilitating a unified assessment of the correlation degree. A preset decay threshold is set as the reciprocal of the natural constant. This threshold can robustly distinguish between significantly correlated and irrelevant states of a sequence, and find the satisfying... Minimum number of lag steps Combine it with the comprehensive time acquisition interval Multiplying them together yields the duration of the risk process. The product of the lag steps and the time interval can be directly converted into the actual duration of the risk process, quantifying the time scale of dynamic changes in risk. Next, the first ratio of the spatial dimension is calculated. ,in The minimum sampling interval threshold required to capture periodic signals in the corresponding sampling theorem corresponds to the repetition spacing of terraced landforms. The ratio directly reflects whether spatial sampling can cover the periodic structure of the terraces; the second ratio is calculated in the time dimension. Similarly, The minimum sampling interval threshold for capturing time-dependent dynamic signals, and the duration of the risk process. The ratio reflects whether the time sampling can keep up with the changing pace of the risk process. Finally, the maximum value of the first ratio and the second ratio is selected as the spatiotemporal sampling aliasing index. The severity of spatiotemporal aliasing is determined by the dimension with the worst matching between sampling capability and object features, and taking the maximum value can comprehensively characterize the overall aliasing level.

[0033] In a preferred embodiment, the use of the spatiotemporal sampling aliasing index Generate monitoring reliability decay factor Specifically, it includes: Set the preset attenuation adjustment constant The monitoring reliability decay factor is calculated using the natural exponential function. The calculation formula is as follows:

[0034] in, This represents the operation of the natural exponential function.

[0035] When generating the monitoring reliability decay factor, the focus is on the correlation between the spatiotemporal sampling aliasing index and the observation reliability. Through a monotonically increasing function mapping, the aliasing degree is transformed into a quantifiable factor that can be directly used for subsequent constraints. First, a preset decay adjustment constant is set. This constant controls the rate of reliability decay. Its value is calibrated using multiple sets of measured data to ensure that the decay level matches the actual observational distortion, avoiding deviations in subsequent processing due to over- or under-adjustment. Next, the spatiotemporal sampling aliasing index is calculated. The difference between the value and the value of one, when When the difference is 0, it means that the sampling capability exactly matches the characteristics of the observed object, aliasing does not occur, and the reliability does not need to be attenuated; when At that time, the difference increases with the degree of aliasing. Subsequently, the attenuation adjustment constant is... Multiplying this difference yields the input value of the exponential function. Linear scaling is then used to adapt the difference range to the response characteristics of the natural exponential function, ensuring smooth and discriminative changes in the decay factor under different degrees of aliasing. Finally, this product is used as the exponent to calculate the value of the natural exponential function, thus obtaining the monitoring reliability decay factor. The monotonically increasing property of the natural exponential function ensures that the more severe the aliasing (…), the better. The larger the value, the lower the reliability decay factor, and its value range is always positive, which is consistent with the characteristics of a reliability parameter.

[0036] In a preferred embodiment, the monitoring reliability decay factor is... As a constraint, anti-aliasing signal restoration processing is performed on the multi-source heterogeneous sensing data to obtain a confidence correction risk value. Specifically, it includes: Utilizing the terraced landform repeating interval Construct a signal restoration basis function model and monitor the signal. Expanded to:

[0037] Construct an optimization objective function that includes a data fitting error term and a smoothing regularization term, and solve for the optimal restoration model parameter vector. :

[0038] use The reconstruction yielded the restoration monitoring signal of the terraced planting micro-units. Calculate its time window Internal fluctuation energy :

[0039] Calculate the reliability-corrected risk value :

[0040] in, These are the slope aspect projection coordinates. The harmonic order is... This is the weight matrix. For the perception mapping matrix, The regularization coefficient is... For difference operators, This represents the window mean.

[0041] When performing anti-aliasing signal restoration and obtaining confidence-corrected risk values, the focus is on the correlation between the periodic structure of terraces and the degree of aliasing. Through structured basis function modeling, constrained optimization solutions, and confidence normalization, the aliasing-distorted observed signals are restored to true risk-related signals. First, the analyzed terrace repeating intervals are used... Construct a signal restoration basis function model and monitor the signal. It is decomposed into a superposition of a slowly varying background term and a periodic harmonic term. The slowly varying background term is adopted... Characterization, in which The slope aspect projection coordinates, in a quadratic polynomial form, can adequately fit long-term, slow fluctuations caused by seasonal variations and human management; the periodic harmonic term is obtained through... Build, To fix the harmonic order, this form directly matches the periodic structure of the terraces along the slope, and can capture meter-level high-frequency risk signals. Corresponding to different order periodic frequencies, the main fluctuation characteristics of the terraced structure are ensured to be covered. Next, an optimization objective function is constructed, including a data fitting error term and a smoothing regularization term. The data fitting error term employs... ,in This is a weight matrix (the weights for missing data are set to 0, and the weights for valid data are assigned according to the noise level). The perception mapping matrix (transforming the values ​​of the basis functions at the observation points into matrix form). These are observations from multi-source heterogeneous sensing data. The parameter vector of the basis functions is used; this error term ensures the consistency between the restored signal and the observed data; the smoothing regularization term is set to... , To fix the regularization coefficients, For the difference operator (used to suppress drastic fluctuations in the parameter vector), monitor the confidence decay factor. As a weight, the more severe the aliasing ( Larger micro-units achieve stronger smoothing constraints, avoiding the generation of spurious high-frequency signals due to noise interference. Solving for the optimal parameter vector This achieves a balance between data fitting and anti-aliasing. Subsequently, the optimal parameter vector is... Substituting back into the basis function model, in the corresponding terraced planting micro-unit The average value is taken to obtain the restored monitoring signal. This step converts the model output into a continuous signal at the unit level, consistent with the micro-unit partitioning logic. Afterwards, the restoration monitoring signal is calculated within a predetermined time window. Internal fluctuation energy ,in The length of the time window. The mean of the restored signal within the window is used, and the fluctuation energy quantifies the intensity of abnormal fluctuations in the signal, which is a core indicator characterizing risk. Finally, the fluctuation energy... With monitoring reliability decay factor Take the ratio to obtain the reliability-corrected risk value. This ratio calculation can normalize the wave energy under different degrees of aliasing, especially for severe aliasing ( The reliability of the fluctuation energy of the large unit is reduced to ensure that the final risk value can truly reflect the actual risk level within the micro unit.

[0042] In a preferred embodiment, the risk value is adjusted based on the confidence level. Generate early warning events and, based on the monitoring reliability decay factor... Calculate the weight of responsibility attribution Specifically, it includes: Using logical functions Calculate the probability of early warning :

[0043] in, These are model constants. For natural logarithm operations, The spatiotemporal sampling aliasing index; Obtain the terraced planting micro-units For batch Contribution Calculate the intermediate weights :

[0044] The intermediate weights are normalized to calculate the responsibility attribution weights. :

[0045] in, For the contribution batch The set of units.

[0046] When generating early warning events and calculating the responsibility attribution weights, a synergistic consideration of risk intensity and observation reliability is undertaken. Probabilistic modeling quantifies the corrected risk into an early warning signal, while responsibility weights are allocated based on contribution and reliability decay factors to ensure consistency between early warning and attribution. First, an early warning probabilistic logic function is constructed, incorporating a reliability-corrected risk value and a spatiotemporal sampling aliasing index. It can map input values ​​to the [0,1] interval, which precisely matches the range of probability values, thus representing the likelihood of risk occurrence. The function incorporates a confidence-corrected risk value. The linear term is because It has undergone reliability normalization and can accurately reflect the risk level within the micro-unit; the linear relationship directly reflects the positive correlation trend between risk intensity and warning probability; a spatiotemporal sampling aliasing index has been added. logarithmic terms This is because logarithmic operations can smooth out extreme fluctuations in aliasing levels, preventing sudden changes in warning probabilities due to excessively large or small aliasing values, thus affecting model constants. Calibration was performed using multiple sets of measured data to ensure that the function output matched the actual risk occurrence patterns, ultimately achieving [the desired result]. The probability of an early warning is calculated, and an early warning event is generated based on this probability, so that the early warning judgment is simultaneously linked to the risk itself and the reliability of the observation. Next, the responsibility attribution weight is calculated, first by obtaining the terraced planting micro-units. For a specific batch of tea Contribution The contribution directly reflects the actual participation of the micro-unit in batch production; using the contribution... Divide by the micro-unit in the corresponding time window of the batch Internal monitoring reliability decay factor To obtain the intermediate weights This step involves lowering the weight of severely aliased micro-units by using a reliability decay factor to prevent low-reliability observational data from dominating responsibility determination; subsequently, data belonging to the same tea batch... The intermediate weights of all micro-units are normalized, i.e. ,in To contribute the set of micro-units in this batch, the normalization operation makes the sum of the responsibility weights of all micro-units equal to 1, ensuring the completeness and rationality of responsibility allocation. The final responsibility attribution weights reflect both the differences in contribution and the credibility of observation, thus achieving the unity of responsibility attribution and early warning logic.

[0047] In a preferred embodiment, storing the early warning event, the responsibility tracing and allocation weight, and the recovery parameters into the blockchain system for traceability specifically includes: Construct a dual-ledger structure to generate Fast Ledger transaction payloads. With Evidence Ledger Transaction Load ; Define the Fast Ledger transaction payload for:

[0048] Define the evidentiary ledger transaction payload for:

[0049] in, For the event payload hash, This serves as the content addressing identifier for the original data within the distributed file system. For the restoration model parameters The hash fingerprint.

[0050] When storing early warning events, responsibility tracing allocation weights, and recovery parameters into the blockchain system, a balance between storage efficiency and auditability is considered. A dual-ledger structure separates core information from complete evidence, ensuring that the tracing process can quickly query key data and verify it through complete link recalculation, while also guaranteeing data immutability and consistency. First, a dual-ledger storage structure is constructed, including a fast ledger and an evidence ledger. The fast ledger focuses on low-load, high-frequency query scenarios, while the evidence ledger emphasizes the retention of complete evidence and process recalculation. The collaboration between the two avoids the inefficiency caused by storing massive amounts of data in a single ledger and prevents the mixing of core information and redundant data from affecting the convenience of tracing. Next, the transaction payload of the fast ledger is defined. The eventID and batchID are unique identifiers for the event and batch, ensuring accurate association with the corresponding object during tracing; timestamp With monitoring unit Defining the spatiotemporal attribution of data; spatiotemporal sampling aliasing index With the probability of early warning It is the core calculation result for early warning and responsibility determination, providing key evidence for tracing; event payload hash The data is serialized from the original event data and then calculated using SHA256, which can quickly verify the consistency between the on-chain data and the original data, preventing tampering. Subsequently, the evidence ledger is processed for data storage. Since the original data (such as multi-source remote sensing images and sensor time-series data) is large in volume, directly storing it on the blockchain would consume significant storage resources and reduce transaction efficiency. Therefore, it is written to a distributed file system (such as IPFS) to obtain a unique content address identifier. ,pass The original data can be retrieved at any time for recalculation; the hash fingerprint of the model parameters can be restored. The core parameter features for anti-aliasing signal recovery are preserved, ensuring that the parameters have not been tampered with and the signal recovery process can be reproduced through subsequent verification using this fingerprint; the reliability attenuation factor is monitored. As a core constraint parameter throughout anti-aliasing restoration and liability weighting calculation, its inclusion in the evidence ledger ensures the traceability of liability apportionment and risk value calculation. Finally, the transaction payload of the evidence ledger is defined. eventID enables precise association with Quick Ledger. and , Together, they form a complete chain of evidence, ensuring that during tracing, not only can the result be confirmed, but also the rationality and accuracy of the entire process can be recalculated, achieving the tracing goal of verifiable results, repeatable processes, and reliable data.

[0051] It should be noted that a digital elevation model (DEM) is a digital simulation of the ground terrain using limited terrain elevation data. Complete terrain elevation information for the tea plantation area is collected through legal and compliant means such as satellite remote sensing, UAV mapping, or ground surveying. DEM data with a resolution of at least 1 meter is preferred to ensure the accuracy of slope orientation identification.

[0052] It should be noted that high-resolution orthophotos are remote sensing images that have undergone geometric correction and topographic distortion removal, used to help verify the topographic details of tea gardens and the distribution characteristics of terraced fields. Image data with a spatial resolution of at least 0.5 meters are preferred through commercial satellite remote sensing, drone aerial photography, and other methods.

[0053] It should be noted that the dominant direction vector of the slope This is a unit vector representing the main direction of terrace extension in a tea garden area, reflecting the distribution trend of terraces along the slope aspect. Based on a digital elevation model, the slope aspect information of the entire region is extracted. It is determined by statistically analyzing the dominant mode of slope aspect distribution; that is, the slope aspect of all slope units is statistically analyzed, and the direction with the highest frequency is taken as the dominant direction. This is then standardized into a unit vector (module length of 1). For example, if the slope aspect of a tea garden is mainly concentrated in the northeast-southwest direction, then... This is the unit vector corresponding to that direction.

[0054] It should be noted that spatial location These are the planar coordinates of any point within the tea garden area, used to locate the terrain. The coordinate system adopts a nationally unified Cartesian coordinate system (such as the CGCS2000 plane coordinate system), and the coordinates of any point can be obtained through positioning technologies such as GPS and BeiDou.

[0055] It should be noted that the slope aspect projection coordinates Spatial point Dominant direction vector on the slope The projection values ​​on the surface are used to transform two-dimensional planar coordinates into a one-dimensional ordered sequence. This is calculated through the inner product operation of the spatial point coordinates and the dominant direction vector of the slope, i.e. The value is a real number.

[0056] It should be noted that the micro-unit segmentation length This is a fixed length used to divide the terrace planting micro-units along the dominant direction of the slope, ensuring that all micro-units are of uniform scale. The preferred value is 2 meters, which is determined based on the typical geometric scale of the terrace (approximately 1.2 meters wide and 0.8 meters high). This value can fully cover the individual terrace structural unit while ensuring the fineness of the micro-unit division, avoiding the neglect of topographic heterogeneity due to excessively long divisions or the computational redundancy caused by excessively short divisions.

[0057] It should be noted that terraced planting micro-units It is a set of discrete points formed by equally spaced division along the dominant direction of the slope, and it serves as the basic unit for all subsequent parameter calculations and risk assessments. Each micro-unit is defined as satisfying All spatial points, among which The index of the segment (a non-negative integer, such as 0, 1, 2, etc.).

[0058] It should be noted that the segment sequence number index It is a non-negative integer identifier used to uniquely identify each terrace planting micro-unit, with values ​​starting from 0 and increasing sequentially until the entire tea garden area is covered. The assignment rule is to number each equally spaced segment sequentially according to the slope aspect projection coordinates in ascending order.

[0059] It should be noted that terraced planting micro-units It is the basic processing unit for analyzing the repeating spacing of terraced landforms, and... The essence remains the same, only the descriptions are simplified in different technical steps, for focusing on the aspect profile analysis of individual micro-units. Each Corresponding unique segment sequence number index This includes the tea garden area corresponding to a fixed range of slope aspect projection coordinates.

[0060] It should be noted that the repeating spacing of the terraced landforms It is a single terraced planting micro-unit The repeat length of the terrace structure along the slope is the distance between the centers of two adjacent terrace faces (or walls). It is calculated using a high-order program sequence of slope curvature, and then determined through autocorrelation analysis. The optimal value range is 1 to 3 meters (consistent with the typical geometric scale of mountain terraces), for example, a micro-unit. The meter indicates that the terraced fields in this area have a structure of terrace surface + terrace wall that repeats every 2 meters along the slope.

[0061] It should be noted that the fixed sampling interval This is the interval between data collection of elevation data along the slope profile of the micro-units planted in the terraced fields, used to obtain a continuous elevation sequence. The preferred value is 0.1 meters, which ensures the continuity of the elevation sequence, accurately captures elevation changes at the terrace walls, and avoids data redundancy due to overly dense sampling.

[0062] It should be noted that high-level sequences It is a sequence of elevation data collected along the slope profile of micro-units planted in terraced fields, used to construct a slope curvature sequence. The sampling point index (a non-negative integer). For the first Elevation values ​​of each sampling point (unit: meters).

[0063] It should be noted that the slope curvature sequence Based on high-program sequence The sequence obtained through second-order difference calculation is used to highlight the differences in elevation changes between the terrace walls and the terrace surface. Sampling point index (values ​​from 1 to ...) , (Total number of sampling points) The sign and magnitude of the value reflect the direction and degree of curvature of the slope. The slope wall exhibits a larger curvature value due to abrupt changes in elevation, while the curvature value is smaller at the surface of the slope. For example, the sampling point index corresponding to a certain slope wall... ,That It is significantly greater than the curvature value of the sampling point at the trapezoidal surface.

[0064] It should be noted that the sampling point index It is used to identify high-order columns. and slope curvature sequence The value is a non-negative integer representing a single sampling point, starting from 0, with a maximum value of 1 less than the total number of sampling points. The assignment rule is to number the points sequentially along the slope profile, starting with the first sampling point. Subsequent sampling points are arranged at fixed sampling intervals. Increasing sequentially.

[0065] It should be noted that the curvature autocorrelation function It is a measure of the slope curvature sequence Different lag steps A similarity function is used to identify the repetition cycle of terraced structures. This is the autocorrelation lag step (a non-negative integer). The range of values ​​is ,when When the corresponding distance is equal to the repeating interval of the terraced landform, A peak will occur.

[0066] It should be noted that the autocorrelation lag steps It is the curvature autocorrelation function The input parameter represents the number of steps between two curvature values ​​in the sequence. The value is a non-negative integer.

[0067] It should be noted that the preset search range It is about finding the curvature autocorrelation function. The range of lag steps for the first significant non-zero peak is used to define the search interval for the repeating spacing of the terraced landforms. A preferred value is... , (Corresponding to an actual distance of 0.5 meters to 10 meters), this range covers the typical repeating scale of mountain terraces, which can effectively eliminate false peaks caused by noise.

[0068] It should be noted that peak lag... It is the curvature autocorrelation function Within the preset search range The number of lag steps corresponding to the first significant non-zero peak is used to calculate the terraced landform repetition spacing. By traversing the preset search range Value, select to make The maximum number of lag steps.

[0069] It should be noted that the reservation time window It is the time interval used to statistically analyze the sampling characteristics of multi-source heterogeneous sensing data, defined as ,in The start time of the time window. The time window length. Preferred time window length. The time frame of 1 hour (i.e., 1 day) can cover the daily variation cycle of the microclimate in the mountain tea garden (such as the drainage of cold air at night and the dissipation of fog during the day) and ensure the sufficiency of effective observation data.

[0070] It should be noted that the number of multiple data sources It is the scheduled time window The total number of data source types for collecting data from micro-units of terraced fields is used to fuse the spatial resolution of different observation methods. A common value is 4, corresponding to four typical data sources: satellite remote sensing (Sentinel-2), synthetic aperture radar (Sentinel-1), unmanned aerial vehicles (UAVs), and ground sensors.

[0071] It should be noted that single-source spatial resolution It is the first Spatial observation accuracy of a data source refers to the distance between two adjacent observation points (or pixels). The optimal value differs for different data sources: ground sensors... meters, drone Meters, Synthetic Aperture Radar (Sentinel-1) Meters, satellite remote sensing (Sentinel-2) rice.

[0072] It should be noted that the source weight coefficient It is the first The contribution weight of different data sources to the calculation of the overall spatial resolution is used to reflect the observation reliability of different data sources. The preferred values ​​are: Ground sensor... drones Synthetic Aperture Radar Satellite remote sensing A higher weight value indicates higher observation accuracy and reliability of the data source.

[0073] It should be noted that the overall spatial resolution It is the scheduled time window A fusion index of spatial sampling capabilities from multiple data sources is used to quantify the micro-units of terraced planting. The spatial observation accuracy is calculated by taking the square root of the reciprocal of the square of the single-source spatial resolution, and the result is a positive number (unit: meters).

[0074] It should be noted that the effective observation timestamp sequence It is the scheduled time window The set of observation times arranged in chronological order after removing missing data. For the number of valid observations, For the first The time of each valid observation (format: year-month-day hour:minute:second).

[0075] It should be noted that the number of valid observations It is a valid observation timestamp sequence The total number of observation times reflects the predetermined time window. The continuity of internal data observations. Its values ​​are affected by factors such as cloud cover, equipment malfunction, etc., and optimal selection is crucial. (i.e., at least 8 valid observations within 24 hours).

[0076] It should be noted that the time difference set It is the set of time intervals between two adjacent observation timestamps in the effective observation timestamp sequence, used to calculate the comprehensive time acquisition interval. Timestamp index (values ​​from 1 to ...) ), For the first A time interval (unit: hour or minute).

[0077] It should be noted that the overall time acquisition interval It is the scheduled time window The median of the effective observation time interval within the terraced fields is used to quantify the planting micro-units. The frequency of time observations. Calculate the median of the set of time differences, taking a positive value (in hours).

[0078] It should be noted that the timestamp index It is used to identify valid observation timestamp sequences. A positive integer representing a single observation time, ranging from 1 to... The assignment rule is to number them sequentially according to the timestamp sequence, with the earliest observation time being... The latest one is .

[0079] It should be noted that the temperature series Terraced planting micro-units One of the microclimate time series, recording a predetermined time window Near-surface temperature data at different times within the region are used to calculate the duration of risk processes. For the observation time, For a moment Temperature value (unit: °C).

[0080] It should be noted that the temperature series after mean removal... It is a temperature sequence The series, after removing slow-changing trends, is used to focus on risk-related temperature fluctuations. Temperature series calculation. within the scheduled time window The mean value within the range is then subtracted from the temperature value at each time point, i.e. .

[0081] It should be noted that the normalized autocorrelation function It measures the temperature series after the mean has been removed. Different lag steps A similarity function is used to determine the duration of a risk process. The number of lag steps (a non-negative integer). The range of values ​​is The closer the value is to 1, the higher the similarity of the sequences at that lag step.

[0082] It should be noted that the preset attenuation threshold (Approximately equal to 0.3679) is the criterion for normalized autocorrelation function. The critical value for whether there is significant decay is used to determine the minimum number of lag steps. This threshold is set based on statistical regularities and can robustly distinguish between significantly correlated and irrelevant states of a sequence. At that time, it is considered that the sequences no longer have significant similarity at that lag step number.

[0083] It should be noted that the minimum lag step count It is the normalized autocorrelation function Attenuation to the preset attenuation threshold The following minimum lag steps are used to calculate the duration of a risk process. (Iteration) The value of is selected to make The smallest value.

[0084] It should be noted that the risk process lasts for a considerable period of time. Terraced planting micro-units Typical duration of internal risk-related microclimate processes (such as cold air convergence and frost). Minimum lag steps. With integrated time acquisition interval The product of, i.e. .

[0085] It should be noted that the first ratio It is the overall spatial resolution Repeating spacing with terraced landforms This is a derived index used to quantify the degree of matching between spatial sampling capability and terraced structure. The value is positive; when the value is greater than 1, it indicates that the spatial sampling resolution is insufficient to cover the cyclical structure of the terraces, resulting in spatial aliasing.

[0086] It should be noted that the second ratio It is the comprehensive time acquisition interval With the continuous timeliness of risk processes This is a derived indicator used to quantify the degree of matching between time sampling capability and the dynamics of risk processes. The value is positive; when the value is greater than 1, it indicates that the time sampling frequency is insufficient to capture changes in risk processes, resulting in time aliasing.

[0087] It should be noted that the spatiotemporal sampling aliasing index It is a quantified terraced planting micro-unit within the scheduled time window The core indicator of the degree of aliasing in internal observation signals reflects the level of degradation in observability. The maximum value between the first ratio and the second ratio is selected, and the value is positive; the larger the value, the more severe the aliasing.

[0088] It should be noted that the attenuation adjustment constant It is a factor for controlling the reliability decay of monitoring. A preset constant for the decay rate is used to match the relationship between aliasing level and reliability decay. A preferred value is 0.7, which has been calibrated using multiple sets of measured data to ensure reliability. The range of variation is reasonable, and will not be affected by... Excessive size leads to rapid decay of the reliability, and will not be affected by Too small a value results in insignificant reliability adjustment.

[0089] It should be noted that the reliability decay factor is monitored. Based on the spatiotemporal sampling aliasing index The generated reliability metrics are used to characterize the reliability of the observed data. They are calculated using the natural exponential function, i.e. The value is a positive number. The larger The larger the value, the lower the reliability.

[0090] It should be noted that the monitoring signal These are physical quantity signals characterizing the environmental state of micro-units in terraced planting, covering monitoring objects such as canopy temperature, near-ground relative humidity, and leaf surface moisture. For space-time coordinates, These are the slope aspect projection coordinates. For the observation time, This represents the signal value at the corresponding coordinates.

[0091] It should be noted that the coefficient of the slowly varying background term , , These are the coefficients of the slowly varying background term in the signal restoration basis function model, used to fit long-term, slow fluctuations such as seasonal changes and human intervention. The coefficient values ​​change over time at the observation time and are obtained by optimizing the objective function.

[0092] It should be noted that the harmonic order This is the order of the periodic harmonic term in the signal reconstruction basis function model, used to cover the main periodic frequencies of the terraced structure. A value of 3 is preferred, as this order accurately captures the fundamental period of the terraces (…). It can also take into account the influence of higher-order harmonics and avoid the effects of... Too small a value leads to insufficient fitting of the periodic signal or Excessive size leads to computational complexity.

[0093] It should be noted that the periodic harmonic term coefficient , These are the coefficients of the periodic harmonic terms in the signal reconstruction basis function model, used to fit the high-frequency fluctuations corresponding to the terraced structure. The harmonic order (values ​​from 1 to 1) ), The coefficient values ​​change over time at the observation time and are obtained by optimizing the objective function.

[0094] It should be noted that the weight matrix This is the matrix used in the objective function to distinguish between valid observations and missing data, thereby improving the accuracy of data fitting. The matrix dimension is the same as the number of observations. The matrix elements corresponding to valid observations are weighted according to the noise level (the lower the noise, the higher the weight), while the matrix elements corresponding to missing data are 0.

[0095] It should be noted that the observed values It is the scheduled time window The vector of actual observation results from the multi-source heterogeneous sensing data is used to construct the data fitting error term by comparing it with the model fitting results. The vector dimension is consistent with the number of valid observation data, and each element is the actual observation value of the corresponding observation point.

[0096] It should be noted that the perceptual mapping matrix It is to recover the parameter vector of the basis function model of the signal. This is a matrix mapped to the observation space, used to construct the relationship between the model fit and the actual observations. The matrix dimension is the number of valid observations × the parameter vector dimension, and each element is the value of the basis function at the corresponding observation point. For example, if the basis function has 5 parameters and there are 3 valid observation points, then... It is a 3×5 matrix, and the element values ​​are determined by the calculation results of the basis functions at each observation point.

[0097] It should be noted that the restoration model parameter vector It is the set of all coefficients in the signal restoration basis function model, including the coefficients of the slowly varying background term and the coefficients of the periodic harmonic term, i.e. The vector dimension is ( (This refers to the harmonic order).

[0098] It should be noted that the optimal restoration model parameter vector This is the parameter vector of the restoration model that minimizes the optimization objective function, used to reconstruct accurate restoration monitoring signals. It is solved using penalized least squares, i.e. Vector dimension and Consistent.

[0099] It should be noted that the regularization coefficient These are the coefficients of the smoothing regularization term in the objective function, used to suppress the parameter vector of the restored model. To avoid drastic fluctuations and overfitting, a value of 0.1 is preferred (which can be fine-tuned according to the data noise level). This value balances data fitting error and parameter smoothness, preventing excessive fluctuations. Too large a value leads to underfitting of the model, and will not be affected by... Too small a value will cause excessive fluctuations in parameters.

[0100] It should be noted that the difference operator It is a matrix operator used in the objective function to calculate the smoothness of the parameter vector, and is used to measure the parameter vector of the restored model. The rate of change. Matrix dimension and parameter vector. The dimensions are consistent and can be constructed using first- or second-order differences.

[0101] It should be noted that the monitoring signal was restored. Based on the optimal restoration model parameter vector The reconstructed signal is used to eliminate aliasing distortion and restore the true risk-related signal. Substitute the signal back to restore the basis function model, and plant micro-units in terraced fields. Take the average value above. The physical meaning and the original monitoring signal Consistent.

[0102] It should be noted that the reservation time window length It is the time span of the predetermined time window, that is... In Used to calculate wave energy The preferred value is 24 hours (unit: hour), which corresponds to the predetermined time window. The preferred lengths are consistent.

[0103] It should be noted that the mean of the restored signal within the window It is a restoration monitoring signal within the scheduled time window The average value within a certain range is used to calculate the fluctuation energy of the signal. exist Integral by dividing ,Right now .

[0104] It should be noted that wave energy It is a restoration monitoring signal within the scheduled time window A quantitative indicator of the intensity of abnormal fluctuations within the region is used to characterize the potential level of risk. Calculation With window mean The square of the difference in Integral within, then divided by ,For example When the fluctuations are large, A larger value indicates a higher potential level of risk.

[0105] It should be noted that the reliability-corrected risk value It is a risk quantification index normalized by observational reliability, used to accurately characterize the micro-units of terraced planting. The actual level of risk. Fluctuation energy. With monitoring reliability decay factor The ratio, i.e. .

[0106] It should be noted that logical functions The sigmoid function is used to adjust the confidence level for risk. The activation function, mapped to the early warning probability, is used to transform the risk value into a probability value in the interval [0,1]. Its expression is: ,in For input variables (including) and (linear combination).

[0107] It should be noted that the model constants , , These are preset parameters of the early warning probability logic function, used to calibrate the relationship between risk value, aliasing index, and early warning probability. They are obtained through training with multiple sets of measured data (including cases where risks occurred and cases where they did not), and the optimal values ​​are [values ​​to be filled in]. , , (This can be slightly adjusted based on data from different tea-producing regions).

[0108] It should be noted that the natural logarithm operation It is used in the early warning probability logic function to handle the spatiotemporal sampling aliasing index. Mathematical operations used for smoothing Extreme fluctuations. Its function is to... The large-scale changes are transformed into relatively gentle logarithmic changes, avoiding the influence of... Excessive values ​​can cause sudden changes in the probability of issuing a warning.

[0109] It should be noted that the probability of early warning Based on reliability-corrected risk value With spatiotemporal sampling aliasing index The calculated probability of risk occurrence is used to generate early warning events. This is achieved through a logical function mapping, i.e. The value ranges from [0,1]. The closer the value is to 1, the higher the probability of the risk occurring.

[0110] It should be noted that this applies to specific batches of tea. This refers to a batch of tea products harvested and processed from tea gardens within a specific timeframe; it is the core object of traceability. Each batch Corresponding unique batch identifier It covers the entire process from harvesting to sales.

[0111] It should be noted that the amount of contribution Terraced planting micro-units For a specific batch of tea The total physical contribution reflects the degree of participation of micro-units in batch production. It is obtained through statistical analysis of harvesting record events, including measurement indicators such as fresh leaf weight and the number of harvested baskets, with fresh leaf weight (unit: kilogram) being the preferred statistical standard.

[0112] It should be noted that the batch Corresponding time window It is related to a specific batch of tea. The time window corresponding to the harvesting process is used to determine the monitoring reliability decay factor during the production period of that batch. Its time range covers the entire batch. The optimal harvesting cycle length and predetermined time window. Consistent (24 hours).

[0113] It should be noted that the intermediate weight Terraced planting micro-units For a specific batch of tea The initial responsibility weights are used for subsequent normalization processing. Micro-unit contribution. With this micro unit Internal monitoring reliability decay factor The ratio, i.e. .

[0114] It should be noted that the contribution batch micro-unit set It is all to a specific batch of tea Provides a collection of micro-units for terraced planting of fresh leaves, used for unified batch calculation. The responsibility attribution weights. Elements in the set are micro-unit identifiers (such as...). , wait).

[0115] It should be noted that the weighting of responsibility attribution... Terraced planting micro-units For a specific batch of tea The final responsibility percentage is used to accurately attribute production responsibility. (Intermediate weighting) and The ratio of the sum of the intermediate weights of all micro-units is taken in the range [0,1], and the sum of the weights of all micro-units is 1.

[0116] It's important to note that Quick Ledger is a ledger used in the blockchain's dual-ledger structure to store low-load, high-frequency query data, ensuring efficient traceability queries. Its stored content focuses on core, critical fields and does not include massive amounts of raw data, enabling rapid response to user traceability query requests.

[0117] It should be noted that the evidence ledger is the ledger used in the blockchain's dual-ledger structure to store complete evidence data, ensuring the auditability and recalculation of the tracing process. The stored content includes complete evidence such as the addressing identifier of the original data and the hash of the restoration parameters. Although its query frequency is lower than that of the fast ledger, it can provide authoritative verification for tracing results.

[0118] It should be noted that the Quick Ledger transaction payload It is a structured data carrier written to the Fast Ledger, containing core identifiers of events and batches, key calculation results, and data fingerprints. Its structure is defined as follows: Each field contains low-load information.

[0119] It should be noted that the evidentiary ledger transaction payload It is a structured data carrier written into the evidence ledger, containing event identifiers, complete evidence addressing information, and core constraint parameters. Its structure is defined as follows: It is used to link complete evidence with core parameters.

[0120] It should be noted that the event identifier This is a string identifier used to uniquely identify each event in the entire process, such as early warning events and data collection events, to achieve precise association between the quick ledger and the evidence ledger. The encoding rule is event type + date + sequence number.

[0121] It should be noted that batch identification It is used to uniquely identify a specific tea batch. This string identifier is used to associate an event with its corresponding tea batch. The encoding rule is batch number + date + serial number.

[0122] It should be noted that timestamps This records the time information of an event, used to clarify the time at which the event occurred. The format is uniformly year-month-day hour:minute:second.

[0123] It should be noted that the monitoring unit This is a micro-unit identifier for the terraced planting area corresponding to the event, used to clearly identify the spatial location of the event. Identification rules are consistent with those for terraced planting micro-units. Consistent, directly using the segment sequence number index. Logo.

[0124] It should be noted that the event payload hash It is the hash fingerprint of the original event payload (sensor data, image data, etc.), used to verify the consistency between the on-chain data and the original data. It is calculated using the SHA256 hash algorithm and is a 64-bit hexadecimal string.

[0125] It should be noted that raw data refers to the original acquisition results of multi-source heterogeneous sensing data, including satellite remote sensing imagery, UAV imagery, and time-series data from ground sensors, used for recalculation and verification during the traceability process. Due to the large volume of data, it is not directly written to the blockchain but is stored in a distributed file system.

[0126] It should be noted that a distributed file system is a decentralized storage system used to store raw data, with IPFS (InterPlanetary File System) being the preferred choice. It features content addressing, decentralization, and tamper resistance. Each uploaded raw data file receives a unique content address identifier. ,pass The original data can be retrieved at any time.

[0127] It should be noted that the content addressing identifier It is a unique identifier for the original data within the distributed file system, used to associate the original data with the blockchain. It is generated by the distributed file system based on the content of the original data and is in string format.

[0128] It should be noted that the hash fingerprint of the restored model parameters It is the parameter vector of the optimal restoration model. The hash value is used for verification. The integrity and tamper-proof nature of the data are verified. The hash value is calculated using the SHA256 hash algorithm and is a 64-bit hexadecimal string.

[0129] It is important to note that all input data described in this solution is acquired in real-time through legal and compliant hardware interfaces with the user's full knowledge, explicit consent, and active cooperation. The preset parameters, prior constants, and statistical means are all derived from publicly available scientific literature data, de-identified general research datasets, or calibration data from laboratory environments, and do not contain any unauthorized sensitive third-party information. The system's data processing is limited to local or volatile memory computation transmitted via encrypted channels. There is no illegal collection, theft, or retention of user biometric data or infringement of user privacy without the user's knowledge. All parameter calls and generation comply with the principles of data minimization, legality, legitimacy, and necessity.

[0130] 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. A digital early warning and blockchain-based responsibility traceability method for the entire tea production process, characterized in that: include: Construct a distribution map of terraced planting micro-units and analyze the terraced landform repeating spacing along the slope of each terraced planting micro-unit; Acquire multi-source heterogeneous sensing data for the terraced planting micro-units, set a predetermined time window, calculate and determine the comprehensive spatial resolution and comprehensive temporal acquisition interval within the predetermined time window, and obtain the duration of risk processes for the terraced planting micro-units; Based on the ratio of the comprehensive spatial resolution to the terraced landform repetition spacing, and the ratio of the comprehensive temporal acquisition interval to the duration of the risk process, the spatiotemporal sampling aliasing index is calculated. The monitoring reliability attenuation factor is generated using the spatiotemporal sampling aliasing index, and the monitoring reliability attenuation factor is used as a constraint condition to perform anti-aliasing signal restoration processing on the multi-source heterogeneous sensing data to obtain restoration model parameters and reliability correction risk values. An early warning event is generated based on the reliability correction risk value, and the corresponding responsibility traceability allocation weight for the tea batch is calculated based on the monitoring reliability decay factor. The early warning event, the responsibility traceability allocation weight, and the restoration model parameters are stored in the blockchain system for traceability.

2. The method for digital early warning and blockchain-based traceability of the entire tea production process according to claim 1, characterized in that, Construct a distribution map of micro-units for terraced planting, including: A digital elevation model and high-resolution orthophotos of the tea garden area were collected, and the dominant direction vector of the slope was identified based on the digital elevation model. For any spatial point within the tea garden area, its coordinates are projected onto the dominant direction vector of the slope to calculate the slope aspect projection coordinates. A fixed micro-unit segmentation length is set, and the tea garden area is equally segmented along the axis of the dominant direction vector of the slope using the slope aspect projection coordinates to generate multiple discrete sets of terraced planting micro-units. The distribution map of terraced planting micro-units is constructed based on the multiple discrete sets of terraced planting micro-units.

3. The method for digital early warning and blockchain-based traceability of the entire tea production process according to claim 1, characterized in that, Analysis of the terraced landform repeating intervals along the slope of each terraced planting micro-unit, including: Sampling was performed along the slope profile of each of the terraced planting micro-units to obtain high-order sequences; Based on the high-order program sequence, the second-order difference is calculated to construct the slope curvature sequence; Perform autocorrelation analysis on the slope curvature sequence to calculate the curvature autocorrelation function; Within a preset search range, the first significant non-zero peak of the curvature autocorrelation function is identified, and the lag distance corresponding to the first significant non-zero peak is used as the repeating interval of the terraced landform.

4. The method for digital early warning and blockchain-based traceability of the entire tea production process according to claim 1, characterized in that, The calculation and determination of the integrated spatial resolution and integrated temporal acquisition interval within the predetermined time window includes: Obtain multiple data sources for the terraced planting micro-units within the predetermined time window, obtain the single-source spatial resolution of each data source, and set the corresponding source weight coefficient. Calculate the reciprocal square of the spatial resolution of each single source, sum the reciprocals of the squares using the source weight coefficients, and take the square root of the reciprocal of the weighted sum to obtain the overall spatial resolution; Obtain the sequence of valid observation timestamps arranged in chronological order within the predetermined time window, and calculate the set of time differences between two adjacent valid observation timestamp sequences; Calculate the median of the time difference set and use the median as the integrated time acquisition interval.

5. The method for digital early warning and blockchain-based traceability of the entire tea production process according to claim 1, characterized in that, The duration of the risk process of the terraced planting micro-units is obtained, and the spatiotemporal sampling aliasing index is calculated, including: The microclimate time series of the terraced planting micro-units is obtained and the mean is removed, and the normalized autocorrelation function is calculated. Find the minimum hysteresis step when the normalized autocorrelation function decays to below a preset decay threshold, and use the product of the minimum hysteresis step and the comprehensive time acquisition interval as the duration of the risk process; Calculate a first ratio of twice the overall spatial resolution to the terraced landform repeating interval, and calculate a second ratio of twice the overall temporal acquisition interval to the duration of the risk process; The maximum value between the first ratio and the second ratio is selected as the spatiotemporal sampling aliasing index.

6. The method for digital early warning and blockchain-based traceability of the entire tea production process according to claim 1, characterized in that, The monitoring reliability attenuation factor is generated using the spatiotemporal sampling aliasing index, including: Set an attenuation adjustment constant and calculate the difference between the spatiotemporal sampling aliasing index and the value one; Calculate the product of the attenuation adjustment constant and the difference, use the product as an exponent to calculate the natural exponential function value, and use the natural exponential function value as the monitoring reliability attenuation factor.

7. The method for digital early warning and blockchain-based traceability of the entire tea production process according to claim 1, characterized in that, Using the monitoring reliability attenuation factor as a constraint, anti-aliasing signal restoration processing is performed on the multi-source heterogeneous sensing data to obtain restoration model parameters and reliability correction risk values, including: Construct a signal restoration basis function model that includes a slowly varying background term and a periodic harmonic term based on the repeating interval of the terraced landform; Construct an optimization objective function, which includes a data fitting error term and a smoothing regularization term weighted by the monitoring reliability decay factor; Solve for the restoration model parameters that minimize the optimization objective function, and use the restoration model parameters to reconstruct the restoration monitoring signal of the terraced planting micro-unit; Calculate the fluctuation energy of the restored monitoring signal within the predetermined time window, calculate the ratio of the fluctuation energy to the monitoring reliability attenuation factor, and use the ratio as the reliability correction risk value.

8. The method for digital early warning and blockchain-based traceability of the entire tea production process according to claim 1, characterized in that, Based on the aforementioned reliability-corrected risk value, an early warning event is generated, and the corresponding responsibility traceability allocation weight for the tea batch is calculated according to the aforementioned monitoring reliability decay factor, including: Construct a warning probability logic function that includes a linear term of the confidence correction risk value and a logarithmic term of the spatiotemporal sampling aliasing index, calculate the warning probability, and generate the warning event based on the warning probability; The contribution of the terraced planting micro-unit to the specific tea batch is obtained, and the ratio of the contribution to the monitoring reliability decay factor is calculated to obtain the intermediate weight. The intermediate weights of all terraced planting micro-units belonging to the same tea batch are normalized to obtain the responsibility traceability allocation weight of each terraced planting micro-unit for that tea batch.

9. The method for digital early warning and blockchain-based responsibility traceability of the entire tea production process according to claim 8, characterized in that, The warning event, the responsibility tracing and allocation weights, and the recovery model parameters are stored in the blockchain system for traceability, including: Construct a dual-ledger storage structure that includes a fast ledger and an evidence ledger; Extract the event identifier, timestamp, and corresponding event payload of the warning event, and extract the batch identifier of the specific tea batch; Write the event identifier, the batch identifier, the timestamp, the spatiotemporal sampling aliasing index, the warning probability, and the hash fingerprint of the event payload into the fast ledger; The multi-source heterogeneous sensing data is written as raw data into the distributed file system and the content addressing identifier is obtained. The content addressing identifier, the hash fingerprint of the restoration model parameters, and the monitoring reliability decay factor are written into the evidence ledger.

Images