Agricultural industry o2o intelligent supply chain collaborative management platform and construction method thereof

Abstract

The application discloses an agricultural industry O2O intelligent supply chain collaborative management platform and a construction method thereof, and the platform comprises a farmer end data acquisition module, an intelligent prediction and scheduling module, an O2O transaction matching engine, a cold chain logistics dynamic optimization module, a block chain traceability storage module and a collaborative control middle platform. The construction method comprises the following steps: collecting multi-dimensional agricultural data and performing space-time feature coding; adopting a hybrid deep learning and self-adaptive optimization algorithm to generate a production and sales matching strategy; combining a cold chain transportation energy consumption model and a path real-time adjustment mechanism to realize logistics optimization; and utilizing the non-tamperable characteristics of a block chain to record full-link data, so as to realize agricultural product traceability and transaction credit guarantee. The application can effectively improve the response speed of the agricultural supply chain, reduce the loss rate and improve the transaction transparency, and is suitable for various agricultural industry modes.

Description

Technical Field

[0001] This invention belongs to the field of agricultural industry technology, specifically relating to an O2O intelligent supply chain collaborative management platform for the agricultural industry and its construction method. Background Technology

[0002] With the advancement of agricultural modernization and informatization, agricultural production and distribution are shifting from a traditional, decentralized model towards intensive, intelligent, and integrated online and offline operations. In recent years, in particular, the application of the O2O (Online to Offline) model in agricultural product distribution has gradually increased. Consumers can order online and receive fresh agricultural products offline, while producers can connect with a wider market through the platform. However, the current agricultural O2O supply chain system still has many shortcomings, hindering further improvements in efficiency and effectiveness.

[0003] First, in the data collection stage, traditional agricultural supply chains often rely on manual statistics and periodic reporting, making it difficult to grasp real-time changes in the farmland environment and crop growth status. Even in some production areas where IoT sensors have been introduced, there are often problems such as limited data collection parameters and insufficient spatiotemporal resolution, making it impossible to accurately reflect the differences in crops at different growth stages, let alone effectively link with subsequent forecasting and scheduling. In addition, manual data entry is prone to omissions and errors, causing backend systems to make decisions based on distorted data, affecting the overall supply chain's responsiveness and accuracy.

[0004] Secondly, in terms of production and sales forecasting and scheduling, existing systems mostly use single time series models or linear regression methods. These methods have limited forecasting accuracy when faced with complex factors such as the seasonality of agricultural production, sudden climate changes, and market fluctuations. At the same time, scheduling recommendations often lack comprehensive consideration of factors such as real-time market orders and logistics capabilities, resulting in a low degree of matching between production and sales, and the occurrence of local oversupply or shortages.

[0005] Third, in the O2O transaction matching process, most existing platforms adopt simple price-first or time-first rules, ignoring the need for multi-objective optimization, such as improving the matching success rate while reducing delivery delays and default risks. This one-dimensional matching method is prone to causing uneven resource allocation, affecting the balance of interests among all parties in the industry chain, and is also difficult to cope with high-frequency, multi-category, and multi-specification transaction scenarios.

[0006] Fourth, in cold chain logistics, the preservation of agricultural products places stringent demands on transportation routes, temperature control strategies, and energy management. Existing cold chain systems largely rely on static maps and fixed routes for route planning, lacking the ability to dynamically adjust based on real-time traffic, weather, and temperature sensitivity. This can easily lead to excessively long transportation times or temperature control failures, thereby increasing loss rates. Simultaneously, energy management often focuses only on a single cost indicator, failing to comprehensively consider the coupling relationship between route length, temperature difference penalties, and unit energy consumption.

[0007] Fifth, regarding traceability and reliability assurance, although some agricultural product supply chains have introduced QR codes or RFID tags to record production information, these records are easily tampered with or lost, failing to achieve true tamper-proof and end-to-end traceability. Once a problem arises, it is difficult to quickly pinpoint the responsible party, impacting consumer trust and regulatory efficiency.

[0008] Sixth, in terms of overall collaborative control, the existing system's functional modules are mostly loosely connected, lacking a unified central platform to coordinate data and instruction flows. When an anomaly occurs in a certain link, it often fails to quickly trigger global optimization or fault-tolerance mechanisms, leading to a decline in the overall operational efficiency of the supply chain, and even a chain reaction.

[0009] In summary, the existing agricultural O2O supply chain system has significant shortcomings in areas such as data real-time performance and diversity, comprehensive accuracy of forecasting and scheduling, multi-objective optimization of transaction matching, dynamic adaptability of cold chain logistics, reliability of traceability, and global collaborative capabilities. These issues limit the potential for improvement in response speed, loss control, transaction transparency, and overall efficiency of the agricultural supply chain, and also hinder the in-depth development of digitalization and intelligentization in the agricultural industry. Therefore, it is necessary to develop an agricultural O2O intelligent supply chain collaborative management platform and its construction method that integrates multi-dimensional data collection, intelligent forecasting and scheduling, multi-objective transaction matching, dynamic optimization of the cold chain, blockchain traceability, and global collaborative control to address the aforementioned deficiencies and meet the operational needs of modern agricultural supply chains for efficiency, reliability, and transparency. Summary of the Invention

[0010] The purpose of this invention is to provide an O2O intelligent supply chain collaborative management platform for the agricultural industry and its construction method.

[0011] To achieve the above objectives, this invention employs the following technical solution: an agricultural industry O2O intelligent supply chain collaborative management platform and its construction method. The platform's overall architecture comprises six core modules: a farmer-side data collection module, an intelligent forecasting and scheduling module, an O2O transaction matching engine, a cold chain logistics dynamic optimization module, a blockchain traceability and evidence storage module, and a collaborative control platform. The farmer-side data collection module is responsible for acquiring farmland environmental parameters, crop growth status, yield forecast data, and available supply time windows from IoT sensors, drone remote sensing equipment, and manual input interfaces. This raw data forms the basis for all subsequent intelligent analysis and decision-making. The intelligent forecasting and scheduling module encodes the collected multidimensional data using spatiotemporal features and inputs it into a hybrid model integrating convolutional neural networks and bidirectional long short-term memory networks to generate supply and demand forecast curves and scheduling suggestions, thereby optimizing production arrangements in both time and space dimensions. The O2O transaction matching engine, based on the forecast curves and real-time market order information, uses a multi-objective optimization function to calculate the optimal matching scheme and automatically generates electronic contracts, achieving seamless online and offline integration. The cold chain logistics dynamic optimization module combines transportation distance, temperature and time sensitivity index, and vehicle energy consumption model to adjust delivery routes and temperature control strategies in real time, ensuring the quality stability of agricultural products during transportation. The blockchain traceability and evidence storage module hashes data from each stage of production, transaction, and logistics and writes it to the consortium blockchain, achieving tamper-proof and traceability, enhancing the credibility of the entire chain. The collaborative control platform uniformly schedules the data and instruction flows of the above modules and triggers fault tolerance and re-optimization mechanisms in abnormal situations, ensuring the stability and coordination of the overall system operation. The platform's overall operation follows a collaborative control equation, achieving dynamic optimization decisions for the supply chain state by minimizing the weighted average of logistics costs, delay time, and loss costs. This defines the platform's complete modular architecture and collaborative operation mechanism, enabling efficient collaboration among functional units under a unified control strategy. It forms a closed-loop management system from data collection to transaction fulfillment, logistics execution, and traceability and evidence storage, significantly improving the overall response speed and operational reliability of the supply chain. Simultaneously, it achieves global optimization through mathematical models, avoiding resource waste or risk accumulation caused by local optima.

[0012] Furthermore, the farmer-side data collection module collects key indicators such as soil moisture content, air temperature, and light intensity when gathering farmland environmental parameters. These parameters directly affect crop growth and yield potential. Crop growth status is quantified using the NDVI vegetation index calculation formula, which reflects vegetation cover density and health status using the ratio of reflectance in the near-infrared band to the red band. The collection frequency is not fixed but dynamically adjusted according to the crop growth stage. The adjustment formula takes into account the crop growth stage index, resulting in more intensive collection during key growth stages and a reduced frequency during maturity, ensuring data representativeness while conserving collection resources. This strategy of multi-parameter collection and dynamic frequency adjustment throughout the growth stage improves data timeliness and accuracy, enabling subsequent prediction models to perform calculations based on more refined growth status information. This enhances prediction accuracy and scheduling rationality while reducing unnecessary energy consumption and storage pressure.

[0013] Furthermore, the core algorithm of the intelligent forecasting and scheduling module is a hybrid model integrating convolutional neural networks (CNNs) and bidirectional long short-term memory (LSTM) networks. CNNs excel at extracting spatial features, while LSTM networks are adept at capturing the dependencies between time series data. The combination of these two technologies can simultaneously handle the spatial distribution and temporal evolution characteristics of agricultural data. The model input is a multidimensional data matrix encoded with spatiotemporal features, and the output is the supply and demand forecast. An error correction term calculated using adaptive Kalman filtering is introduced to correct the forecast results in real time, addressing sudden environmental changes or data noise. Through this hybrid model structure and error correction mechanism, the robustness and accuracy of the forecasts are significantly improved. It maintains high forecast quality in complex and ever-changing agricultural environments, providing a reliable basis for scheduling decisions and reducing supply and demand imbalances caused by forecast bias.

[0014] Furthermore, the O2O transaction matching engine employs a multi-objective optimization function, incorporating matching success rate, delivery time deviation, and default penalty value into a unified optimization framework, and introducing dynamic weights to adapt to different market scenarios. Matching success rate reflects the efficiency of transaction completion, delivery time deviation reflects the timeliness of fulfillment, and default penalty value measures transaction risk; all three jointly determine the final matching scheme. The goal of the optimization function is to maximize overall benefits, ensuring transaction volume while controlling latency and risk. This multi-objective optimization mechanism achieves a comprehensive balance in transaction matching, avoiding the biases caused by optimizing a single indicator. This allows the platform to flexibly adjust its matching strategy under changing market demands, improving transaction success rate while reducing default and delay risks, and enhancing the willingness of all parties in the industry chain to cooperate.

[0015] Furthermore, the path planning of the cold chain logistics dynamic optimization module adopts an improved A / B algorithm. The algorithm, whose cost function adds a temperature difference penalty factor and a unit energy consumption index to the traditional distance calculation, ensures that route selection considers not only the shortest distance but also temperature control requirements and energy efficiency. The temperature difference penalty factor reflects the impact of potential temperature fluctuations on agricultural products along the route, while the unit energy consumption index reflects the economic efficiency of transportation. This module updates the route every 15 minutes based on real-time traffic, weather, and vehicle status. The update condition is based on a threshold ratio of the actual route deviation to the planned route length, ensuring the route is always in an optimal or near-optimal state. Through the improved route planning algorithm and dynamic update mechanism, dual optimization of temperature control and energy consumption in cold chain transportation is achieved, effectively reducing transportation losses and costs while improving delivery timeliness and flexibility, adapting to complex transportation environments.

[0016] Furthermore, the blockchain traceability and evidence storage module adopts a consortium blockchain structure, with nodes including key participants in the supply chain such as farmers, processing enterprises, logistics companies, and sales platforms. The data upload process involves packaging local data and calculating the Merkle root, sending the Merkle root along with a timestamp and digital signature to the consensus node, and writing it into a block after consensus is reached. The block header includes the hash of the previous block and the current state root, ensuring the integrity and immutability of the chain structure. Through the consortium blockchain structure and rigorous upload process, tamper-proof and traceable data across the entire supply chain is achieved, enhancing consumer trust in product quality and providing regulatory authorities with a reliable data source, thus contributing to improved transparency and credibility throughout the supply chain.

[0017] Furthermore, the collaborative control platform incorporates an anomaly detection unit, employing a sliding window-based statistical process control method to monitor key operational indicators in real time. When a monitored indicator exceeds control limits, the system automatically triggers a re-optimization process. The control limits are calculated based on the mean and standard deviation within the window, enabling timely detection of potential anomalies and the implementation of measures to prevent their spread and impact on the overall system. Through real-time anomaly detection and rapid re-optimization mechanisms, the system's fault tolerance and stability are significantly improved, allowing for intervention at the early stages of problems and preventing supply chain disruptions or severe losses.

[0018] Furthermore, the platform supports multilingual interfaces and multi-currency settlement. The settlement exchange rate is calculated using a real-time weighted average method, with the weight determined by the trading volume of each currency. This ensures that the exchange rate reflects the actual trading situation and avoids discrepancies caused by a single exchange rate source. Through multilingual and multi-currency support, the platform possesses international application capabilities, enabling it to serve cross-border trading scenarios. Simultaneously, the real-time weighted average exchange rate guarantees the fairness and accuracy of settlement, enhancing the trading experience for international users.

[0019] Furthermore, the platform construction methodology comprises eight steps, starting with equipment deployment and sensor calibration, followed by the establishment of a multi-dimensional database and spatiotemporal feature encoding, then model training and validation, matching rule configuration, path model establishment, blockchain node deployment, middleware integration and testing, and finally, trial operation and iterative optimization, forming a complete implementation process. This provides a systematic and replicable construction method, ensuring the platform's complete functionality and smooth integration of each stage, reducing implementation difficulty and risk, and facilitating its promotion and application in different regions and industries.

[0020] Furthermore, the model accuracy verification in the construction method adopts a weighted approach. The scores are biased towards increasing recall rates to meet the needs of supply chain risk early warning. The value is set to 1.5. The path model update condition is based on a threshold ratio of actual path deviation to planned path length, ensuring the timeliness and effectiveness of path adjustments. Through targeted accuracy verification indicators and path update conditions, the platform performs better in risk warning and logistics optimization, enabling it to quickly identify potential risks and adjust strategies, thereby improving the robustness and adaptability of the supply chain.

[0021] This invention provides an O2O intelligent supply chain collaborative management platform for the agricultural industry and its construction method, which has the following beneficial effects: The agricultural industry O2O intelligent supply chain collaborative management platform and its construction method proposed in this invention have made systematic innovations to address the shortcomings of existing technologies and have many significant advantages, which are specifically reflected in the following aspects.

[0022] Firstly, regarding data acquisition, this invention integrates IoT sensors, UAV remote sensing equipment, and manual input interfaces through a farmer-side data acquisition module, enabling multi-dimensional real-time acquisition of farmland environmental parameters, crop growth status, yield forecasts, and supply time windows. In particular, the use of the NDVI vegetation index calculation formula to quantitatively assess crop growth status, combined with a formula for dynamically adjusting the acquisition frequency during the growing season, results in more refined data in terms of temporal resolution and spatial coverage. This provides high-quality input for subsequent forecasting and scheduling, significantly improving the accuracy and timeliness of decision-making.

[0023] Secondly, in terms of intelligent forecasting and scheduling, this invention constructs a hybrid model integrating convolutional neural networks and bidirectional long short-term memory networks, and performs spatiotemporal encoding on the input features, thereby simultaneously capturing the spatial distribution characteristics and temporal evolution patterns of agricultural data. The prediction results are further improved in accuracy through adaptive Kalman filtering error correction. This model can not only output supply and demand forecast curves, but also generate scheduling suggestions by combining real-time market orders, ensuring a high degree of alignment between production arrangements and market demand, and reducing the risks of inventory backlog and stockouts.

[0024] Third, regarding O2O transaction matching, this invention designs a multi-objective optimization function that incorporates matching success rate, delivery time deviation, and default penalty value into a unified optimization framework, and adapts to different market scenarios through dynamic weight adjustment. This matching mechanism can ensure transaction success rate while taking into account timeliness and risk control, improving the satisfaction and stability of all parties in the industry chain, and is especially suitable for agricultural product trading scenarios with multiple categories, specifications, and high frequency.

[0025] Fourth, in terms of cold chain logistics, this invention adopts an improved A... The algorithm performs route planning and incorporates a temperature difference penalty factor and a unit energy consumption index into the cost function. This ensures that route selection not only considers the shortest distance but also takes into account temperature control requirements and energy efficiency. Coupled with a real-time update mechanism every 15 minutes, it can dynamically adjust delivery plans based on changes in traffic, weather, and vehicle status, significantly reducing transportation losses and energy consumption, and extending the shelf life of agricultural products.

[0026] Fifth, regarding traceability and trust assurance, this invention is based on a consortium blockchain structure. Data from each stage of production, transaction, and logistics is hashed and then written into the blockchain, achieving tamper-proof and end-to-end traceability. Through the organic combination of Merkle roots, timestamps, digital signatures, and consensus mechanisms, it ensures that data cannot be altered once it is on the blockchain. Consumers and management departments can query authentic and reliable traceability information at any time, significantly enhancing product trust and brand value.

[0027] Sixth, regarding global collaborative control, this invention establishes a collaborative control platform to uniformly manage the data and instruction flows of each module, and incorporates a sliding window-based statistical process control anomaly detection unit. When monitored indicators exceed control limits, the system can immediately trigger a re-optimization process, achieving rapid fault tolerance and recovery, and ensuring the stable operation of the entire supply chain. This centralized collaborative mechanism effectively avoids information silos and response delays between modules, improving the overall resilience and adaptability of the system.

[0028] Seventh, the present invention also considers the needs of international applications in the design of platform functions, supports multi-language interfaces and multi-currency settlement, and uses the real-time weighted average method to calculate the settlement exchange rate, ensuring the fairness and accuracy of exchange rate conversion in cross-border transactions, and expanding the application scope and market space of the platform.

[0029] Eighth, in terms of construction methodology, this invention provides a complete process from equipment deployment, data collection, model training, rule configuration, path model establishment, blockchain deployment to platform integration. Furthermore, it employs a weighted Fβ score in model accuracy verification, biased towards improving recall to meet the needs of supply chain risk early warning. This systematic construction method not only ensures the integrity of the platform's functions but also provides a replicable implementation path for different regions and types of agricultural industries.

[0030] Ninth, all key algorithms and model parameters of this invention have been originally designed and mathematically derived, avoiding duplication with existing technologies and ensuring the novelty and inventiveness of the patent. For example, the cold chain path cost function, the transaction matching multi-objective optimization function, and the anomaly detection control limit formula are all specially designed for the specific needs of the agricultural supply chain and have significant industry adaptability.

[0031] Tenth, in summary, this invention, through the organic integration of multi-dimensional data collection, intelligent prediction and scheduling, multi-objective transaction matching, cold chain dynamic optimization, blockchain traceability, and global collaborative control, has achieved a comprehensive improvement in the agricultural industry's O2O supply chain in terms of response speed, loss control, transaction transparency, operational stability, and internationalization capabilities. It can effectively promote the digital transformation and high-quality development of the agricultural industry and has broad application prospects and significant economic and social benefits. Attached Figure Description

[0032] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0033] Figure 1 This is a diagram illustrating the overall operational logic of the platform of this invention. Figure 2 This is a logic diagram of the farmer-side data collection module of the present invention; Figure 3 This is a logic diagram of the intelligent prediction and scheduling module of the present invention; Figure 4 This is the logic of the O2O transaction matching engine in this invention. Detailed Implementation

[0034] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses consistent with some aspects of this disclosure as detailed in the appended claims.

[0035] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0036] How to use: The use of the farmer-side data collection module: During the platform initialization phase, IoT sensors, drone remote sensing equipment, and manual data entry interfaces are first deployed in the farmland. The sensors continuously collect data on soil moisture content (θ) and air temperature. Light intensity Environmental parameters, such as near-infrared reflectance, are obtained by UAV remote sensing equipment through spectral measurements. Reflectivity in the red light band and calculate Vegetation Index:

[0037] The sampling frequency is based on the crop growth stage index. Dynamic adjustment, the formula is:

[0038] Farmers supplement yield forecasts and available supply time windows through a manual input interface, and all data is uploaded to the collaborative control platform in real time.

[0039] The intelligent prediction and scheduling module is used as follows: After data collection is completed, the platform performs spatiotemporal feature encoding on the multidimensional data to form an input feature matrix. The matrix is ​​then input into a hybrid model that integrates a convolutional neural network and a bidirectional long short-term memory network to obtain the predicted value:

[0040] in Error correction terms are calculated using adaptive Kalman filtering. The model outputs supply and demand forecast curves and scheduling suggestions to guide production planning and resource allocation.

[0041] The O2O transaction matching engine works by: acquiring market order information in real time, inputting the prediction curve and order data into a multi-objective optimization function to calculate the optimal matching solution.

[0042] in To improve the matching success rate, Due to delivery time deviation, This is the penalty value for breach of contract. , , The weighting is dynamic. Upon successful matching, an electronic contract is automatically generated and sent to both parties.

[0043] The use of the cold chain logistics dynamic optimization module: Based on the matching results, the platform calls the cold chain logistics dynamic optimization module and adopts improved A... The algorithm performs path planning, and the cost function is:

[0044] in The distance between nodes. As the temperature difference penalty factor, Energy consumption per unit , To adjust parameters, the system checks the actual path deviation every 15 minutes. With planned path length If the proportion exceeds the threshold Then the path will be recalculated.

[0045] The blockchain traceability and evidence storage module works as follows: During production and transactions, data from each stage is hashed to generate a local Merkle root, which is then sent to the consortium blockchain consensus node along with a timestamp and digital signature. Once consensus is reached, the data is written into a block. The block header contains the hash of the previous block and the current state root, ensuring that the data is immutable and fully traceable.

[0046] The collaborative control platform is used to uniformly manage the data and instruction flows of each module and monitor key indicators in real time. It employs a sliding window statistical process control method, triggering a re-optimization process when an indicator exceeds control limits. The control limit calculation formula is as follows:

[0047] in The mean within the window. This represents the standard deviation. In case of anomalies, the platform can reschedule module operating parameters or activate fault tolerance mechanisms.

[0048] Multilingual and Multicurrency Settlement: The platform supports multilingual interfaces and multicurrency settlement. The settlement exchange rate is calculated using a real-time weighted average method.

[0049] in For the first Trading volume of various currencies The corresponding exchange rate will be displayed. Users can select their language and settlement currency in the platform settings interface.

[0050] Platform build process: During the initial deployment, follow these steps: (1) Deploy data collection equipment at the farmer's end and calibrate sensor parameters; (2) Establish a multidimensional agricultural database and perform spatiotemporal feature encoding; (3) Train the hybrid prediction model and verify its accuracy; (4) Configure O2O matching rules and multi-objective optimization parameters; (5) Establish a cold chain energy consumption and path model and embed a real-time adjustment mechanism; (6) Deploy consortium blockchain nodes and define data on-chain standards; (7) Integrate the collaborative control platform and test its fault tolerance performance; (8) Go online for trial operation and conduct iterative optimization.

[0051] Model accuracy validation and path update conditions: After model training is complete, a weighted average is used. Precision verification of the fractions:

[0052] in The approach prioritizes increasing recall rates to meet the demands of supply chain risk early warning. The path model update condition is:

[0053] When this condition is met, the system immediately replans the cold chain transportation route.

[0054] Global collaborative operation: During the platform's operation, each module works collaboratively according to the above steps and formulas to form a complete closed loop from data collection, prediction and scheduling, transaction matching, cold chain optimization, traceability and evidence storage to global control, ensuring the efficient, stable and transparent operation of the agricultural industry O2O intelligent supply chain.

[0055] Example: 1. Example of Farmer-Side Data Collection Module – Multi-Variety Planting Environment in Hilly Areas In hilly areas, the terrain is highly undulating, and crops are diverse and sparsely distributed. When deploying IoT sensors in this environment, the platform uses sensor nodes of varying densities based on different slope aspects and altitudes to capture spatial differences in soil moisture content, air temperature, and light intensity. UAV remote sensing equipment covers the entire planting area along its flight path, acquiring near-infrared and red light reflectance data and calculating... Vegetation Index:

[0056] Because different crops have different growth cycles, the sampling frequency is dynamically adjusted according to the crop growth stage index:

[0057] Farmers supplement the yield forecast and available supply time window for each variety through a manual input interface. The data is transmitted back to the collaborative control platform in real time to ensure that comprehensive and accurate raw data can still be obtained in environments with complex terrain and diverse varieties.

[0058] 2. Intelligent Prediction and Scheduling Module Implementation Example – Coastal Humid Climate Environment

[0059] In humid coastal climate zones, temperature and humidity fluctuate frequently, significantly impacting crop growth due to the maritime climate. Under these conditions, the platform performs spatiotemporal feature encoding on the collected multidimensional data, forming an input feature matrix. This matrix is ​​then input into a hybrid model fusing convolutional neural networks and bidirectional long short-term memory networks to obtain predicted values.

[0060] The error correction term is calculated using an adaptive Kalman filter to offset data noise caused by climate fluctuations. The model outputs supply and demand forecast curves and scheduling suggestions to help farmers rationally arrange harvesting and marketing times during the wet and rainy season, reducing unsold inventory or quality decline caused by sudden weather changes.

[0061] 3. O2O Transaction Matching Engine Implementation Example – High-Altitude Seasonal Market Activity Environment

[0062] In high-altitude regions, market activity varies significantly with the seasons, with the peak tourist season overlapping with the agricultural harvest season. Under these conditions, the platform acquires real-time online order and offline market information, inputting the forecast curve and order data into a multi-objective optimization function.

[0063] During peak tourist seasons, the system dynamically increases the weight of matching success rates to ensure more orders are fulfilled; during off-seasons, it focuses on reducing penalties for breach of contract and delivery deviations to maintain supply chain stability. Upon successful matching, an electronic contract is automatically generated and pushed to both parties, achieving efficient integration between online orders and offline delivery in high-altitude areas.

[0064] 4. Example of dynamic optimization module for cold chain logistics – high-temperature and dry desert edge environment

[0065] In the hot and arid regions on the edge of deserts, agricultural products are prone to moisture loss and spoilage during transportation. Under these conditions, the platform invokes the cold chain logistics dynamic optimization module, employing improved A... The algorithm performs path planning, and the cost function is:

[0066] The system adjusts the temperature difference penalty factor based on real-time meteorological data, prioritizing routes with small temperature variations and reducing transportation links that are exposed to high temperatures for extended periods. At regular intervals, the system checks the ratio of actual route deviation to planned route length; if it exceeds a threshold, the route is replanned to ensure the freshness and quality of agricultural products in a high-temperature, dry environment.

[0067] 5. Implementation Example of Blockchain Traceability and Evidence Preservation Module – Multi-Party Cross-Domain Collaboration Environment

[0068] In multi-party, cross-regional collaborative environments, such as cross-provincial agricultural product supply chains, multiple entities are involved, including farmers, processing enterprises, logistics companies, and sales platforms. In this environment, the platform hashes data from each stage of production, transaction, and logistics to generate a local Merkle root, which is then sent to the consortium blockchain consensus node along with a timestamp and digital signature. Once consensus is reached, the data is written into a block, with the block header containing the hash of the previous block and the current state root, achieving end-to-end tamper-proofing and traceability. Regardless of geographical or administrative boundaries, consumers and regulatory agencies can access complete and authentic traceability information, enhancing trust and transparency in cross-regional supply chains.

[0069] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An agricultural industry O2O intelligent supply chain collaborative management platform and its construction method, characterized in that, include: Farmer-side data acquisition module: used to collect farmland environmental parameters, crop growth status, yield forecast data, and available supply time windows for farmers through IoT sensors, drone remote sensing equipment, and manual input interfaces; Intelligent prediction and scheduling module: used to encode the spatiotemporal features of the collected multidimensional data and input it into a hybrid model that integrates convolutional neural networks and bidirectional long short-term memory networks to generate supply and demand prediction curves and scheduling suggestions; O2O transaction matching engine: It is used to calculate the optimal matching scheme based on the prediction curve and real-time market order information, and automatically generate electronic contracts. Cold chain logistics dynamic optimization module: used to adjust delivery routes and temperature control strategies in real time based on transportation distance, temperature and time sensitivity index and vehicle energy consumption model; Blockchain traceability and evidence storage module: used to write data from each stage of production, transaction, and logistics into the consortium blockchain after hash processing, so as to achieve tamper-proof and traceability; Collaborative Control Platform: Used to uniformly schedule the data and instruction flows of the above modules, and trigger fault tolerance and re-optimization mechanisms in abnormal situations; The overall operation of the platform follows the following collaborative control equation: In the formula, For a moment The supply chain collaboration state decision vector, For logistics cost function, It is a time delay function. Let the loss cost function be... , , The dynamic weighting coefficient is determined by the real-time supply-demand ratio and the market volatility index.

2. The agricultural industry O2O intelligent supply chain collaborative management platform and its construction method according to claim 1, characterized in that, The farmer-side data collection module includes farmland environmental parameters such as soil moisture content. air temperature Light intensity Crop growth status through Vegetation index calculation: in For near-infrared reflectivity, The red light band reflectance is used; the sampling frequency is dynamically adjusted according to the crop growth period, and the formula is: This is an index representing the crop growth stage. , This is an empirical coefficient.

3. The platform according to claim 1, characterized in that, The hybrid model structure of the intelligent prediction and scheduling module is as follows: in for The multidimensional input feature matrix at time step 1. For predicted values, This is the error correction term, calculated by adaptive Kalman filtering.

4. The agricultural industry O2O intelligent supply chain collaborative management platform and its construction method according to claim 1, characterized in that, The multi-objective optimization function of the O2O transaction matching engine is: in To improve the matching success rate, The deviation from the ideal delivery time. This is the penalty value for breach of contract. , , For dynamic weights.

5. The agricultural industry O2O intelligent supply chain collaborative management platform and its construction method according to claim 1, characterized in that, The path adjustment algorithm used in the cold chain logistics dynamic optimization module is based on an improved A... The algorithm's cold chain constrained path planning has the following cost function: in The distance between nodes. As the temperature difference penalty factor, Energy consumption per unit , To adjust the parameters.

6. The agricultural industry O2O intelligent supply chain collaborative management platform and its construction method according to claim 1, characterized in that, The blockchain traceability and evidence storage module adopts a consortium blockchain structure, with nodes including farmers, processing enterprises, logistics companies, and sales platforms; the data upload process includes: Local data is packaged and Merkle roots are calculated; Send the Merkle root along with the timestamp and digital signature to the consensus node; Once consensus is reached, the data is written into a block. The block header contains the hash of the previous block and the current state root, ensuring chain integrity.

7. The agricultural industry O2O intelligent supply chain collaborative management platform and its construction method according to claim 1, characterized in that, The collaborative control platform has a built-in anomaly detection unit and adopts a statistical process control method based on a sliding window. When the monitored index exceeds the control limit, a re-optimization process is triggered. The control limit calculation formula is as follows: in The mean within the window. The standard deviation is denoted as .

8. The agricultural industry O2O intelligent supply chain collaborative management platform and its construction method according to claim 1, characterized in that, The platform supports multilingual interfaces and multi-currency settlements, with exchange rates calculated using a real-time weighted average method. in For the first Trading volume of various currencies This corresponds to the exchange rate.

9. The agricultural industry O2O intelligent supply chain collaborative management platform and its construction method according to claim 1, characterized in that, The platform construction method includes the following steps: (1) Deploy data collection equipment at the farmer's end and calibrate sensor parameters; (2) Establish a multidimensional agricultural database and perform spatiotemporal feature encoding; (3) Train the hybrid prediction model and verify its accuracy; (4) Configure O2O matching rules and multi-objective optimization parameters; (5) Establish a cold chain energy consumption and path model and embed a real-time adjustment mechanism; (6) Deploy consortium blockchain nodes and define data on-chain standards; (7) Integrate the collaborative control platform and test its fault tolerance performance; (8) Go online for trial operation and conduct iterative optimization.

10. The agricultural industry O2O intelligent supply chain collaborative management platform and its construction method according to claim 9, characterized in that, In step (3), the model accuracy verification uses a weighted average. Fraction: in The model is biased towards improving recall rate to meet the needs of supply chain risk early warning; in step (5), the path model is updated every 15 minutes, and the update conditions are: This represents the actual path deviation. For the planned path length, This represents the threshold ratio.

Images