Method and system for dynamic pricing and tracing of agricultural products based on integration of plant growth mechanism and multi-dimensional spatio-temporal data

Abstract

The application belongs to the field of agricultural information technology and big data processing technology, and relates to a dynamic pricing and tracing method and system for agricultural products by integrating crop growth mechanism and multi-dimensional space-time data. The method improves the authenticity and credibility of the tracing data by collecting multi-source heterogeneous data, combining the data organization architecture constructed by space-time index and data completion processing. The method generates a unique full-link tracing identifier as the index core to construct a digital archive of the full life cycle of agricultural products, establishes a cross-node and cross-system data correlation architecture, and realizes automatic and accurate correlation of full-link data and construction of complete evidence chain. The method extracts quality key features from the digital archive and predicts the quality status through an algorithm model, combines market, supply and demand, and credit data to construct a multi-factor pricing model, converts environmental data into quantifiable quality indicators and scientific dynamic guidance prices, solves the problems of lack of quality quantization and reasonable pricing mechanism in the prior art, and realizes digital rights protection of agricultural assets.

Description

Technical Field

[0001] This application belongs to the field of agricultural information technology and big data processing technology, and more specifically, it relates to a method and system for dynamic pricing and traceability of agricultural products that integrates crop growth mechanisms and multi-dimensional spatiotemporal data. Background Technology

[0002] With the rapid development of the digital economy and the upgrading of consumption structure, consumers' demand for agricultural products has shifted from simply "food quality" to "food safety." They are paying significantly more attention to information such as the authenticity of the origin, the growing environment, and the intrinsic quality of agricultural products. Against this backdrop, agricultural product traceability technology has emerged as a core means to ensure food safety and enhance brand credibility. It also provides important support for the digital transformation of regional public brands, helping to solve the market pain point of high-quality agricultural products not fetching high prices.

[0003] Currently, the agricultural product traceability system has gone through two main development stages: the first stage is the identification stage based on barcodes / QR codes, which assigns a unique ID to agricultural products, records the basic information of producers and product attributes, and realizes the basic traceability function of knowing the place of origin by scanning the code; the second stage is the process tracking stage based on RFID and logistics information, which focuses on recording the node information of agricultural products in the processing, storage and logistics links to ensure that the flow of products can be traced. The application of the above technologies has solved the basic problem of traceability of agricultural product sources to a certain extent and promoted the initial development of agricultural digitalization.

[0004] However, in practical industrial applications, existing traceability technologies still face many core bottlenecks, making it difficult to meet the needs of full life-cycle management and scientific value assessment of agricultural products. Specifically, this is reflected in the following three aspects, which limit the formation of a premium pricing mechanism for high-quality agricultural products: First, the data source is too singular and it is difficult to verify the authenticity of the data. Existing traceability systems mainly rely on text data such as fertilization records and delivery notes entered manually. They lack the ability to integrate unstructured multi-source heterogeneous data such as environmental perception data obtained by IoT sensors and image data obtained by drones / satellites / cameras. The single manual data lacks cross-verification with objective data, resulting in low data production costs and insufficient traceability credibility.

[0005] Secondly, there is a lack of data-based quality quantification and value assessment mechanisms. Existing systems only serve as data recorders and cannot uncover the value behind the data. The systems cannot use accumulated growth environment data (such as accumulated temperature and light) to quantify and rate the intrinsic quality of agricultural products through algorithmic models, nor can they provide dynamic guidance prices based on market conditions. This results in good products lacking data support and failing to obtain reasonable brand premiums in the market.

[0006] Third, the data correlation across the entire chain is weak, resulting in information silos. Throughout the entire life cycle from planting to sales, due to the lack of a unified coding standard, resource data (land environment), subject data (farmer qualifications), and product data (circulation batches) are often separated in different systems, making it difficult to achieve automated and accurate correlation and confirmation of rights. This leads to a break in the chain of evidence and makes it impossible to form complete digital assets. Summary of the Invention

[0007] This invention provides a method and system for dynamic pricing and traceability of agricultural products that integrates crop growth mechanisms and multidimensional spatiotemporal data. It aims to solve the technical problems in agricultural product traceability technology, such as the susceptibility of data sources to tampering, the difficulty in spatiotemporal alignment of multi-source heterogeneous data, the lack of a quality quantification assessment mechanism based on the growth environment, and the weak correlation of data across the entire chain.

[0008] On the one hand, this invention provides a method for dynamic pricing and traceability of agricultural products that integrates crop growth mechanisms and multidimensional spatiotemporal data, including the following steps: Multi-source heterogeneous data acquisition: Collect environmental perception time-series data, unstructured image data, and structured business management data of the target agricultural products throughout their entire life cycle from production and processing to distribution; Data fusion and traceability identifier generation: A data organization architecture is constructed based on spatiotemporal index. The collected multi-source heterogeneous data is aligned on the time axis and matched in spatial dimension. Missing value completion processing is performed on the environmental perception time series data. After data fusion is completed, a unique full-link traceability identifier is generated. With the full-link traceability identifier as the index core, a digital archive binding the entire life cycle data of agricultural products is established to realize a one-to-one correspondence between data and agricultural products. Quality assessment and dynamic pricing: Key feature information affecting the quality of agricultural products is screened and extracted from the digital archives, input into the trained algorithm model to predict the quality status of agricultural products, and then market benchmark information, supply and demand dynamics and production entity credit data are integrated to construct a multi-factor pricing model. Based on the quality prediction results and the weights of each factor, the dynamic guidance price of agricultural products is calculated. Data storage and information display: A hybrid architecture of off-chain storage and on-chain anchoring is adopted. The original data and digital archives are stored in a distributed file system. After hash digest processing of the data, a Merkle tree is constructed. Only the root hash is written to the blockchain to complete the storage. At the same time, a mapping relationship between the full-chain traceability identifier and the stored data is established for terminal devices to scan and parse. The traceability trajectory and value assessment information of agricultural products are displayed in a visual way.

[0009] This invention achieves precise spatiotemporal alignment and fusion of multi-source data by collecting environmental perception time-series data, unstructured image data, and structured business management data, combined with a data organization architecture constructed using spatiotemporal indexing and data completion processing. This effectively overcomes the shortcomings of existing technologies, such as single data sources and lack of fusion capabilities, and improves the authenticity and credibility of traceability data. Furthermore, by generating a unique end-to-end traceability identifier and using it as the core of the index to construct a digital archive of the entire lifecycle of agricultural products, a cross-node, cross-system data association architecture is established, breaking down information silos and achieving automated, precise association of end-to-end data and construction of a complete chain of evidence, thus solving the problem of end-to-end data correlation. This addresses the technical challenges of weak evidence chains and broken evidence chains. By extracting key quality features from digital archives and predicting quality status using algorithmic models, and combining market, supply and demand, and credit data to construct a multi-factor pricing model, environmental data is transformed into quantifiable quality indicators and scientifically dynamic guiding prices. This overcomes the predicament of existing technologies lacking quality quantification and reasonable pricing mechanisms, and the problem of high quality not commanding high prices. By adopting a hybrid architecture of off-chain storage and on-chain anchoring, storage costs are significantly reduced while ensuring data immutability and security. Combined with visualization, the value and traceability information of agricultural products are presented intuitively, solving the defects of high data storage costs, insufficient security, and poor display effects.

[0010] Preferably, the environmental sensing time-series data includes air temperature and humidity, soil parameters, and light intensity data, which are continuously collected at a preset frequency; The image data includes spectral images, radar data, and high-definition video data, which are collected regularly to record the growth and processing status of agricultural products. The business management data includes agricultural operation records, harvesting batch information, logistics documents, and quality inspection report data, which are extracted synchronously from the corresponding production management system.

[0011] Preferably, the spatiotemporal index adopts a data cube architecture, with time axis alignment achieved through a sliding window aggregation algorithm, and spatial dimension matching completed based on the coordinates of agricultural production plots. Data completion employs a differentiated strategy, with short-term missing data completed through interpolation, and long-term missing data retained after being marked with confidence levels.

[0012] Preferably, the full-chain traceability identifier is generated by integrating the production entity identifier, product batch identifier, geospatial code, and timestamp. The generation process involves hash operation and XOR operation in sequence. The geospatial code is converted from the coordinates of the center point of the agricultural product production plot into a fixed-precision spatial hash string to ensure the uniqueness and collision prevention of the identifier.

[0013] Preferably, the key characteristic information of the agricultural product quality includes the effective accumulated temperature, the diurnal temperature range cumulative index, the light availability characteristics, and the soil parameter compliance rate.

[0014] Preferably, the key characteristic information of agricultural product quality is extracted using pre-defined characteristic extraction rules for different categories of agricultural products. That is, environmental time-series data and image data that are directly related to the intrinsic quality of agricultural products are screened from the full life cycle digital archive, and extracted using a pre-defined algorithm in combination with the growth mechanism of the corresponding category of agricultural products.

[0015] Preferably, the algorithm model in the quality assessment and dynamic pricing adopts the gradient boosting algorithm or the random forest algorithm, takes the key feature information of the extracted agricultural product quality as input, takes the physical and chemical quality indicators of agricultural products as training targets, and realizes the quantitative prediction of the quality status of agricultural products after training with sample data.

[0016] Preferably, the multi-factor pricing model obtains the quality premium factor through the normalized quality prediction results, calculates the supply and demand factor by combining inventory balance and market search popularity, and combines the blockchain credit score of the production entity to assign corresponding weights to each factor before calculating the total premium coefficient. Finally, it superimposes the market benchmark price to obtain the dynamic guidance price.

[0017] Preferably, the data hash digest is extracted from the original data and digital archives, and only the root hash is uploaded to the chain after constructing the Merkle tree, which reduces the on-chain storage cost while ensuring that the data is immutable.

[0018] Preferably, the visualization is presented in the form of a value radar chart, which maps multi-dimensional data to corresponding coordinate axes to show the value difference of agricultural products relative to standard products and the traceability of the entire supply chain. The multi-dimensional data includes accumulated temperature, temperature difference, quality score, credit rating and price composition.

[0019] On the other hand, the present invention provides a dynamic pricing and traceability system for agricultural products that integrates crop growth mechanisms and multidimensional spatiotemporal data, including a processor and a memory. The memory stores a computer program that can run on the processor. The processor calls the computer program stored in the memory to execute the dynamic pricing and traceability method for agricultural products that integrates crop growth mechanisms and multidimensional spatiotemporal data described in the present invention.

[0020] The beneficial effects of this invention include: This invention achieves precise spatiotemporal alignment and fusion of multi-source data by collecting environmental perception time-series data, unstructured image data, and structured business management data, combined with a data organization architecture constructed using spatiotemporal indexing and data completion processing. This effectively overcomes the shortcomings of existing technologies, such as single data sources and lack of fusion capabilities, and improves the authenticity and credibility of traceability data. Furthermore, by generating a unique end-to-end traceability identifier and using it as the core of the index to construct a digital archive of the entire lifecycle of agricultural products, a cross-node, cross-system data association architecture is established, breaking down information silos and achieving automated, precise association of end-to-end data and construction of a complete chain of evidence, thus solving the problem of end-to-end data correlation. This addresses the technical challenges of weak evidence chains and broken evidence chains. By extracting key quality features from digital archives and predicting quality status using algorithmic models, and combining market, supply and demand, and credit data to construct a multi-factor pricing model, environmental data is transformed into quantifiable quality indicators and scientifically dynamic guiding prices. This overcomes the predicament of existing technologies lacking quality quantification and reasonable pricing mechanisms, and the problem of high quality not commanding high prices. By adopting a hybrid architecture of off-chain storage and on-chain anchoring, storage costs are significantly reduced while ensuring data immutability and security. Combined with visualization, the value and traceability information of agricultural products are presented intuitively, solving the defects of high data storage costs, insufficient security, and poor display effects. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 The overall system architecture diagram provided for embodiments of the present invention.

[0023] Figure 2 The following is a detailed flowchart provided for an embodiment of the present invention.

[0024] Figure 3 A schematic diagram of the data structure of the end-to-end value traceability code provided in an embodiment of the present invention. Detailed Implementation

[0025] To make the technical problems, technical solutions, and beneficial effects to be solved by this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and are not intended to limit the scope of this application.

[0026] Example 1

[0027] See Figure 1 and Figure 2As shown, the method for dynamic pricing and traceability of agricultural products, which integrates crop growth mechanisms and multidimensional spatiotemporal data, includes the following steps: S1. Multi-source heterogeneous data acquisition: Deploy a sensing layer at the target agricultural product production base to collect three types of heterogeneous data: Environmental time-series data: Through field IoT sensors, time-series data such as air temperature and humidity, soil electrical conductivity (EC value), and light intensity are collected at a preset frequency (e.g., every 10 minutes).

[0028] Impact on unstructured data: Regularly collect multispectral images, radar data, or high-definition video data during key crop growth periods (such as flowering and fruiting stages) using drones, satellites, or field cameras.

[0029] Structured business data: Text data such as agricultural operation records (e.g., fertilization, irrigation), harvest batches, and logistics tracking numbers are obtained from enterprise ERP or production management systems.

[0030] S2. Data Fusion and Traceability Identifier Generation: Addressing the challenges of inconsistent frequencies (e.g., high-frequency continuous IoT data, low-frequency discrete impact data, and event-triggered business data) and spatial heterogeneity in agricultural scenarios, this embodiment employs spatiotemporal cube mapping technology for data governance, specifically including the following steps: Build end-to-end value traceability code: such as Figure 3 As shown, the end-to-end value traceability code adopts a hierarchical hash structure. Its core is anchored to the product batch ID, and the outer layer surrounds the theme credit ID and resource plot GeoHash. To achieve a unique mapping from physical assets to digital assets, the generation algorithm of the end-to-end value traceability code (VTC) is defined, which adopts the SHA-256 hash algorithm, combined with geolocation-based hash (GeoHash) spatial encoding technology, and ensures the global uniqueness and collision resistance of the identifier through XOR operation. ; In the formula: The unified social credit code represents the production entity (enterprise); Indicates the product batch number; This means converting the latitude and longitude coordinates of the plot's center into a string with a precision of 9 bits, approximately 4.8 meters, thus achieving spatial dimension binding; This represents the XOR operation, used to increase the obfuscation and unpredictability of the encoding; This represents the assigned timestamp, used to prevent replay attacks and ensure the uniqueness of the time dimension.

[0031] Spatiotemporal alignment and missing data completion for heterogeneous frequency data: Constructing a data cube with VTC as the index key and time t and space s as axes; A standard time granularity Δt is set (1 hour in this embodiment). For high-frequency IoT data (sampling frequency of 10 minutes) in the environmental time series data, a sliding window aggregation algorithm is used to align them to the standard time axis. Specifically, for the k-th environmental feature within the time window [t, t+Δt], the aligned value... : ; In the formula: This represents the value of the k-th environmental feature at the i-th time point within the time window [t, t+Δt]; N is the total number of time points.

[0032] For low-frequency image data (sampling frequency) The forward hold strategy is employed, which associates the environmental data at time t with the feature vector of the most recently captured image on the time axis. Until new image data This generates a frequency mismatch problem.

[0033] To address data gaps caused by sensor malfunctions, if the number of consecutive missing points m < 6 (i.e., less than 6 hours), Lagrange interpolation is used for completion to preserve the smoothness of the growth curve. For example, let the missing point be x, and use the known data points before and after it. Constructing a polynomial: ; In the formula, L(x) represents the estimated value after completion; x represents the time point to be completed, (x i ,y i ) and (x j ,y j ) represents the nearest, valid sensor readings before and after the missing time point x; y j Indicates time x j The actual sensor value measured; k represents the number of reference points, which is assigned the total number of known data points participating in the interpolation calculation minus 1; This represents the weight term of the Lagrange polynomial.

[0034] For long-term missing data with m≥6, the system automatically marks the time period as "low data confidence" and does not perform forced completion to ensure the authenticity of the traceability.

[0035] S3. Assessment and Dynamic Pricing: Using artificial intelligence technology, the cleaned (i.e., the data processed in step S2) time-series environmental data is mapped to the intrinsic quality score and market guidance price of agricultural products. The specific steps are as follows: Feature engineering extraction algorithm: Extract key feature vectors X of image product quality (such as sugar content and hardness) from the cleaned environmental data (i.e., the data processed in step S2), including: effective accumulated temperature, average diurnal temperature range, number of frost-free days, soil trace element compliance rate, etc.; it should be noted that the feature vector extraction can be selected according to different agricultural products. Effective accumulated temperature : ; In the formula: This represents the cumulative effective accumulated temperature value; It indicates the number of days in the growing season, from the flowering period (start) to the harvest period (end); This indicates the biological zero temperature of the corresponding crop (e.g., Aksu apples are set to 10℃). This indicates the daily maximum / minimum temperature, which is the daily extreme temperature value collected by IoT weather stations.

[0036] Diurnal temperature range cumulative index : ; In the formula: Indicates the cumulative temperature difference index; This indicates the critical period of ripening, specifically the key time window for sugar conversion in fruit (such as 30 days before harvest). It indicates the temperature difference between day and night.

[0037] Light efficiency characteristics : ; In the formula: Indicates the light intensity per hour; This represents the cloud attenuation coefficient calculated using satellite cloud images; h represents the hour index, traversing the entire growth cycle.

[0038] Construct a feature vector X=[GDD, , , ,…](in A regression prediction model that takes soil electrical conductivity as input and physicochemical indicators (such as sugar content and fruit diameter) as target variable Y.

[0039] The regression prediction model uses either the XGBoost (eXtreme Gradient Boosting) algorithm or the random forest algorithm to train the model by minimizing the objective function. ; in Represents the loss function. Represents the training error term; i represents the sample index; Indicates the true label; Indicates the predicted label; This represents the regularization term (to prevent overfitting); k represents the index of the tree; The function structure of the k-th tree; This indicates the complexity of the k-th tree; the model outputs the predicted value. (e.g., predicted sugar content of 18.5 Brix).

[0040] Then, establish a dynamic pricing model based on three-dimensional factors: quality, supply and demand, and credit, and output a suggested guidance price. : ; In the formula: Indicates the premium factor; Indicates the market benchmark price; Among them, premium coefficient Calculated based on the following formula: ; In the formula: Indicates predicted quality; Indicates the national standard value; Indicates the current inventory level; This represents the overall online search popularity index. When search volume is high and inventory is low, the premium coefficient increases. This represents the performance credit score of the production entity on the blockchain. , and This represents the weighting coefficient, which is dynamically adjusted by the platform based on its operational strategy (e.g., ...). =0.6, =0.3, =0.1); This embodiment utilizes The function will predict quality Compared with national standard values The deviation is normalized to the (0,1) interval, reflecting "good products at good prices".

[0041] S4. Data Storage and Information Display: To address the issues of high cost and low efficiency in storing big data on the blockchain, a hybrid architecture of off-chain storage and on-chain anchoring is adopted.

[0042] Off-chain storage: Stores the raw image data and high-frequency IoT logs generated by S2 in a distributed file system (IPFS) and obtains the content addressing hash. .

[0043] On-chain evidence storage: Quality Assessment Report Summary The VTC code is used to construct leaf nodes, generate Merkle tree root hashes, and only the root hashes are written into the consortium blockchain blocks.

[0044] It ensures the immutability of all data (any change in any bit of data will cause the root hash to change), and reduces the cost of putting data on the chain by more than 99%.

[0045] After parsing the full-chain value traceability code (VTC) at the front end, the feature vector calculated by S3 is retrieved, which includes GDD (accumulated temperature), DTR (temperature difference), and Credit. Dimensions such as (predicted sugar content) are mapped onto the coordinate axes of the radar chart, intuitively displaying the "value envelope" of the product relative to the standard product.

[0046] Example 2 This embodiment provides a dynamic pricing and traceability system for agricultural products that integrates crop growth mechanisms and multidimensional spatiotemporal data. The system includes a processor and a memory. The memory stores a computer program that can run on the processor. The processor calls the computer program stored in the memory to execute the dynamic pricing and traceability method for agricultural products that integrates crop growth mechanisms and multidimensional spatiotemporal data described in Embodiment 1 above. The memory includes at least one type of readable storage medium, such as flash memory, card-type memory (e.g., SD or DX memory), magnetic memory, etc. In some embodiments, a processor may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits packaged with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips.

[0047] Experimental Example 1: This embodiment selects the core experimental area (Area A, center coordinates E:80.2°, N:41.1°, area of ​​500 mu) of Hongqipo Farm in Aksu region, Xinjiang as the implementation object, and the implementation period covers the entire growing season of 2025 (from flowering period in April to harvesting period in October) to verify the complete process of the method described in this invention.

[0048] Step S1: Multi-source heterogeneous data acquisition based on Figure 1 The perception layer architecture shown deploys integrated air-to-ground perception devices in the target orchard: Category A (Environmental Time Series Data): Five RK520-01 agricultural meteorological stations are deployed in a grid pattern in Area A, with a sampling frequency of f=10min, meaning that air temperature, air humidity, light intensity (Lux), soil moisture, and electrical conductivity are automatically collected every ten minutes. Data; special attention should be paid to monitoring the Frost's Descent data in mid-to-late October (a key meteorological indicator for the formation of the "rock sugar heart").

[0049] Category B (Unstructured Image Data): Using DJI Mavic 3 Multispectral Edition (M3M) drones, four aerial photography sessions were conducted during the fruit tree flowering period in April, the fruit setting period in July, the fruit enlargement period in August, and the ripening period in October to acquire visible light and multispectral images with a resolution better than 5cm; Hikvision AI monitoring PTZ cameras were deployed at key intersections in the orchard to record videos of personnel entering and exiting and harvesting operations.

[0050] Category C (Business Structured Data): Obtain fertilization records (mainly organic fertilizer), pruning records, and expected harvest plans for the plot of land from the ERP system of the resident enterprise.

[0051] Collected 30,810 data points on air temperature and humidity from April 1 to October 31, 2025, 160 GB of drone imagery data, and 8 fertilization records from the ERP system.

[0052] Step S2: According to the appendix Figure 3 The system extracts the defined data structure: Hongqipo Farm Main Entity ID (last 6 digits of Social Credit Code: 35003B); Extract the product batch ID (20251024-A01); Calculate the GeoHash of the plot: Convert the center coordinates (80.2, 41.1) of area A into a 9-bit GeoHash string txw37mv9q; Date and timestamp: Convert the harvest time October 25, 2025, 17:09:06 to a millisecond-level timestamp 1761383346000.

[0053] Perform a hash operation: Based on the formula VTC=SHA256("35003B"⊕"20251024-A01"⊕"txw37mv9q"⊕"1761383346000"), a unique traceability code is generated: DFEc1ycWWcSX3wUDHP.

[0054] Spatiotemporal alignment: The system establishes a data cube indexed by DFEc1ycWWcSX3wUDHP; to address the inconsistency in frequency between UAV imagery (monthly data) and meteorological data (minute-level), a forward hold strategy is adopted to map and associate the NDVI vegetation index (0.78) captured on July 15 with each meteorological data from July 15 to August 15.

[0055] Missing data completion: Data loss (6 points) was detected due to a power outage of the sensor between 14:00 and 15:00 on June 12. The system automatically called the Lagrange interpolation method to fit the average temperature of 28.5℃ for that period using data from the hour before and after the sensor, thus filling in the data gaps.

[0056] Step S3: As shown in Figure 2, the value assessment and pricing process in this embodiment specifically includes three sub-steps: feature engineering extraction, model inference, and dynamic pricing. Feature calculation: Effective accumulated temperature (GDD): Set the biological zero degree for Akzu apples =10℃; The system iterates through data from April to October, using the formula The calculated effective accumulated temperature over the entire cycle is 3450℃ (higher than the standard value of 3200℃).

[0057] Frost observation: Extract meteorological data from October and identify 3 days with temperatures below 0℃ for more than 3 hours, which meet the key meteorological conditions for the formation of "ice sugar heart".

[0058] Quality prediction: Input the above feature vector X=[3450,3,0.78,…] into the pre-trained XGBoost regression model.

[0059] Output: Predict the sugar content of this batch of apples. =18.2 Brix, hardness 7.5 kg / cm2, overall rating is "Extra Grade".

[0060] Dynamic pricing: Base price acquisition: Web crawler retrieves the average price of premium Fuji apples on JD.com and Tmall platforms on the current day. =8.5 yuan / kg.

[0061] Coefficient calculation: setting weights =0.7 (quality weight) =0.3 (supply and demand weight).

[0062] Quality premium: Sigmoid (18.2−14.0)≈0.8 (sweetness level far exceeds the national standard of 14 degrees).

[0063] Supply and demand factors: The current online search volume for "rock sugar heart" has surged, and the inventory is in short supply, contributing 0.2 to the factor.

[0064] The final premium coefficient α≈0.85.

[0065] Final pricing: =8.5×(1+0.85)=15.7 yuan / kg.

[0066] Step S4: On-chain evidence storage and visualization Hybrid storage: Store 160GB of raw image data in IPFS and obtain the content hash. .

[0067] On-chain: Only [VTC, Sugar Content Prediction: 18.2, Suggested Price: 15.7, The Merkle root hash is generated by packaging and written to block height 102488 of the consortium blockchain.

[0068] On the user side: Consumers scan the “Full-chain Value Traceability Code (VTC)” on the packaging box and retrieve on-chain data through the mini-program to display the value radar charts of “sweetness 18.2 (30% higher than the average)” and “having experienced 3 frosts”.

[0069] Experiment Example 2: Digital Origin and Value Anchoring of Nilek Cold-Water Fish (Salmon) (This embodiment aims to verify the universality and technical robustness of the present invention in the aquaculture industry scenario.) This embodiment selects a digital salmon (rainbow trout) aquaculture base (Area B, center coordinates of the net cages in the water area E: 82.5°, N: 43.8°) in Nileke County, Ili Prefecture, Xinjiang as the object.

[0070] Step S1: Construction of Underwater Multidimensional Sensing Network Hardware deployment: Three YSI multi-parameter water quality analyzers are deployed at different depths in the deep-sea cages, with a sampling frequency of f=30min, to monitor water temperature, dissolved oxygen, and pH value; two underwater high-definition cameras are deployed to capture videos of fish swimming at regular intervals every day.

[0071] Data acquisition: 26,000 water quality time-series data points and 500GB of underwater image data were collected during the aquaculture cycle (2023-2025, a total of 26 months).

[0072] Step S2: Fisheries data fusion based on spatiotemporal cube Generate end-to-end value traceability code: Extract the entity ID (last six digits of the aquaculture enterprise's credit code: 48621L); Extract the fish fry batch ID (Fry-202305); Calculate the GeoHash of the net cage: Convert the center coordinates of the net cage into a 9-bit string qznxi48lm; Date and timestamp: Convert the harvest time 2025-05-21 20:18:16 to a millisecond-level timestamp 1747829896000.

[0073] VTC generation: Use the SHA-256 algorithm to generate a unique digital identifier (AczQ0msDx3xoTUf0Hg) for this batch of salmon.

[0074] Spatiotemporal alignment and fusion: Constructing a water body data cube with VTC as the key.

[0075] Dynamic alignment: In response to the uncertainty of the fish's swimming position, the system uses computer vision algorithms from underwater cameras to lock the layer of the fish's main activity depth, and uses the water quality sensor data of that layer (such as dissolved oxygen at a depth of 10 meters) as the effective environmental feature, filtering out invalid data from the surface.

[0076] Step S3: Quality Value Assessment Based on Cold Water Environment Characteristics Fisheries characteristic engineering calculations: Effective Cold-Water Hours (ECWH): The optimal water temperature range for rainbow trout growth is defined [10℃, 14℃]; the cumulative duration of water temperature within this range over the entire cycle is calculated using an integral algorithm. ; Calculation results: This batch of ECWH reached 16,500 hours (accounting for 94%), far exceeding that of ordinary farms.

[0077] Dissolved oxygen stress index: The percentage of time with dissolved oxygen below 6 mg / L was calculated, and the result was 0.1%, indicating that the fish population was rarely under hypoxic stress and there was no risk of meat spoilage.

[0078] Quality prediction model: Construct an XGBoost regression model with input features X=[ECWH, mean dissolved oxygen, feed conversion rate, ...].

[0079] Predicted output: The predicted fat content of the adult fish is 12.5%, and the meat firmness is 8.5N, which meets the international "raw food grade" standard.

[0080] Dynamic pricing: Benchmark Price: Average price of imported salmon captured from fresh food platforms =65 yuan / kg.

[0081] Coefficient calculation: Setting quality weights =0.8. Since the "raw food grade" index is significantly better than the average level of domestic freshwater fish, the quality premium coefficient α is calculated to be 0.35.

[0082] Final pricing: =65×(1+0.35)=87.7 yuan / kg.

[0083] Step S4: Digital Asset Proof Selected clips from 500GB of underwater video are stored in IPFS, and their hashes are obtained. .

[0084] [VTC, Fat percentage: 12.5%, Grade: Raw food grade] Packaging and uploading to the blockchain. Consumers can scan the code to view real-time evidence of "fish swimming in clear, cold water" and verify the "entire cold water accumulation time" data.

[0085] Comparative Example 1: Traditional Manual Input Tracing Method To highlight the technical advantages of the present invention in terms of "data authenticity" and "value quantification capability", the following comparative examples are provided.

[0086] Subjects: Ordinary orchards (traditional growers) adjacent to Experiment 1.

[0087] Technical solution: Use a common QR code traceability system available on the market.

[0088] Data source: After harvesting, farmers manually entered text information such as "Origin: Aksu", "Harvest Date: October 25", and "Fertilizer: Compound Fertilizer" into the background; no IoT sensors or drone images were used.

[0089] Value assessment: No algorithmic model; product pricing depends entirely on the subjective judgment of the buyer or on market conditions, with a uniform price of 8.5 yuan / kg.

[0090] Table 1. Comparison of Results

[0091] In summary, this invention employs a technical architecture combining spatiotemporal cube data fusion, multi-factor dynamic pricing, and hybrid off-chain and on-chain evidence storage, resulting in the following significant beneficial effects: 1. This invention solves the technical challenges of "difficulty in frequency alignment" and "chaotic spatial matching" in multi-source heterogeneous agricultural data. Traditional methods struggle to handle the temporal misalignment between high-frequency IoT data (minute-level) and low-frequency image data (month-level), resulting in rigid data correlations. This invention constructs a spatiotemporal cube indexed by a full-link value traceability code and utilizes sliding window aggregation and Lagrange interpolation algorithms to achieve precise alignment and completion of heterogeneous data in the micro-spatiotemporal dimension. This not only ensures the continuity and integrity of environmental feature vectors but also improves data fusion accuracy by orders of magnitude compared to traditional coarse matching, providing a high-quality data foundation for subsequent AI inference.

[0092] 2. This invention represents a leap from "experience-based valuation" to "mathematical model pricing," granting agricultural products a scientific premium. Existing traceability systems only record information and cannot quantify value; pricing still relies on subjective experience or market fluctuations, leading to the failure of the premium pricing mechanism for high-quality products. This invention innovatively introduces the XGBoost quality prediction model based on environmental feature engineering and maps the prediction results to a dynamic pricing formula through the Sigmoid function. This method transforms the abstract concepts of "good production area and good weather" into calculable and verifiable "mathematical premium coefficients," providing regional public brands with data-driven pricing power and effectively solving the "lemon market" problem (bad money drives out good).

[0093] 3. This invention overcomes the performance bottlenecks of "large data volume" and "high storage cost" in blockchain evidence storage. Agricultural traceability involves massive amounts of high-definition images and time-series logs; directly storing these on the blockchain would lead to a bloated blockchain, high costs, and low query efficiency. This invention adopts a hybrid architecture of "IPFS off-chain storage + Merkle root hash on-chain"; by storing only a lightweight root hash to anchor the entire data, it retains the immutable, judicial-grade evidentiary validity of the blockchain while reducing storage costs by more than 99%, greatly improving the system's retrieval response speed and commercial feasibility.

[0094] 4. Possesses strong robustness and versatility across categories and scenarios; the feature extraction framework proposed in this invention has extremely strong generalization ability; whether it is the "effective accumulated temperature" in planting or the "cold water accumulated time" in fisheries, they can all be uniformly incorporated into the "environmental integral model" for calculation; this means that the system can be quickly replicated from Aksu apples (planting) to Nileke salmon (farming) and Tianlai fragrant cattle (livestock) without reconstructing the underlying architecture, and has broad industrial promotion value.

[0095] The above are merely preferred embodiments of this application and are not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for dynamic pricing and traceability of agricultural products that integrates crop growth mechanisms and multidimensional spatiotemporal data, characterized in that... Includes the following steps: Multi-source heterogeneous data acquisition: At least environmental perception time-series data, unstructured image data and structured business management data of the target agricultural products throughout their entire life cycle from production and processing to distribution; Data fusion and traceability identifier generation: A data organization architecture is constructed based on spatiotemporal index. The collected multi-source heterogeneous data is aligned on the time axis and matched in spatial dimension. Missing value completion processing is performed on the environmental perception time series data. After data fusion is completed, a unique full-link traceability identifier is generated. With the full-link traceability identifier as the index core, a digital archive binding the entire life cycle data of agricultural products is established to realize a one-to-one correspondence between data and agricultural products. Quality assessment and dynamic pricing: Key feature information affecting the quality of agricultural products is screened and extracted from the digital archives, input into the trained algorithm model to predict the quality status of agricultural products, and then market benchmark information, supply and demand dynamics and production entity credit data are integrated to construct a multi-factor pricing model. Based on the quality prediction results and the weights of each factor, the dynamic guidance price of agricultural products is calculated. Data storage and information display: A hybrid architecture of off-chain storage and on-chain anchoring is adopted. The original data and digital archives are stored in a distributed file system. After hash digest processing of the data, a Merkle tree is constructed. Only the root hash is written to the blockchain to complete the storage. At the same time, a mapping relationship between the full-chain traceability identifier and the stored data is established for terminal devices to scan and parse. The traceability trajectory and value assessment information of agricultural products are displayed in a visual way.

2. The method for dynamic pricing and traceability of agricultural products that integrates crop growth mechanisms and multidimensional spatiotemporal data according to claim 1, characterized in that, The environmental sensing time-series data includes air temperature and humidity, soil parameters and light intensity data, which are continuously collected at a preset frequency. The image data includes spectral images, radar data, and high-definition video data, which are collected regularly to record the growth and processing status of agricultural products. The business management data includes agricultural operation records, harvesting batch information, logistics documents, and quality inspection report data, which are extracted synchronously from the corresponding production management system.

3. The method for dynamic pricing and traceability of agricultural products that integrates crop growth mechanisms and multidimensional spatiotemporal data according to claim 1, characterized in that, The spatiotemporal index adopts a data cube architecture. Time axis alignment is achieved through a sliding window aggregation algorithm, and spatial dimension matching is completed based on the coordinates of agricultural production plots. Data completion adopts a differentiated strategy: if it belongs to the preset short-term missing data, it is completed by interpolation; if it belongs to the preset long-term missing data, it is marked with a confidence level and then retained.

4. The method for dynamic pricing and traceability of agricultural products that integrates crop growth mechanisms and multidimensional spatiotemporal data according to claim 1, characterized in that, The full-chain traceability identifier is generated by integrating the production entity identifier, product batch identifier, geospatial code, and timestamp. The generation process involves hash operation and XOR operation in sequence. The geospatial code is converted from the coordinates of the center point of the agricultural product production plot into a fixed-precision spatial hash string to ensure the uniqueness and collision prevention of the identifier.

5. The method for dynamic pricing and traceability of agricultural products that integrates crop growth mechanisms and multidimensional spatiotemporal data according to claim 1, characterized in that, The key characteristics of agricultural product quality include effective accumulated temperature, diurnal temperature range index, light availability characteristics, and soil parameter compliance rate.

6. The method for dynamic pricing and traceability of agricultural products that integrates crop growth mechanisms and multidimensional spatiotemporal data according to claim 1, characterized in that, The key characteristic information of agricultural product quality is extracted according to the pre-defined characteristic extraction rules for different categories of agricultural products. That is, environmental time-series data and image data that are directly related to the intrinsic quality of agricultural products are screened from the full life cycle digital archive, and extracted by pre-defined algorithms in combination with the growth mechanism of the corresponding category of agricultural products.

7. The method for dynamic pricing and traceability of agricultural products that integrates crop growth mechanisms and multidimensional spatiotemporal data according to claim 1, characterized in that, The algorithm model used in the quality assessment and dynamic pricing adopts either the gradient boosting algorithm or the random forest algorithm. It takes the key feature information of the extracted agricultural product quality as input and the physicochemical quality indicators of agricultural products as training targets. After training with sample data, it realizes the quantitative prediction of the quality status of agricultural products.

8. The method for dynamic pricing and traceability of agricultural products that integrates crop growth mechanisms and multidimensional spatiotemporal data according to claim 1, characterized in that, The multi-factor pricing model obtains the quality premium factor through normalized quality prediction results, calculates the supply and demand factor by combining inventory balance and market search popularity, and adds the blockchain credit score of the production entity to assign corresponding weights to each factor before calculating the total premium coefficient. Finally, it superimposes the market benchmark price to obtain the dynamic guidance price.

9. The method for dynamic pricing and traceability of agricultural products that integrates crop growth mechanisms and multidimensional spatiotemporal data according to claim 1, characterized in that, The data hash digest is extracted from the original data and digital archives. After constructing the Merkle tree, only the root hash is uploaded to the chain, which reduces the on-chain storage cost while ensuring that the data is immutable.

10. A dynamic pricing and traceability system for agricultural products that integrates crop growth mechanisms and multidimensional spatiotemporal data, characterized in that: The device includes a processor and a memory, wherein the memory stores a computer program that can run on the processor, and the processor calls the computer program stored in the memory to execute the method for dynamic pricing and traceability of agricultural products that integrates crop growth mechanisms and multidimensional spatiotemporal data as described in any one of claims 1 to 9.

Images