A big data-based recycling system and method for scheduling and sharing of recycled packages

Abstract

The application discloses a kind of based on big data's circulation package sharing scheduling recovery system and method, it is related to data analysis technical field, the method includes following steps: obtaining the flow state data and inventory change data of circulation package, the data is verified by blockchain;Key flow event is hashed, and hierarchical storage is executed according to data access frequency and storage demand, based on flow event hash association constructs whole life cycle traceability chain, extracts each node historical flow data from blockchain, constructs feature dataset, identifies demand fluctuation mode to construct feature early warning information set;Real-time monitoring node flow data, match abnormal mode and calculate the comprehensive scheduling urgency coefficient of corresponding node, filter high priority scheduling node, according to high priority scheduling node query blockchain idle empty box, generate scheduling single chain;Recovery node comparison information, standard then generate push recovery single, the application realizes circulation package sharing scheduling recovery.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, specifically a circular packaging shared scheduling and recycling system and method based on big data. Background Technology

[0002] Traditional recycling packaging management methods often face numerous problems when adapting to the full-process management of new material recycling boxes and achieving shared scheduling: First, shared scheduling lacks big data analysis support, resulting in low efficiency in supply and demand matching. Traditional methods rely on manual statistics of the inventory, idle, and in-transit status of new material recycling boxes at each node, failing to leverage big data to uncover fluctuations in box demand at each node. This leads to supply and demand imbalances among multiple nodes, with both idle and shortage issues coexisting, making true shared scheduling difficult to achieve. Second, box inventory relies on manual operation, lacking a reliable data traceability system. The flow data of new material recycling boxes, including inbound, outbound, and transfer data, is scattered and lacks unified storage and verification using blockchain technology. Data is easily tampered with or lost, and manual box-by-box verification is not only time-consuming and labor-intensive, but also prone to large inventory errors and cannot achieve automated inventory. Furthermore, traditional recycling processes lack quantifiable thresholds, easily leading to untimely recycling or empty transport. Summary of the Invention

[0003] The purpose of this invention is to provide a big data-based cyclic packaging shared scheduling and recycling system and method to solve the problems raised in the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a cyclic packaging shared scheduling and recycling method based on big data, the method comprising the following steps:

[0005] Obtain circulation status data and inventory change data of reusable packaging boxes, and verify the data through blockchain;

[0006] The collected circulation status data and inventory change data are processed by hashing key circulation events using blockchain, and hierarchical storage is performed according to data access frequency and evidence storage requirements. A full life cycle traceability chain is built based on the circulation event hash association.

[0007] Extract historical transaction data from each node on the blockchain to construct a feature dataset;

[0008] Based on the aforementioned feature dataset, demand fluctuation patterns are identified, and a feature-based early warning information set is constructed.

[0009] Real-time monitoring of node flow data, identification of abnormal patterns matching the feature warning information set, calculation of the comprehensive scheduling urgency coefficient of nodes that trigger the warning mode, and selection of high-priority scheduling nodes based on this.

[0010] Based on the query of idle empty bins on the blockchain by high-priority scheduling nodes, a scheduling order is generated and uploaded to the chain; the bin information is read by the recycling node and compared with the records on the chain. When the cumulative recycling bin quantity reaches the threshold, a recycling task order is generated and pushed.

[0011] The process involves acquiring data on the circulation status and inventory changes of reusable packaging boxes, and verifying this data using blockchain technology. Specific steps include:

[0012] The system acquires circulation status data and inventory change data for reusable packaging boxes at each node. The circulation status data includes: box identification, entry / exit time, and operation node location collected by RFID reading and writing devices; and handover type and handler information recorded by handover terminals. The inventory change data includes: real-time empty box inventory and in-transit box quantity at each node; and the number of boxes leaving the warehouse, the number of boxes entering the warehouse, and the box dwell time for each node during historical periods extracted from the historical database.

[0013] The collected data is verified for consistency and integrity using blockchain. Specifically, this includes: verifying the completeness of key fields for each data entry based on preset data format rules; verifying whether the data has been tampered with during transmission based on hash algorithms; and removing duplicate uploads or data with abnormal timestamps. For data that passes verification, the data is categorized based on the current circulation status of the container and node attributes, and each data entry is labeled with a collection timestamp and geographical location tag. The categorized data is then written into the blockchain distributed ledger according to a preset storage strategy, with data related to the real-time location of the container written into a high-speed access channel and data related to historical circulation records written into a permanent evidence block.

[0014] The collected circulation status data and inventory change data are processed using blockchain to perform key circulation event hashing. Layered storage is implemented based on data access frequency and evidence storage requirements. A full lifecycle traceability chain is constructed based on the circulation event hash associations. Specific steps include:

[0015] Hash the key circulation events of each cycle of packaging boxes to generate a unique evidence identifier for that circulation event. The key circulation events include the box entry and exit events, cross-node transfer events, and recycling handover events.

[0016] Based on the access frequency and evidence storage requirements of the container data, a hierarchical storage strategy is implemented using blockchain: real-time location updates and current status change data of the container are written to the main chain; historical transfer records and past handover voucher data of the container are transferred to the side chain or distributed storage network according to preset storage rules; at the same time, a unique timestamp generated by a blockchain timestamp server is assigned to each piece of on-chain data to record the generation time or on-chain processing time of the data; based on the hash association between each transfer event, a full lifecycle traceability chain is constructed by including the hash value of the previous record in each new block or new record.

[0017] Extracting historical transaction data from each node on the blockchain to construct a feature dataset involves the following steps:

[0018] In the constructed full lifecycle traceability chain, the historical circulation data of the reusable packaging boxes of each node in the past preset time period is extracted. The historical circulation data includes: daily outbound box quantity, daily inbound box quantity, average box dwell time, and node business type identifier.

[0019] The extracted historical data is cleaned and normalized to remove outliers caused by abnormal data collection, and missing data is filled in by interpolation to generate a standardized time series dataset.

[0020] Feature extraction is performed on the standardized time-series dataset to obtain a set of feature information for each node. The feature information includes: periodic features, trend features, and volatility features.

[0021] The periodicity characteristic refers to the fluctuation pattern of box usage statistics on a weekly, monthly, or quarterly basis; the trend characteristic refers to the long-term growth slope or decline slope fitted by linear regression; and the volatility characteristic refers to statistical indicators such as variance, standard deviation, and kurtosis that reflect the degree of data dispersion.

[0022] The feature information of all nodes is aggregated to construct a global feature dataset.

[0023] Identifying demand fluctuation patterns and constructing a feature-based early warning information set involves the following steps:

[0024] Cluster analysis is performed on the global feature dataset to divide nodes with similar binning demand patterns into several node category clusters; where binning demand pattern represents the node binning feature vector composed of extracted periodic features, trend features, and volatility features.

[0025] Nodes with similar bin usage patterns are divided into several node clusters, as follows:

[0026] The similarity between the feature vectors of each node is calculated using a preset clustering algorithm (e.g., K-means, DBSCAN, or hierarchical clustering). Nodes with similarity higher than a preset threshold are grouped into the same cluster. The preset threshold can be set according to the type of clustering algorithm and the characteristics of the data distribution. For example, for the DBSCAN algorithm, the neighborhood radius ε can be determined based on the k-distance graph. For distance-based thresholds, a similarity threshold can be set based on business experience.

[0027] For each node cluster, feature information of nodes that have triggered emergency allocation or caused container shortage events in the past is extracted to form a feature warning information set for that type of node; the feature warning information set contains several feature items, each feature item corresponds to a feature extracted from the historical flow data, and a corresponding warning threshold range is set;

[0028] The aforementioned warning threshold range refers to a preset critical value or critical interval for each feature item to determine whether the real-time feature value is abnormal. It can be a one-way threshold (e.g., triggering a warning if the value is greater than or less than a certain value) or a two-way interval threshold (e.g., the feature value is normal if it is within the preset interval, and triggers a warning if it exceeds the interval). The threshold range can be determined based on the statistical distribution of historical data, for example, by taking the mean of historical data plus or minus a certain number of standard deviations, or by taking a certain percentile (e.g., the 90th percentile) of historical normal data as the threshold. It can also be set directly based on business experience. For example, for the feature item of dwell time, it can be set to not exceed 48 hours according to the node turnover requirements. Those skilled in the art can flexibly set appropriate threshold ranges for different feature items according to actual data characteristics and business needs. This invention does not impose specific numerical limitations on this.

[0029] Write the feature warning information set of each node category cluster into the blockchain.

[0030] Real-time monitoring of node flow data and identification of abnormal patterns matching the feature-based early warning information set. Specific steps include:

[0031] By using the real-time on-chain node flow data, and based on the feature extraction method of the historical flow data, the real-time feature vector of each node within the current time window is calculated.

[0032] Based on the category cluster to which each node belongs, the corresponding feature warning information set is retrieved, and each feature item in the real-time feature vector is matched with the corresponding feature item in the feature warning information set;

[0033] If the feature values ​​of a node in its real-time feature vector exceed the warning threshold range of the corresponding feature item in the feature warning information set for two consecutive time windows, then the node is determined to have triggered the warning mode, and the two real-time records corresponding to the node in these two time windows are extracted.

[0034] The comprehensive scheduling urgency coefficient of the node that triggers the early warning mode is calculated, and the specific steps include:

[0035] Obtain the two real-time records corresponding to the node that triggered the early warning mode within two consecutive time windows, denoted as R. a and R b , where R a The time was earlier than R b ; respectively from R a and R b Extract the set of features that trigger the early warning mode, denoted as F. a and F b ;

[0036] Calculate F a Relative to F b The set of distinguishing features F(ab) = F a -F a ∩F b , and F b Relative to F a The set of distinguishing features F'(ab) = F b -F a ∩F b ;

[0037] For each feature term in F(ab), calculate its deviation from the corresponding feature term in the feature warning information set of the category cluster to which the node belongs, and take the product of all deviations as the first comprehensive scheduling urgency coefficient β. a β is defined as follows: a =∏ f∈Fab θ f ; where θ f The deviation of feature term f is represented by the ratio of the real-time feature value to the warning threshold.

[0038] Similarly, calculate the second integrated scheduling urgency coefficient β corresponding to F'(ab). b ;

[0039] If β b >β a And β b -β a If the value is greater than η, then the node is determined to be a high-priority scheduling node, and its identifier is added to the list of nodes to be scheduled. Here, η represents the preset comprehensive urgency coefficient threshold.

[0040] The specific value of the preset comprehensive urgency coefficient threshold can be set according to actual business needs and historical data analysis results. As an optional implementation method, it is possible to analyze cases in historical scheduling data that successfully avoid box shortage events, statistically analyze the corresponding comprehensive urgency coefficient difference distribution, and select the percentile that enables the recall rate in historical samples to reach a preset target (such as 90%) as the preset comprehensive urgency coefficient threshold η. The recall rate is defined as the proportion of the number of emergency events that are correctly identified as needing priority scheduling to the total number of actual emergency events.

[0041] Based on the query of the source nodes of idle empty boxes on the blockchain by high-priority scheduling nodes, a scheduling task order is generated and stored on the blockchain; at the recycling node, the box information is read in batches by RFID devices and compared with the on-chain records. When the recycling threshold is reached, a recycling task order is generated and pushed. The specific steps include:

[0042] Based on the high-priority scheduling node as the target node, query the blockchain for available source nodes with idle empty boxes, determine the amount of boxes to be allocated based on the amount of boxes missing from the target node, generate a scheduling task order containing the amount of boxes to be allocated, source node identifier, and target node identifier, and store the scheduling task order on the blockchain for evidence.

[0043] The quantity of boxes to be allocated can be determined based on the shortage of boxes at the target node. The shortage of boxes can be calculated by the difference between the predicted demand for boxes at the target node and the current inventory. As an optional implementation, the shortage of boxes can be directly used as the quantity of boxes to be allocated. As another optional implementation, the quantity of boxes to be allocated, including a safety margin, can be set by combining the historical box usage patterns and transportation time of the target node.

[0044] At the recycling node, the RFID reader reads the box identifiers of the reusable packaging boxes in batches, compares them with the recycling records stored on the blockchain, realizes inventory, and updates the box status to be recycled.

[0045] The system tracks the cumulative number of recycling bins at each recycling node in real time. When the cumulative number of recycling bins reaches the preset recycling batch threshold, a recycling task order is generated, which includes the recycling node identifier, the number of recycling bins, and a list of bin identifiers. The recycling task order is then pushed to the recycling executor.

[0046] The preset recycling batch threshold refers to the critical recycling bin quantity that triggers the recycling task. Its specific value can be determined comprehensively based on factors such as the load capacity of the recycling vehicle, the storage capacity of the recycling node, and transportation costs. As an optional implementation method, the full load bin quantity of the recycling vehicle can be used as the threshold.

[0047] A big data-based shared scheduling and recycling system for reusable packaging includes: a data acquisition and verification module, a blockchain evidence storage and traceability module, a feature analysis and early warning module, and an intelligent scheduling and recycling module. The output of the data acquisition and verification module is connected to the input of the blockchain evidence storage and traceability module; the output of the blockchain evidence storage and traceability module is connected to the input of the feature analysis and early warning module; and the output of the feature analysis and early warning module is connected to the input of the intelligent scheduling and recycling module.

[0048] The data acquisition and verification module includes an RFID acquisition unit, a handover record unit, and a blockchain verification unit. The RFID acquisition unit acquires flow data including box identification and entry / exit time. The handover record unit records the handover type and handler information, and synchronously collects node inventory data. The blockchain verification unit performs consistency and integrity verification on the data and removes abnormal data.

[0049] The blockchain evidence storage and traceability module includes a hash processing unit, a hierarchical storage unit, and a traceability construction unit. The hash processing unit generates a unique evidence storage identifier for key circulation events of the container. The hierarchical storage unit stores data on the main chain and side chain according to access frequency. The traceability construction unit builds a traceability chain for the entire life cycle of the container based on hash association.

[0050] The feature analysis and early warning module includes a dataset construction unit, a clustering analysis unit, and a real-time monitoring unit. The dataset construction unit extracts historical data from the blockchain and processes it to generate a feature dataset. The clustering analysis unit divides node categories into clusters and constructs a feature early warning information set. The real-time monitoring unit matches real-time node data with early warning information to identify abnormal patterns.

[0051] The intelligent scheduling and recycling module includes a priority calculation unit, a scheduling task unit, and a recycling execution unit. The priority calculation unit calculates the comprehensive scheduling urgency coefficient of abnormal nodes and filters high-priority nodes. The scheduling task unit queries idle empty boxes, generates a scheduling order, and uploads it to the blockchain. The recycling execution unit completes the box inventory through RFID. When the cumulative recycling amount reaches the target, a recycling task order is generated and pushed.

[0052] Compared with the prior art, the beneficial effects of the present invention are:

[0053] 1. By using blockchain to verify the consistency and integrity of the circulation and inventory data of new material recycling boxes, and combining hash processing to build a full life cycle traceability chain, and relying on RFID devices to read box information in batches and compare it with the data on the chain to realize inventory, unlike the existing technology of manually checking boxes and data being scattered and easily tampered with, this invention realizes the reliable storage and inventory of box data, avoids the problems of data loss and tampering, and improves inventory efficiency and the traceability of circulation data;

[0054] 2. This invention relies on big data mining of historical circulation data of each node in the cyclic packaging process to construct a set of feature early warning information and calculate the urgency coefficient of node comprehensive scheduling. After screening high-priority nodes, it matches idle empty boxes on the blockchain to generate scheduling orders. Unlike the existing technology that manually counts supply and demand and schedules based on experience, this invention can identify abnormal box usage needs of nodes, realize intelligent matching of cyclic packaging supply and demand, avoid the problem of idle boxes and missing boxes at nodes, and improve the shared scheduling efficiency and resource reuse rate of cyclic packaging. Attached Figure Description

[0055] Figure 1 This is a flowchart illustrating a big data-based cyclic packaging shared scheduling and recycling method according to the present invention. Detailed Implementation

[0056] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] like Figure 1 As shown, this invention provides a technical solution: a cyclic packaging shared scheduling and recycling method based on big data, which includes the following steps:

[0058] Obtain circulation status data and inventory change data of reusable packaging boxes, and verify the data through blockchain;

[0059] The collected circulation status data and inventory change data are processed by hashing key circulation events using blockchain, and hierarchical storage is performed according to data access frequency and evidence storage requirements. A full life cycle traceability chain is built based on the circulation event hash association.

[0060] Extract historical transaction data from each node on the blockchain to construct a feature dataset;

[0061] Based on the aforementioned feature dataset, demand fluctuation patterns are identified, and a feature-based early warning information set is constructed.

[0062] Real-time monitoring of node flow data, identification of abnormal patterns matching the feature warning information set, calculation of the comprehensive scheduling urgency coefficient of nodes that trigger the warning mode, and selection of high-priority scheduling nodes based on this.

[0063] Based on the query of idle empty bins on the blockchain by high-priority scheduling nodes, a scheduling order is generated and uploaded to the chain; the bin information is read by the recycling node and compared with the records on the chain. When the cumulative recycling bin quantity reaches the threshold, a recycling task order is generated and pushed.

[0064] The process involves acquiring data on the circulation status and inventory changes of reusable packaging boxes, and verifying this data using blockchain technology. Specific steps include:

[0065] The system acquires circulation status data and inventory change data for reusable packaging boxes at each node. The circulation status data includes: box identification, entry / exit time, and operation node location collected by RFID reading and writing devices; and handover type and handler information recorded by handover terminals. The inventory change data includes: real-time empty box inventory and in-transit box quantity at each node; and the number of boxes leaving the warehouse, the number of boxes entering the warehouse, and the box dwell time for each node during historical periods extracted from the historical database.

[0066] The collected data is verified for consistency and integrity using blockchain technology.

[0067] Specifically, this includes: verifying the completeness of key fields for each data item based on preset data format rules; verifying whether the data has been tampered with during transmission based on hash algorithms; and removing duplicate uploads or data with abnormal timestamps. For data that passes verification, the data is classified based on the current circulation status of the container and node attributes, and each data item is labeled with a collection timestamp and geographical location tag. The classified data is written into the blockchain distributed ledger according to a preset storage strategy, with data related to the real-time location of the container written into a high-speed access channel and data related to historical circulation records written into a permanent evidence block.

[0068] The collected circulation status data and inventory change data are processed using blockchain to perform key circulation event hashing. Layered storage is implemented based on data access frequency and evidence storage requirements. A full lifecycle traceability chain is constructed based on the circulation event hash associations. Specific steps include:

[0069] Hash the key circulation events of each cycle of packaging boxes to generate a unique evidence identifier for that circulation event. The key circulation events include the box entry and exit events, cross-node transfer events, and recycling handover events.

[0070] Based on the access frequency and evidence storage requirements of the container data, a hierarchical storage strategy is implemented using blockchain: real-time location updates and current status change data of the container are written to the main chain; historical transfer records and past handover voucher data of the container are transferred to the side chain or distributed storage network according to preset storage rules; at the same time, a unique timestamp generated by a blockchain timestamp server is assigned to each piece of on-chain data to record the generation time or on-chain processing time of the data; based on the hash association between each transfer event, a full lifecycle traceability chain is constructed by including the hash value of the previous record in each new block or new record.

[0071] Extracting historical transaction data from each node on the blockchain to construct a feature dataset involves the following steps:

[0072] In the constructed full lifecycle traceability chain, the historical circulation data of the reusable packaging boxes of each node in the past preset time period is extracted. The historical circulation data includes: daily outbound box quantity, daily inbound box quantity, average box dwell time, and node business type identifier.

[0073] The extracted historical data is cleaned and normalized to remove outliers caused by abnormal data collection, and missing data is filled in by interpolation to generate a standardized time series dataset.

[0074] Feature extraction is performed on the standardized time-series dataset to obtain a set of feature information for each node. The feature information includes: periodic features, trend features, and volatility features.

[0075] The periodicity characteristic refers to the fluctuation pattern of box usage statistics on a weekly, monthly, or quarterly basis; the trend characteristic refers to the long-term growth slope or decline slope fitted by linear regression; and the volatility characteristic refers to statistical indicators such as variance, standard deviation, and kurtosis that reflect the degree of data dispersion.

[0076] The feature information of all nodes is aggregated to construct a global feature dataset.

[0077] Identifying demand fluctuation patterns and constructing a feature-based early warning information set involves the following steps:

[0078] Cluster analysis is performed on the global feature dataset to divide nodes with similar binning demand patterns into several node category clusters; where binning demand pattern represents the node binning feature vector composed of extracted periodic features, trend features, and volatility features.

[0079] Nodes with similar bin usage patterns are divided into several node clusters, as follows:

[0080] The similarity between the feature vectors of each node is calculated using a preset clustering algorithm (e.g., K-means, DBSCAN, or hierarchical clustering). Nodes with similarity higher than a preset threshold are grouped into the same cluster. The preset threshold can be set according to the type of clustering algorithm and the characteristics of the data distribution. For example, for the DBSCAN algorithm, the neighborhood radius ε can be determined based on the k-distance graph. For distance-based thresholds, a similarity threshold can be set based on business experience.

[0081] For each node cluster, feature information of nodes that have triggered emergency allocation or caused container shortage events in the past is extracted to form a feature warning information set for that type of node; the feature warning information set contains several feature items, each feature item corresponds to a feature extracted from the historical flow data, and a corresponding warning threshold range is set;

[0082] The aforementioned warning threshold range refers to a preset critical value or critical interval for each feature item to determine whether the real-time feature value is abnormal. It can be a one-way threshold (e.g., triggering a warning if the value is greater than or less than a certain value) or a two-way interval threshold (e.g., the feature value is normal if it is within the preset interval, and triggers a warning if it exceeds the interval). The threshold range can be determined based on the statistical distribution of historical data, for example, by taking the mean of historical data plus or minus a certain number of standard deviations, or by taking a certain percentile (e.g., the 90th percentile) of historical normal data as the threshold. It can also be set directly based on business experience. For example, for the feature item of dwell time, it can be set to not exceed 48 hours according to the node turnover requirements. Those skilled in the art can flexibly set appropriate threshold ranges for different feature items according to actual data characteristics and business needs. This invention does not impose specific numerical limitations on this.

[0083] Write the feature warning information set of each node category cluster into the blockchain.

[0084] Real-time monitoring of node flow data and identification of abnormal patterns matching the feature-based early warning information set. Specific steps include:

[0085] By using the real-time on-chain node flow data, and based on the feature extraction method of the historical flow data, the real-time feature vector of each node within the current time window is calculated.

[0086] Based on the category cluster to which each node belongs, the corresponding feature warning information set is retrieved, and each feature item in the real-time feature vector is matched with the corresponding feature item in the feature warning information set;

[0087] If the feature values ​​of a node in its real-time feature vector exceed the warning threshold range of the corresponding feature item in the feature warning information set for two consecutive time windows, then the node is determined to have triggered the warning mode, and the two real-time records corresponding to the node in these two time windows are extracted.

[0088] The comprehensive scheduling urgency coefficient of the node that triggers the early warning mode is calculated, and the specific steps include:

[0089] Obtain the two real-time records corresponding to the node that triggered the early warning mode within two consecutive time windows, denoted as R. a and R b , where R a The time was earlier than R b ; respectively from Ra and R b Extract the set of features that trigger the early warning mode, denoted as F. a and F b ;

[0090] Calculate F a Relative to F b The set of distinguishing features F(ab) = F a -F a ∩F b , and F b Relative to F a The set of distinguishing features F'(ab) = F b -F a ∩F b ;

[0091] For each feature term in F(ab), calculate its deviation from the corresponding feature term in the feature warning information set of the category cluster to which the node belongs, and take the product of all deviations as the first comprehensive scheduling urgency coefficient β. a β is defined as follows: a =∏ f∈Fab θ f ; where θ f The deviation of feature term f is represented by the ratio of the real-time feature value to the warning threshold.

[0092] Similarly, calculate the second integrated scheduling urgency coefficient β corresponding to F'(ab). b ;

[0093] If β b >β a And β b -β a If the value is greater than η, then the node is determined to be a high-priority scheduling node, and its identifier is added to the list of nodes to be scheduled. Here, η represents the preset comprehensive urgency coefficient threshold.

[0094] The specific value of the preset comprehensive urgency coefficient threshold can be set according to actual business needs and historical data analysis results. As an optional implementation method, it is possible to analyze cases in historical scheduling data that successfully avoid box shortage events, statistically analyze the corresponding comprehensive urgency coefficient difference distribution, and select the percentile that enables the recall rate in historical samples to reach a preset target (such as 90%) as the preset comprehensive urgency coefficient threshold η. The recall rate is defined as the proportion of the number of emergency events that are correctly identified as needing priority scheduling to the total number of actual emergency events.

[0095] Based on the query of the source nodes of idle empty boxes on the blockchain by high-priority scheduling nodes, a scheduling task order is generated and stored on the blockchain; at the recycling node, the box information is read in batches by RFID devices and compared with the on-chain records. When the recycling threshold is reached, a recycling task order is generated and pushed. The specific steps include:

[0096] Based on the high-priority scheduling node as the target node, query the blockchain for available source nodes with idle empty boxes, determine the amount of boxes to be allocated based on the amount of boxes missing from the target node, generate a scheduling task order containing the amount of boxes to be allocated, source node identifier, and target node identifier, and store the scheduling task order on the blockchain for evidence.

[0097] The quantity of boxes to be allocated can be determined based on the shortage of boxes at the target node. The shortage of boxes can be calculated by the difference between the predicted demand for boxes at the target node and the current inventory. As an optional implementation, the shortage of boxes can be directly used as the quantity of boxes to be allocated. As another optional implementation, the quantity of boxes to be allocated, including a safety margin, can be set by combining the historical box usage patterns and transportation time of the target node.

[0098] At the recycling node, the RFID reader reads the box identifiers of the reusable packaging boxes in batches, compares them with the recycling records stored on the blockchain, realizes inventory, and updates the box status to be recycled.

[0099] The system tracks the cumulative number of recycling bins at each recycling node in real time. When the cumulative number of recycling bins reaches the preset recycling batch threshold, a recycling task order is generated, which includes the recycling node identifier, the number of recycling bins, and a list of bin identifiers. The recycling task order is then pushed to the recycling executor.

[0100] The preset recycling batch threshold refers to the critical recycling bin quantity that triggers the recycling task. Its specific value can be determined comprehensively based on factors such as the load capacity of the recycling vehicle, the storage capacity of the recycling node, and transportation costs. As an optional implementation method, the full load bin quantity of the recycling vehicle can be used as the threshold.

[0101] In Example 1: Data collection and blockchain verification of reusable packaging boxes are carried out. Radio frequency identification devices are used to collect the box's identity, inbound and outbound information, and operation node location and other circulation status data. The handover terminal synchronously records the handover type of the box and the relevant information of the person in charge. At the same time, real-time empty box inventory, in-transit box quantity and historical inventory change data of each node are collected. All the above-mentioned raw data are uploaded to the blockchain network. The blockchain is used to complete the consistency and integrity verification of the data and eliminate abnormal data such as duplicate uploads and missing information.

[0102] Hash processing is performed on key circulation events such as the entry and exit of reusable packaging boxes, cross-node transfer, and recycling handover to generate unique evidence identifiers for each event. A hierarchical storage strategy is implemented based on the access frequency of box data and evidence storage requirements. Data related to the real-time location update and current status change of the boxes are written to the main chain, while historical circulation records and past handover voucher data are transferred to the side chain or distributed storage network. Each piece of data on the chain is marked with a unique time stamp. Based on the hash association between each circulation event, a full lifecycle traceability chain for reusable packaging boxes is built.

[0103] The historical circulation data of cyclic packaging boxes of each node are extracted from the traceability chain of the blockchain. The extracted data is cleaned and standardized, abnormal data is removed and missing information is filled in. Periodic, trend and fluctuation characteristics in the data are extracted. The feature information of all nodes is summarized to construct a global feature dataset. Cluster analysis is carried out on the dataset to divide nodes with similar box demand patterns into different clusters. Feature information that has triggered emergency allocation or box shortage events in each cluster is extracted. Based on this, a feature warning information set for the corresponding category is constructed, and all sets are written into the blockchain.

[0104] Real-time collection of container flow data at each node; calculation of real-time feature vectors for each node based on the feature extraction method of historical data; retrieval of the corresponding feature warning information set according to the category cluster to which the node belongs; matching of real-time feature vectors with warning information; calculation of comprehensive scheduling urgency coefficient for nodes that trigger warning mode; selection of high-priority scheduling nodes based on coefficient and inclusion of them in the list of nodes to be scheduled.

[0105] Targeting high-priority scheduling nodes, the system queries the blockchain for available source nodes with idle empty boxes. Combining the target node's box usage needs, a scheduling task order is generated. After the task order is stored on the blockchain, the intelligent allocation of boxes is completed. At each recycling node, box information is read in batches using RFID devices and compared with the records to be recycled on the blockchain to achieve automatic inventory. The cumulative number of recycled boxes at each node is counted in real time. When the number of boxes reaches the preset recycling standard, a recycling task order is generated and pushed to the recycling executor, who then completes the subsequent recycling and transportation work.

[0106] A big data-based shared scheduling and recycling system for reusable packaging includes: a data acquisition and verification module, a blockchain evidence storage and traceability module, a feature analysis and early warning module, and an intelligent scheduling and recycling module. The output of the data acquisition and verification module is connected to the input of the blockchain evidence storage and traceability module; the output of the blockchain evidence storage and traceability module is connected to the input of the feature analysis and early warning module; and the output of the feature analysis and early warning module is connected to the input of the intelligent scheduling and recycling module.

[0107] The data acquisition and verification module includes an RFID acquisition unit, a handover record unit, and a blockchain verification unit. The RFID acquisition unit acquires flow data including box identification and entry / exit time. The handover record unit records the handover type and handler information, and synchronously collects node inventory data. The blockchain verification unit performs consistency and integrity verification on the data and removes abnormal data.

[0108] The blockchain evidence storage and traceability module consists of a hash processing unit, a hierarchical storage unit, and a traceability construction unit. The hash processing unit generates a unique evidence storage identifier for key circulation events of the container. The hierarchical storage unit stores data on the main chain and side chain according to the access frequency. The traceability construction unit builds a traceability chain for the entire life cycle of the container based on hash association.

[0109] The feature analysis and early warning module includes a dataset construction unit, a clustering analysis unit, and a real-time monitoring unit. The dataset construction unit extracts historical data from the blockchain and processes it to generate a feature dataset. The clustering analysis unit divides node categories into clusters and constructs a feature early warning information set. The real-time monitoring unit matches real-time node data with early warning information to identify abnormal patterns.

[0110] The intelligent scheduling and recycling module consists of a priority calculation unit, a scheduling task unit, and a recycling execution unit. The priority calculation unit calculates the comprehensive scheduling urgency coefficient of abnormal nodes and filters high-priority nodes. The scheduling task unit queries idle empty boxes, generates a scheduling order, and uploads it to the blockchain. The recycling execution unit completes the box inventory through RFID. When the cumulative recycling amount reaches the target, a recycling task order is generated and pushed.

[0111] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for cyclical packaging, sharing, scheduling, and recycling based on big data, characterized in that: The method includes the following steps: Obtain circulation status data and inventory change data of reusable packaging boxes, and verify the data through blockchain; The collected circulation status data and inventory change data are processed by hashing key circulation events using blockchain, and hierarchical storage is performed according to data access frequency and evidence storage requirements. A full life cycle traceability chain is built based on the circulation event hash association. Hash the key circulation events of each cycle of packaging boxes to generate a unique evidence identifier for the key circulation events. The key circulation events include the box entry and exit events, cross-node transfer events, and recycling handover events. Based on the access frequency and evidence storage requirements of the container data, a hierarchical storage strategy is implemented using blockchain: real-time location updates and current status change data of the container are written to the main chain; historical transfer records and past handover voucher data of the container are transferred to the side chain or distributed storage network according to preset storage rules; at the same time, a unique timestamp generated by a blockchain timestamp server is assigned to each piece of on-chain data to record the generation time or on-chain processing time of the data; based on the hash association between each transfer event, a full lifecycle traceability chain is constructed by including the hash value of the previous record in each new block or new record; Extract historical transaction data from each node on the blockchain to construct a feature dataset; Based on the aforementioned feature dataset, demand fluctuation patterns are identified, and a feature-based early warning information set is constructed. Real-time monitoring of node flow data, identification of abnormal patterns matching the feature warning information set, calculation of the comprehensive scheduling urgency coefficient of nodes that trigger the warning mode, and selection of high-priority scheduling nodes based on this. Based on the query of idle empty bins on the blockchain by high-priority scheduling nodes, a scheduling order is generated and uploaded to the chain; the bin information is read by the recycling node and compared with the records on the chain. When the cumulative number of recycled bins reaches the threshold, a recycling task order is generated and pushed.

2. The method for cyclic packaging, sharing, scheduling, and recycling based on big data according to claim 1, characterized in that: The process involves acquiring data on the circulation status and inventory changes of reusable packaging boxes, and verifying this data using blockchain technology. Specific steps include: The system acquires circulation status data and inventory change data for reusable packaging boxes at each node. The circulation status data includes: box identification, entry / exit time, and operation node location collected by RFID reading and writing devices; and handover type and handler information recorded by handover terminals. The inventory change data includes: real-time empty box inventory and in-transit box quantity at each node; and the number of boxes leaving the warehouse, the number of boxes entering the warehouse, and the box dwell time for each node during historical periods extracted from the historical database. The collected data is verified for consistency and integrity using blockchain technology.

3. The method for cyclic packaging, sharing, scheduling, and recycling based on big data according to claim 2, characterized in that: Extracting historical transaction data from each node on the blockchain to construct a feature dataset involves the following steps: In the constructed full lifecycle traceability chain, the historical circulation data of the reusable packaging boxes of each node in the past preset time period is extracted. The historical circulation data includes: daily outbound box quantity, daily inbound box quantity, average box dwell time, and node business type identifier. The extracted historical data is cleaned and normalized to remove outliers caused by abnormal data collection, and missing data is filled in by interpolation to generate a standardized time series dataset. Feature extraction is performed on the standardized time-series dataset to obtain a set of feature information for each node. The feature information includes: periodic features, trend features, and volatility features. The feature information of all nodes is aggregated to construct a global feature dataset.

4. The method for cyclic packaging, sharing, scheduling, and recycling based on big data according to claim 3, characterized in that: Identifying demand fluctuation patterns and constructing a feature-based early warning information set involves the following steps: Cluster analysis is performed on the global feature dataset to divide nodes with similar binning demand patterns into several node category clusters; where binning demand pattern represents the node binning feature vector composed of extracted periodic features, trend features, and volatility features. For each node cluster, feature information of nodes in that cluster that have triggered emergency allocation or caused box shortage events in the past is extracted to form a feature warning information set for nodes in that cluster; the feature warning information set contains several feature items, each feature item corresponds to a feature extracted from the historical flow data, and has a corresponding warning threshold range; Write the feature warning information set of each node category cluster into the blockchain.

5. The method for cyclic packaging, sharing, scheduling, and recycling based on big data according to claim 4, characterized in that: Real-time monitoring of node flow data and identification of abnormal patterns matching the feature-based early warning information set. Specific steps include: By using the real-time on-chain node flow data, and based on the feature extraction method of the historical flow data, the real-time feature vector of each node within the current time window is calculated. Based on the category cluster to which each node belongs, the corresponding feature warning information set is retrieved, and each feature item in the real-time feature vector is matched with the corresponding feature item in the feature warning information set; If the feature values ​​of a node in its real-time feature vector exceed the warning threshold range of the corresponding feature item in the feature warning information set for two consecutive time windows, then the node is determined to have triggered the warning mode, and the two real-time records corresponding to the node in these two time windows are extracted.

6. The method for cyclic packaging, sharing, scheduling, and recycling based on big data according to claim 5, characterized in that: The comprehensive scheduling urgency coefficient of the node that triggers the early warning mode is calculated, and the specific steps include: Obtain the two real-time records corresponding to the node that triggered the early warning mode within two consecutive time windows, denoted as R. a and R b , where R a The time was earlier than R b ; respectively from R a and R b Extract the set of features that trigger the early warning mode, denoted as F. a and F b ; Calculate F a Relative to F b The set of distinguishing features F(ab) = F a -F a ∩F b , and F b Relative to F a The set of distinguishing features F'(ab) = F b -F a ∩F b ; For each feature term in F(ab), calculate its deviation from the corresponding feature term in the feature warning information set of the category cluster to which the node belongs, and take the product of all deviations as the first comprehensive scheduling urgency coefficient β. a β is defined as follows: a =∏ f∈Fab θ f ; where θ f The deviation of feature term f is represented by the ratio of the real-time feature value to the warning threshold. Similarly, calculate the second integrated scheduling urgency coefficient β corresponding to F'(ab). b ; If β b >β a And β b -β a If the value is greater than η, then the node is determined to be a high-priority scheduling node, and its identifier is added to the list of nodes to be scheduled. Here, η represents the preset comprehensive urgency coefficient threshold.

7. The method for cyclic packaging, sharing, scheduling, and recycling based on big data according to claim 6, characterized in that: Based on the query of the source nodes of idle empty boxes on the blockchain by high-priority scheduling nodes, a scheduling task order is generated and stored on the blockchain; at the recycling node, the box information is read in batches by RFID devices and compared with the on-chain records. When the recycling threshold is reached, a recycling task order is generated and pushed. The specific steps include: Based on the high-priority scheduling node as the target node, query the blockchain for available source nodes with idle empty boxes, determine the amount of boxes to be allocated based on the amount of boxes missing from the target node, generate a scheduling task order containing the amount of boxes to be allocated, source node identifier, and target node identifier, and store the scheduling task order on the blockchain for evidence. At the recycling node, the RFID reader reads the box identifiers of the reusable packaging boxes in batches, compares them with the recycling records stored on the blockchain, realizes inventory, and updates the box status to be recycled. The system tracks the cumulative number of recycling bins at each recycling node in real time. When the cumulative number of recycling bins reaches the preset recycling batch threshold, a recycling task order is generated, which includes the recycling node identifier, the number of recycling bins, and a list of bin identifiers. The recycling task order is then pushed to the recycling executor.

8. A big data-based shared scheduling and recycling system for reusable packaging, applied to the big data-based shared scheduling and recycling method for reusable packaging as described in any one of claims 1-7, characterized in that: The system includes: a data acquisition and verification module, a blockchain evidence storage and traceability module, a feature analysis and early warning module, and an intelligent scheduling and recycling module; the output of the data acquisition and verification module is connected to the input of the blockchain evidence storage and traceability module; the output of the blockchain evidence storage and traceability module is connected to the input of the feature analysis and early warning module; and the output of the feature analysis and early warning module is connected to the input of the intelligent scheduling and recycling module. The data acquisition and verification module includes an RFID acquisition unit, a handover record unit, and a blockchain verification unit. The RFID acquisition unit acquires flow data including box identification and entry / exit time. The handover record unit records the handover type and handler information, and synchronously collects node inventory data. The blockchain verification unit performs consistency and integrity verification on the data and removes abnormal data. The blockchain evidence storage and traceability module includes a hash processing unit, a hierarchical storage unit, and a traceability construction unit. The hash processing unit generates a unique evidence storage identifier for key circulation events of the container. The hierarchical storage unit stores data on the main chain and side chain according to access frequency. The traceability construction unit builds a traceability chain for the entire life cycle of the container based on hash association. The feature analysis and early warning module includes a dataset construction unit, a clustering analysis unit, and a real-time monitoring unit. The dataset construction unit extracts historical data from the blockchain and processes it to generate a feature dataset. The clustering analysis unit divides node categories into clusters and constructs a feature early warning information set. The real-time monitoring unit matches real-time node data with early warning information to identify abnormal patterns. The intelligent scheduling and recycling module includes a priority calculation unit, a scheduling task unit, and a recycling execution unit. The priority calculation unit calculates the comprehensive scheduling urgency coefficient of abnormal nodes and filters high-priority nodes. The scheduling task unit queries idle empty boxes, generates a scheduling order, and uploads it to the blockchain. The recycling execution unit completes the box inventory through RFID. When the cumulative recycling amount reaches the target, a recycling task order is generated and pushed.

Images