Intelligent shelf system and management method thereof

Abstract

The application provides an intelligent shelf system and a management method thereof, and belongs to the technical field of shelf management.The method comprises the following steps: performing multi-type sensing device topology architecture design on an intelligent shelf unit, constructing a multi-type sensing signal fusion network topology architecture data, collecting related information of the shelf in real time based on the multi-type sensing signal fusion network topology architecture data, forming a time sequence shelf original monitoring data stream, and performing preprocessing on the time sequence shelf original monitoring data stream by using a data preprocessing algorithm to generate preprocessed shelf monitoring data.The distributed data storage architecture optimization and the redundant reliable transmission strategy are adopted, so that the safety and availability of data storage are improved, and the rapid storage and access of data are ensured.

Description

TECHNICAL FIELD

[0001] The application provides an intelligent shelf system and a management method thereof, and belongs to the technical field of shelf management. BACKGROUND

[0002] In a traditional shelf management scenario, inventory checking relies on manual operation, which is inefficient and prone to errors; shelf display adjustment lacks scientific basis and is difficult to meet the diversified needs of consumers; at the same time, risks that may occur during shelf operation, such as out-of-stock of goods and disordered placement, cannot be timely warned and handled; these problems result in poor overall effect of shelf management, increase operating costs, and reduce sales efficiency; therefore, an intelligent shelf management method is needed to solve the above problems. SUMMARY

[0003] The application provides an intelligent shelf system and a management method thereof, which are used to solve the problems mentioned in the background.

[0004] The application provides an intelligent shelf management method, which comprises the following steps:

[0005] S1, a multi-type sensing device topology architecture design is performed on an intelligent shelf unit, a multi-type sensing signal fusion network topology architecture data is constructed, real-time collection of related information of the shelf is performed based on the multi-type sensing signal fusion network topology architecture data, time sequence shelf original monitoring data flow is formed, a data preprocessing algorithm is used to preprocess the time sequence shelf original monitoring data flow, and preprocessed shelf monitoring data is generated;

[0006] S2, a distributed data storage node optimization algorithm is used to perform dynamic layout and resource allocation on a storage node based on the multi-type sensing signal fusion network topology architecture data, a distributed data storage architecture is constructed, a multi-path redundant transmission strategy is used to transmit the preprocessed shelf monitoring data to multiple storage nodes at the same time by using the distributed data storage architecture, and a redundant monitoring upload data packet is generated;

[0007] S3, a unique upload credential information is generated for each redundant monitoring upload data packet, difference analysis is performed on the redundant monitoring upload data received by different storage nodes based on the upload credential information, a monitoring upload difference data report is generated, difference conflict resolution is performed on the redundant monitoring upload data according to the monitoring upload difference data report, and accurate shelf unit monitoring data is generated;

[0008] S4, a dynamic data sharding strategy is used to flexibly shard the data according to multiple dimensions based on the characteristics and management requirements of the shelf unit monitoring data, monitoring sharded feature data is generated, an intelligent cache synchronization algorithm is used to cache the commonly used data shards to a cache device based on the monitoring sharded feature data, and the cache data is synchronized in real time according to the update of the data;

[0009] S5. Based on cache management log data, use the cache status evaluation model to comprehensively evaluate the cache status and generate monitoring cache status data; based on the monitoring cache status data, combined with the business objectives and key performance indicators of shelf management, use a comprehensive analysis algorithm to conduct a comprehensive analysis of the management status of the monitoring shard feature data and generate comprehensive data management indicators.

[0010] S6. Based on the comprehensive data management index, the system uses autonomous intelligent decision-making algorithms to intelligently analyze and make decisions on the management of goods on the shelves, the layout of the shelves, and the operational risks. Based on the results of the autonomous intelligent decision-making, the system adjusts and optimizes the shelves in real time through the intelligent control system. At the same time, the results of the decision execution are fed back into the system to form a closed-loop management cycle.

[0011] The present invention proposes an intelligent shelf system, comprising:

[0012] One or more processors;

[0013] Memory, used to store one or more programs.

[0014] Wherein, when the one or more programs are executed by the one or more processors, the one or more processors are made to implement the method described in any one of the above.

[0015] The beneficial effects of this invention are as follows: By fusing multiple types of sensing signals and preprocessing data, various types of information about the shelves can be acquired in real time and accurately, providing comprehensive and reliable data support for management decisions; the adoption of a distributed data storage architecture optimization and redundant and reliable transmission strategy improves the security and availability of data storage, while ensuring rapid data storage and access; autonomous intelligent decision-making based on a comprehensive data management index can automatically generate scientific and reasonable management strategies according to real-time data and market dynamics, realizing intelligent management and optimization of the shelves; and a closed-loop control cycle of decision-making-execution-feedback is formed, which can promptly identify and resolve problems that arise in the process of shelf management, continuously improving the overall effectiveness and operational efficiency of shelf management. Attached Figure Description

[0016] Figure 1 This is a diagram illustrating the steps of the method described in this invention;

[0017] Figure 2 As described in this invention Figure 1 Detailed flowchart of steps in S2. Detailed Implementation

[0018] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0019] One embodiment of the present application, as shown in Figure 1 An intelligent shelf management method, the method comprising:

[0020] S1, a multi-type sensing device topology architecture design is performed on the intelligent shelf unit, a multi-type sensing signal fusion network topology architecture data is constructed, based on the multi-type sensing signal fusion network topology architecture data, the related information of the shelf is collected in real time, the time sequence shelf original monitoring data stream is formed, the data preprocessing algorithm is used to preprocess the time sequence shelf original monitoring data stream, and the preprocessed shelf monitoring data is generated;

[0021] S2, according to the multi-type sensing signal fusion network topology architecture data, a distributed data storage node optimization algorithm is used to dynamically layout and allocate resources to the storage node, and a distributed data storage architecture is constructed; using the distributed data storage architecture, a multi-path redundant transmission strategy is used to transmit the preprocessed shelf monitoring data to multiple storage nodes at the same time, and a redundant monitoring upload data packet is generated;

[0022] S3, a unique upload credential information is generated for each redundant monitoring upload data packet, based on the upload credential information, the redundant monitoring upload data received by different storage nodes is analyzed, a monitoring upload difference data report is generated, the redundant monitoring upload data is analyzed according to the monitoring upload difference data report, and the accurate shelf unit monitoring data is generated;

[0023] S4, according to the characteristics and management requirements of the shelf unit monitoring data, a dynamic data sharding strategy is used to flexibly shard the data according to multiple dimensions, and monitoring sharding feature data is generated; based on the monitoring sharding feature data, an intelligent cache synchronization algorithm is used to cache the commonly used data shards to a cache device, and the cache data is synchronized in real time according to the update of the data;

[0024] S5, according to the cache management log data, a cache state evaluation model is used to comprehensively evaluate the cache state, and monitoring cache state data is generated; based on the monitoring cache state data, combined with the business objectives and key performance indicators of the shelf management, a comprehensive analysis algorithm is used to analyze the management state of the monitoring sharding feature data, and a comprehensive data management index is generated;

[0025] S6, based on the comprehensive data management index, an autonomous intelligent decision algorithm is used to intelligently analyze and decide the shelf product control, display layout and operation risk; according to the autonomous intelligent decision result, the shelf is adjusted and optimized in real time through the intelligent control system; at the same time, the decision execution result is fed back to the system, forming a closed-loop control cycle.

[0026] The working principle of the above technical solution is that a multi-type sensing device topology architecture design is performed on the intelligent shelf unit, and a plurality of devices such as a pressure sensing device, a visual recognition module, an electronic tag reader, an infrared sensor, and an environmental parameter monitoring sensor are integrated to construct a multi-type sensing signal fusion network topology architecture data. The architecture data records in detail the installation position of each type of sensing device on the shelf, the mutual connection relationship, the sensing range, and the working parameters and other information, providing a basis for subsequent data acquisition and processing. Based on the multi-type sensing signal fusion network topology architecture data, real-time acquisition of related information such as shelf product weight data, product image data, product electronic tag information, product in-out shelf state data, and shelf environmental parameter data is performed to form time-series shelf original monitoring data flow. The product weight data can reflect the number change of the product; the product image data is used to identify the type, appearance state, and the like of the product; the product electronic tag information includes detailed attributes, production date, shelf life, and other information of the product; the product in-out shelf state data records the in-out time of the product; and the shelf environmental parameter data includes temperature, humidity, and light intensity. A data preprocessing algorithm is used to preprocess the time-series shelf original monitoring data flow, including data cleaning, data normalization, and data noise reduction operations. The data cleaning removes noise, outliers, and repeated data in the data. The data normalization converts data of different dimensions to a unified standard range, facilitating subsequent analysis and processing. The data noise reduction uses a filtering algorithm to reduce random noise in the data and improve the quality of the data to generate preprocessed shelf monitoring data.

[0027] According to the multi-type sensing signal fusion network topology architecture data, a distributed data storage node optimization algorithm is used to dynamically layout and allocate resources to the storage nodes by comprehensively considering factors such as storage capacity, processing capability, network bandwidth, geographical location distribution, and data access frequency, thereby constructing a distributed data storage architecture. The architecture can realize near storage and fast access of data, and improve the performance and availability of data storage. Using the distributed data storage architecture, a multi-path redundant transmission strategy is used to simultaneously transmit the preprocessed shelf monitoring data to a plurality of storage nodes to generate redundant monitoring upload data packets. The data is encrypted during transmission to ensure data security. Through redundant transmission, even if some storage nodes or transmission links fail, the integrity and reliability of the data can be ensured, and data loss can be avoided.

[0028] A unique upload credential information is generated for each redundant monitoring upload data packet, which contains data source identification, upload timestamp, data checksum, digital signature, encryption key and other elements; the upload credential information is used to ensure the traceability of the data source, the verifiability of the transmission process, the integrity and authenticity of the data, and the confidentiality of the data; based on the upload credential information, the redundant monitoring upload data received by different storage nodes is analyzed for differences; by comparing the checksum, timestamp and specific content of the data, etc., the data parts with differences are found out, and a monitoring upload difference data report is generated; a data consistency negotiation algorithm and an intelligent voting decision mechanism are used to resolve the differences and conflicts of the redundant monitoring upload data according to the monitoring upload difference data report; in the negotiation process, data interaction and information sharing are carried out between the storage nodes, and factors such as data source reliability and data update time are considered to determine the final correct data version and generate accurate shelf unit monitoring data, ensuring the consistency and accuracy of the data stored in each storage node.

[0029] According to the characteristics and management requirements of the shelf unit monitoring data, a dynamic data sharding strategy is used to flexibly shard the data according to multiple dimensions such as product category, time period, access frequency and data importance, and generate monitoring sharded feature data; the dynamic data sharding strategy can be dynamically adjusted according to the real-time changes of the data and management requirements, for example, for the data of hot-selling products, a finer-grained sharding method is used to improve the processing efficiency and access speed of the data; based on the monitoring sharded feature data, an intelligent cache synchronization algorithm is used to cache frequently used data shards to a cache device, and the cache data is synchronized in real time according to the update of the data; the intelligent cache synchronization algorithm can dynamically adjust the cache strategy according to the access mode and update frequency of the data to improve the hit rate and utilization rate of the cache; at the same time, cache management log data is generated to record the storage location, update time, access frequency, hit rate and cache state of the cache data, providing a basis for the evaluation and optimization of the cache state;

[0030] According to the cache management log data, a cache state evaluation model is used to comprehensively evaluate the cache state by considering indicators such as cache hit rate, cache utilization rate, cache update frequency, data access delay and cache error rate, and generate monitoring cache state data; this data can intuitively reflect the performance and health status of the cache system, providing a decision basis for the optimization and management of the cache; based on the monitoring cache state data, combined with the business objectives and key performance indicators of shelf management, a comprehensive analysis algorithm is used to analyze the management state of the monitoring sharded feature data; considering factors such as product inventory accuracy, display layout rationality, operational risk level, sales performance impact and customer satisfaction, a comprehensive data management index is generated; the comprehensive data management index is a comprehensive indicator for measuring the overall effectiveness of shelf management, which can provide a scientific and comprehensive basis for decision-making in shelf management;

[0031] Based on the comprehensive data management index, the self-intelligent decision algorithm is used to intelligently analyze and decide the goods control, display layout and operation risk of the shelf; the self-intelligent decision algorithm combines the preset rule base, expert knowledge model and machine learning algorithm, automatically generates decision suggestions such as goods replenishment strategy, goods allocation scheme, display adjustment scheme and operation risk early warning measures according to real-time data and market dynamics; according to the self-intelligent decision result, the intelligent control system is used to real-time adjust and optimize the shelf; for example, the goods replenishment process is automatically triggered, the replenishment order is sent to the supplier; the display layout of the shelf is adjusted, the hot-selling goods are placed in a more eye-catching position; the operation risk response mechanism is started, the expired goods are removed, the equipment failure is repaired in time, etc.; at the same time, the decision execution result is fed back to the system to form a closed-loop control cycle, continuously optimize the shelf management strategy and method, and improve the intelligent level and operation efficiency of the shelf management.

[0032] The effect of the above technical solution is that: by integrating multiple types of sensing devices, real-time collection of various data on the shelf can be realized, ensuring that the data accurately reflects the state, quantity and environmental parameters of the goods; through the distributed data storage architecture and multi-path redundant transmission strategy, the data can be quickly accessed, stored nearby, and redundant backup can be realized, avoiding data loss caused by single point failure and enhancing the reliability of data storage;

[0033] The data is encrypted during transmission, combined with the redundant upload and upload credential generation mechanism, to ensure the security, traceability of the source and integrity verification during transmission; through the difference analysis and data consistency negotiation algorithm, the consistency of data between different storage nodes can be ensured, and the conflict of redundant data can be solved through the intelligent voting mechanism, improving the accuracy of the data;

[0034] The intelligent cache synchronization algorithm is adopted, which can dynamically adjust the cache strategy according to the data access frequency and update, improve the cache hit rate and utilization, and optimize the system performance; through the self-intelligent decision algorithm, the replenishment strategy, allocation scheme, display adjustment scheme, etc. can be automatically generated according to real-time data, market dynamics and preset rules, ensuring that the shelf management decision is more flexible, timely and efficient;

[0035] Since the system can automatically perform replenishment, allocation, display adjustment and other operations, the demand for manual operation is significantly reduced, thereby reducing the management cost and consumption of human resources; the dynamic data sharding strategy and the generation of monitoring sharding feature data enable the system to flexibly adjust according to different management needs and data changes, effectively improving the accuracy and flexibility of shelf management;

[0036] By optimizing the shelf layout, accurate replenishment decision and timely operation risk response measures, a better shopping experience and product display can be provided, thereby improving customer satisfaction and sales performance; the distributed architecture and flexible resource allocation strategy provide a good foundation for future expansion and upgrade, which can adapt to larger scale shelf management requirements and ensure the sustainability of the system.

[0037] In one embodiment of the present application, the S1 comprises:

[0038] S11, comprehensive spatial analysis and function planning of the intelligent shelf unit, according to the size, number of layers, product display method and management requirements of the shelf, the installation position of various sensing devices is determined;

[0039] S12, the mutual connection relationship between various sensing devices is clear, and a network topology is constructed by combining wired and wireless communication methods;

[0040] S13, record the sensing range of various sensing devices, determine the area and boundary that each device can effectively monitor through actual test and simulation analysis; set the working parameters according to the working principle and performance characteristics of different types of sensing devices;

[0041] S14, based on the constructed multi-type sensing signal fusion network topology architecture data, start various sensing devices, and collect real-time information of the shelf according to the set sampling frequency;

[0042] S15, the collected various data are preliminarily integrated, and time sequence shelf original monitoring data stream is generated according to time sequence; data cleaning algorithm is used to process the time sequence shelf original monitoring data stream, data normalization method is used to convert different dimension data into unified standard range; data noise reduction algorithm is used to further reduce noise interference in the data; after preprocessing, preprocessed shelf monitoring data are generated.

[0043] The working principle of the above technical solution is: comprehensive spatial analysis and function planning of the intelligent shelf unit, according to the size, number of layers, product display method and management requirements of the shelf, the installation position of various sensing devices is determined; for example, the pressure sensing device is installed below the support structure of each layer of shelf to accurately sense the weight change of goods; the visual recognition module is installed above or on the side of the shelf to ensure that it can fully cover the goods on the shelf and obtain clear image data; the electronic tag reader is installed at the end or side of the shelf to facilitate data interaction with the electronic tag on the goods; the infrared sensor is installed at the entrance and exit position of the shelf to monitor the in-out shelf state of the goods; the environmental parameter monitoring sensor is reasonably distributed around the shelf according to the environmental characteristics of the shelf, and real-time monitors the environmental parameters such as temperature, humidity and light intensity;

[0044] The interconnection relationship between various types of sensing devices is determined, and a network topology is constructed by combining wired and wireless communication methods. For example, for devices with close proximity and large data transmission volume, such as pressure sensing devices and electronic tag readers, wired connection is used to ensure data transmission stability and high speed. For devices with scattered distribution and strong mobility, such as visual recognition modules and infrared sensors, wireless connection methods such as Wi-Fi or Bluetooth are used to improve the flexibility and convenience of device deployment.

[0045] The sensing range of various types of sensing devices is recorded, and through actual testing and simulation analysis, the area and boundary that each device can effectively monitor are determined. For example, the visual range of visual recognition modules and the sensing distance of infrared sensors, providing accurate basic information for subsequent data collection and analysis. According to the working principle and performance characteristics of different types of sensing devices, reasonable working parameters are set. For example, the sensitivity of pressure sensing devices, the resolution and frame rate of visual recognition modules, and the read-write power of electronic tag readers, to ensure that the devices can work normally and obtain accurate and reliable data.

[0046] Based on the constructed multi-type sensing signal fusion network topology architecture data, start various types of sensing devices, and collect real-time information of the shelf according to the set sampling frequency. The pressure sensing device collects real-time weight data of the goods and converts the data into electrical signals for transmission to the data processing unit. The visual recognition module continuously captures images of the goods on the shelf, and the image data is transmitted to the computer for subsequent processing through the image acquisition card. The electronic tag reader periodically reads the information in the goods electronic tag, including the detailed attributes, production date, shelf life, etc. of the goods. The infrared sensor monitors the in-out status of the goods in real time, and when there are goods in and out, it generates corresponding electrical signals and transmits them to the control system. The environmental parameter monitoring sensor continuously collects data such as temperature, humidity, and light intensity of the shelf environment, and uploads the data to the data storage system through the communication interface.

[0047] The collected various data is preliminarily integrated, and time sequence shelf original monitoring data flow is generated in time sequence; the data flow contains comprehensive information of goods weight, image, electronic tag information, in-out shelf state and environment parameter of the shelf at different time, and provides rich data source for subsequent data processing and analysis; data cleaning algorithm is used to process the time sequence shelf original monitoring data flow, and remove noise, abnormal value and repeated data in the data; for example, for weight data collected by the pressure sensing device, sliding average filtering algorithm is used to remove random noise; for image data collected by the visual recognition module, image enhancement and denoising algorithm is used to improve image quality; for information read by the electronic tag reader, data checking and error correction processing is carried out, and accuracy of the data is ensured; data normalization method is used to convert data of different dimensions into a unified standard range; for example, goods weight data, temperature data, humidity data and the like are normalized according to certain formula, so that they have comparability in numerical value, and subsequent analysis and processing are facilitated; data denoising algorithm is used to further reduce noise interference in the data; for time sequence data, such as change data of goods weight with time, Kalman filtering algorithm is used for denoising processing, and smoothness and accuracy of the data are improved; for image data, wavelet transform and the like are used for denoising, and important feature information of the image is retained; after the preprocessing, preprocessed shelf monitoring data is generated, and high-quality data basis is provided for subsequent distributed storage and analysis.

[0048] Effects of the above technical solution are as follows: through overall planning of installation positions of the sensing devices, and setting reasonable working parameters according to performance characteristics of the devices, accurate collection of various sensing devices can be ensured, and information such as weight, image, in-out state and environment parameter of goods on the shelf is acquired in real time, and reliable data support is provided;

[0049] Through adoption of the wired and wireless combined communication mode, connection modes can be flexibly selected according to characteristics of different devices, wiring complexity is reduced, and expandability of the system is improved; for devices which are distributed relatively dispersedly and have strong mobility, wireless connection mode such as Wi-Fi or Bluetooth is adopted, and system expansion in the future is facilitated; for devices which have large transmission volume and short distance, wired connection mode is adopted, data transmission is ensured to be stable and high-speed, interference in wireless transmission is effectively avoided, and reliability and high efficiency of the system are enhanced;

[0050] Through enhancement and denoising processing of the collected image data, key features of the image are retained by using wavelet transform and the like, and definition and effectiveness of the image data are improved, high-quality information is provided for subsequent data analysis and decision; through the data cleaning algorithm, noise and abnormal value in the monitoring data are removed, and accuracy and consistency of the data are ensured; in particular, sliding average filtering and Kalman filtering and the like are used, and smoothness and accuracy of the time sequence data are effectively improved, and reliability of the data is enhanced;

[0051] Through intelligent data cleaning and noise reduction processing, the complexity of manual intervention and traditional data processing process is reduced, the system operation cost is reduced, and the data processing efficiency is improved; by reasonably arranging the environmental parameter monitoring sensor, the temperature, humidity, illumination intensity and other environmental parameters around the shelf are monitored in real time, the shelf management strategy can be adjusted in time to ensure that the goods are stored in a suitable environment and improve the quality management level of the goods;

[0052] Through the optimized perception signal fusion network topology and efficient data storage mechanism, the timely storage and rapid access of a large amount of real-time data can be ensured, and the efficiency and availability of the data storage system are improved; by monitoring and analyzing the data of various sensing devices in real time, combined with data processing algorithms, automatic replenishment, allocation, display and other decisions can be realized, the intelligent level of the system is enhanced, and the automation and intelligence level of the shelf management is improved; through automatic data acquisition and processing, manual operation is reduced, operation simplicity and accuracy are improved, real-time and accurate data support is provided, and user trust and satisfaction with the system are improved.

[0053] An embodiment of the present application, the S15, comprises:

[0054] Integrate various data according to the time stamp; form a complete data record unit, and perform format conversion and unification for the data format difference of different sensing devices;

[0055] According to the preliminary integrated data record unit, arrange it in time sequence to generate a time sequence shelf original monitoring data stream; store the generated time sequence shelf original monitoring data stream into a special time sequence database;

[0056] Use a data cleaning algorithm to remove noise data in the time sequence shelf original monitoring data stream; use an outlier detection technology of a machine learning algorithm to identify outliers in the time sequence shelf original monitoring data stream; use a data normalization method to convert data of different dimensions to a unified standard range; extract valuable feature information from the normalized data;

[0057] Analyze the internal relationship between different types of data, establish a data correlation model; use a multi-source data fusion technology to fuse the data collected by different sensing devices;

[0058] After the above processing steps, preprocessed shelf monitoring data is generated; and the preprocessed shelf monitoring data is stored in a data warehouse or a big data platform.

[0059] The working principle of the above technical solution is that the weight data of goods collected by the pressure sensing device, the image data of goods shot by the visual recognition module, the detailed attribute information of goods read by the electronic tag reader, the in-out shelf state data of goods monitored by the infrared sensor, and the environmental data such as temperature, humidity, and illumination intensity collected by the environmental parameter monitoring sensor are preliminarily integrated according to the time stamp; ensure that different types of data collected at the same time point can be associated together to form a complete data record unit to reflect the comprehensive state of the shelf at that time; in view of the data format difference of different sensing devices, format conversion and unification are carried out; for example, the image data collected by the visual recognition module is converted into a general image format (such as JPEG or PNG), while the key feature information (such as the shape and color distribution of goods) of the image is extracted and converted into a structured data format (such as JSON or XML) for subsequent data processing and analysis; for the text information read by the electronic tag reader, encoding conversion and standardization processing are carried out to ensure the accuracy and consistency of the data;

[0060] According to the preliminarily integrated data record unit, it is arranged in time sequence to generate a time sequence shelf original monitoring data stream; the data stream can clearly show the change of the shelf state with time, providing a basis for subsequent trend analysis and anomaly detection; the generated time sequence shelf original monitoring data stream is stored in a special time sequence database, such as InfluxDB or TimescaleDB; these databases have high efficiency in storing and querying time sequence data, and can meet the real-time storage and fast retrieval requirements of large-scale shelf monitoring data;

[0061] The data cleaning algorithm is used to remove noise data in the original monitoring data stream of the time series shelf. For example, for the weight data collected by the pressure sensing device, noise may be generated due to external interference (such as slight vibration of the shelf). The moving average filtering algorithm can be used to average the data within a certain time window, smooth the noise signal, and improve the accuracy of the data. The anomaly value detection technology of the machine learning algorithm is used to identify the anomaly value in the original monitoring data stream of the time series shelf. For the detected anomaly value, it is processed according to the specific situation. If the anomaly value is caused by equipment failure or data collection error, it can be replaced by normal data at the adjacent time point or repaired by interpolation method. If the anomaly value reflects the real abnormal state of the shelf (such as sudden large loss of goods or sharp change of environmental parameters), it is marked as an abnormal event and the relevant detailed information is recorded for subsequent analysis and processing. It is checked whether there is repeated data record in the original monitoring data stream of the time series shelf, which may be caused by repeated transmission or storage error in the data collection process. For the repeated data, it is deleted to reduce data redundancy and improve the efficiency of data processing. The data normalization method is used to convert data of different dimensions to a unified standard range. For example, for temperature data (the range may be 0-50℃) and weight data (the range may be 0-100kg), the minimum-maximum normalization method is used to normalize them to the interval [0, 1], so that different types of data have comparability. Valuable feature information is extracted from the normalized data to reflect the key state and change trend of the shelf. For example, the number of goods, position distribution, and arrangement neatness of goods are extracted from the image data of the visual recognition module. The weight change rate and weight fluctuation amplitude of goods are extracted from the weight data of the pressure sensing device. The temperature change trend and humidity change frequency are extracted from the data of the environmental parameter monitoring sensor. These features will be important inputs for subsequent data analysis and modeling.

[0062] Analyzing the inherent relationships between different types of data allows for the establishment of data association models. For example, there is a direct correlation between the status of goods entering and leaving the shelves (monitored by infrared sensors) and changes in the quantity of goods (monitored by visual recognition modules or pressure sensors). Changes in environmental parameters (such as temperature and humidity) may affect the quality and weight of goods (monitored by pressure sensors). By establishing these association models, a more comprehensive understanding of the status of the shelves and the reasons for changes can be achieved. Multi-source data fusion technology is employed to integrate data collected by different sensing devices, improving data accuracy and reliability. For instance, combining data from visual recognition modules and pressure sensors allows for a more accurate determination of the actual quantity and status of goods on the shelves. When the visual recognition module cannot accurately identify the quantity of goods due to obstruction, the weight data collected by the pressure sensor can be used as supplementary information for correction. Conversely, when the pressure sensor malfunctions or its data is inaccurate, the data from the visual recognition module can provide a reference. Through data fusion, the advantages of different sensing devices can be fully utilized, their shortcomings compensated for, and more comprehensive and accurate shelf monitoring data generated.

[0063] After the aforementioned data cleaning, normalization, feature extraction, correlation analysis, and fusion processes, preprocessed shelf monitoring data is generated. This data is characterized by high quality, high availability, and high consistency, accurately reflecting the real-time status and trends of the shelves. The preprocessed shelf monitoring data is stored in a data warehouse or big data platform for subsequent querying, analysis, and mining. Simultaneously, a comprehensive data management mechanism is established, including data backup, recovery, and secure access control, to ensure data security and integrity. Furthermore, the preprocessed shelf monitoring data can be further classified and labeled according to different business needs and analytical objectives, facilitating subsequent data retrieval and use.

[0064] The above technical solution achieves the following effects: by performing format conversion, encoding standardization, and outlier detection on data collected from different sensing devices, it ensures consistency in data time synchronization and format, avoids data deviation and error, and improves data quality; by using a moving average filtering algorithm to smooth noisy data, it reduces the impact of external interference on the data and improves the signal-to-noise ratio of the monitoring data.

[0065] Data cleaning, deduplication, and normalization processes reduce data redundancy and format inconsistencies, making subsequent data processing and analysis more efficient and optimizing storage space. By storing time-series data in a dedicated time-series database and combining it with multi-source data fusion technology, the system can store, retrieve, and accurately analyze large-scale shelf monitoring data in real time, thereby enhancing the system's real-time monitoring capabilities.

[0066] By establishing a data association model, we can gain a more comprehensive understanding of the relationships between different types of data, improve the analytical capabilities of shelf monitoring data, and help reveal the underlying reasons for changes in shelf status. By using machine learning algorithms for outlier detection and combining data fusion and complementarity from different sensing devices, we can effectively identify and respond to abnormal situations, improve the intelligent processing capabilities of the monitoring system, and reduce manual intervention.

[0067] The automated data processing workflow reduces the need for manual inspection and adjustment, effectively lowering maintenance costs. By correlating and analyzing data from environmental parameter monitoring sensors with the weight and shelf status data of goods, the system's adaptability to environmental changes is enhanced, and the accuracy of goods status assessment is improved.

[0068] By storing the processed data in a data warehouse or big data platform, subsequent queries, analysis, and mining are supported, providing support for further intelligent decision-making. At the same time, a sound data management mechanism has been established to ensure the security, integrity, and traceability of the data. Based on high-quality data preprocessing and analysis, merchants can be provided with more accurate inventory management, shelf layout optimization, and environmental adjustment strategies, thereby improving operational efficiency and customer satisfaction.

[0069] One embodiment of the present invention, such as Figure 2 As shown, S2 includes:

[0070] S21. Based on the multi-type sensing signal fusion network topology architecture data, collect relevant information from each storage node, and organize and analyze it.

[0071] S22. A distributed data storage node optimization algorithm is adopted to dynamically deploy and allocate resources for storage nodes; a distributed data storage architecture is constructed, dividing storage nodes into different levels and regions for hierarchical data storage and distributed management.

[0072] S23. Using the constructed distributed data storage architecture, a multi-path redundant transmission strategy is adopted to transmit the pre-processed shelf monitoring data to multiple storage nodes simultaneously.

[0073] S24. During transmission, the data is encrypted, redundant monitoring and uploading data packets are generated, and the transmission process is monitored and logged in real time to promptly detect and handle transmission anomalies.

[0074] The working principle of the above technical solution is as follows: based on the multi-type sensing signal fusion network topology architecture data, collect relevant information of each storage node, including storage capacity, processing power, network bandwidth, geographical location distribution and data access frequency, etc.; and organize and analyze it.

[0075] A distributed data storage node optimization algorithm is adopted to comprehensively consider the above factors for dynamic layout and resource allocation of the storage nodes; for example, for a region with high data access frequency, such as a storage node close to a shelf management terminal, more storage resources and processing capacity are allocated to improve the data access speed; for geographically dispersed storage nodes, the network topology structure is optimized to reduce data transmission delay and improve the overall performance of data storage; a distributed data storage architecture is constructed to divide the storage nodes into different levels and regions for hierarchical storage and distributed management of data; for example, hot data (recently frequently accessed data) is stored in a cache layer and cold data (long-term unaccessed data) is stored in a low-speed storage layer to improve the utilization of storage resources and the access efficiency of data;

[0076] Using the constructed distributed data storage architecture, a multi-path redundant transmission strategy is adopted to simultaneously transmit the preprocessed shelf monitoring data to multiple storage nodes; by setting multiple transmission paths in the network topology, such as different network links or redundant connections between storage nodes, it is ensured that the data can be reliably transmitted to the target nodes;

[0077] During transmission, the data is encrypted using a combination of symmetric encryption and asymmetric encryption to ensure data security; for example, the data is encrypted using the AES algorithm to ensure data confidentiality during transmission; the RSA algorithm is used to encrypt and transmit the encryption key to prevent key leakage; a redundant monitoring upload data packet is generated, each data packet containing original data, encryption information, checksum, and other elements; through redundant transmission, even if some storage nodes or transmission links fail, the integrity and reliability of the data can be ensured to avoid data loss; and the transmission process is monitored and logged in real time to timely detect and handle transmission anomalies.

[0078] The effects of the above technical solutions are: through the distributed data storage architecture and dynamic resource allocation algorithm, the layout and resource allocation of the storage nodes are optimized to ensure efficient data storage and fast access; by optimizing the network topology structure and setting multiple path redundant transmission, the data transmission delay is reduced to ensure fast and reliable data transmission between different nodes;

[0079] The combination of symmetric encryption and asymmetric encryption ensures the confidentiality and integrity of the data during transmission, preventing data leakage and tampering; through redundant transmission and monitoring mechanisms, even if some storage nodes or transmission links fail, the integrity and reliability of the data can be ensured to enhance the stability of the system;

[0080] Through multi-path redundant transmission and hierarchical storage strategy, the occupation of network bandwidth is reduced, and the excessive consumption of network bandwidth by high-frequency data access is avoided; through reasonable hierarchical storage and resource optimization, the efficiency of data management is improved, and the maintenance and updating process of the system is simplified;

[0081] A flexible distributed architecture is adopted, so that the system can dynamically adjust the storage nodes and resource configuration according to the needs, and the expansibility and adaptability of the system are enhanced; by storing data according to access frequency and usage, the excessive dependence on high-speed storage resources is reduced, the utilization of storage resources is optimized, and the overall storage cost of the system is reduced;

[0082] By using multi-type sensing signal fusion and data analysis, the system can automatically optimize resource allocation, transmission path and storage layout, improve the overall intelligentization and automation level; by providing high-quality, reliable monitoring data, supporting real-time decision-making and shelf management optimization of merchants, the business efficiency and customer satisfaction are improved.

[0083] An embodiment of the present application, the S22, comprises:

[0084] According to the collected data access frequency information of each storage node, the number of times that each storage node is accessed in different time periods is counted based on a time axis; according to the statistical result of the access frequency, the storage nodes are divided into different heat regions; a threshold value of access frequency is set, the region with access frequency higher than the threshold value is defined as a high-heat region, the region with access frequency lower than the threshold value but with a certain amount of access is defined as a medium-heat region; the region with extremely low access frequency is defined as a low-heat region;

[0085] For the high-heat region, more storage resources and processing capacity are preferentially allocated; for the medium-heat region, storage resources and processing capacity are reasonably allocated according to its access frequency and business demand; for the low-heat region, basic storage resources are allocated;

[0086] The network delay and bandwidth between different geographical locations are analyzed; according to the geographical location analysis and network performance test results, the network topology structure is optimized; the pre-processed shelf monitoring data is evaluated for heat; the data is divided into three categories: hot data, warm data and cold data;

[0087] According to the heat evaluation result of the data, a hierarchical storage architecture is designed; the hot data is stored in the cache layer, the warm data is stored in the intermediate layer, and the cold data is stored in the low-speed storage layer;

[0088] A real-time monitoring system is established to monitor the resource usage, data access and network performance of the storage nodes in real time; according to the feedback information of real-time monitoring, a dynamic resource adjustment algorithm is used to dynamically adjust the resource allocation of the storage nodes.

[0089] The working principle of the above technical solution is: according to the collected data access frequency information of each storage node, the number of times each storage node is accessed in different time periods is counted based on the time axis; for example, taking each hour as a statistical period, the number of data read and write requests received by each storage node in the period is recorded; according to the statistical result of the access frequency, the storage nodes are divided into different heat regions; set the threshold of access frequency, define the region with access frequency higher than the threshold as the high heat region, such as the region where the storage nodes close to the shelf management terminal and frequently queried by users are located; the region with access frequency lower than the threshold but with certain access volume is defined as the medium heat region; the region with extremely low access frequency is defined as the low heat region; through the division of heat regions, the access distribution of data can be clearly understood, which provides a basis for subsequent resource allocation;

[0090] For the high heat region, more storage resources and processing capacity are preferentially allocated; in terms of storage resources, the disk capacity of the storage nodes in this region is increased or higher performance storage devices such as solid state disks (SSD) are used to meet the storage demand of a large amount of data; in terms of processing capacity, more powerful processors and more memory are provided for the storage nodes in this region to improve data processing speed and ensure that the data access requests of users can be quickly responded; for example, part of the storage resources and processing capacity originally allocated to the low heat region are allocated to the high heat region, so that the storage nodes in the high heat region can better cope with high concurrency data access; for the medium heat region, storage resources and processing capacity are allocated reasonably according to its access frequency and business demand; avoid over allocation or insufficient allocation of resources to ensure effective use of resources; resource reservation can be used to reserve certain expansion space for the medium heat region to cope with possible increase in access volume in the future; for the low heat region, basic storage resources are allocated to meet the storage demand of a small amount of data, and low-power storage devices can be used to reduce energy consumption and operating costs;

[0091] Consider the geographical distribution of storage nodes, analyze the network delay and bandwidth between different geographical locations; for example, through network performance testing tools, measure the network delay and bandwidth between different storage nodes, draw a network topology map, and intuitively show the network connection status between storage nodes; according to the results of geographical location analysis and network performance testing, optimize the network topology structure; for storage nodes with dispersed geographical locations, adopt multi-link redundant connection mode to increase the path selection of data transmission and reduce the impact of single point failure on data transmission; for example, establish multiple network links between two storage nodes with far apart geographical locations, when one of the links fails, data can be automatically switched to other links for transmission; at the same time, set the routing strategy between network nodes reasonably, select the optimal transmission path, reduce the data transmission delay, and improve the overall performance of data storage; combine the heat analysis of data access and the update frequency, importance and other factors of data to evaluate the heat of preprocessed shelf monitoring data; divide the data into three categories: hot data, warm data and cold data; hot data refers to data that is frequently accessed recently and has high real-time requirements, such as the in-out warehouse records of goods on the shelf in the past week; warm data refers to data with moderate access frequency and certain real-time requirements, such as inventory statistics of goods in the past month; cold data refers to long-term non-access data with strong historical characteristics, such as shelf environment monitoring data more than a year ago;

[0092] According to the heat evaluation results of data, design a layered storage architecture; store hot data in the cache layer, use memory database or high-speed solid state disk as storage medium to provide extremely fast data access speed; store warm data in the middle layer, use a disk array with moderate performance as storage medium to meet certain data access requirements; store cold data in the low-speed storage layer, use large-capacity tape library or low-cost hard disk as storage medium to reduce storage cost; through the layered storage architecture, storage resources can be reasonably allocated according to the characteristics and access requirements of data, and the utilization rate of storage resources and the access efficiency of data can be improved;

[0093] A real-time monitoring system is established to monitor the resource usage, data access, and network performance of storage nodes in real time. It collects metrics such as CPU utilization, memory utilization, and disk I / O speed, as well as data access latency and throughput. Based on the feedback from real-time monitoring, a dynamic resource adjustment algorithm is used to dynamically adjust the resource allocation of storage nodes. When the resource utilization of a storage node is too high or the data access latency is too large, resources are automatically allocated from other storage nodes with idle resources to that node to improve its processing capacity. For example, when a storage node in a high-intensity area experiences a performance bottleneck, some processing capacity is allocated from low-intensity areas to that node to ensure that data access in the high-intensity area is not affected. Simultaneously, based on changes in data popularity, the storage layer of data is dynamically adjusted, migrating hot data to the high-speed cache layer and cold data to the low-speed storage layer in a timely manner, achieving optimized allocation of storage resources.

[0094] The above technical solution achieves the following results: by dividing storage nodes into hot zones based on data access frequency, the allocation of storage resources can be precisely adjusted to ensure that high-hot zones receive sufficient storage capacity and processing power, thereby improving overall storage efficiency; by optimizing the network topology of storage nodes, adding redundant links, and selecting the optimal transmission path, data transmission latency is significantly reduced, thereby improving data transmission efficiency.

[0095] By introducing a tiered storage architecture and resource reservation strategy, the system can flexibly respond to the growth of data volume and changes in access requirements, improving its adaptability when expanding in the future. Based on the data popularity assessment results, storage resources are rationally allocated, with hot data using high-speed caching and cold data using low-speed storage, maximizing the utilization of storage resources and avoiding resource waste.

[0096] By storing cold data in low-cost hard drives or tape libraries, and using low-power devices to store low-temperature data, energy consumption and equipment investment costs are reduced, and overall operating costs are optimized. By configuring higher-performance storage devices and more powerful processors for high-temperature areas, rapid response to data access requests is ensured, and the overall performance of the system is improved.

[0097] By using multi-link redundancy connections and routing strategy optimization, even if a link fails, data can still be transmitted through other paths, reducing the impact of single-point failures and enhancing the stability and reliability of the system. By dynamically adjusting resource allocation, storage nodes in high-traffic areas can continuously obtain sufficient processing power, effectively avoiding access latency fluctuations caused by insufficient resources.

[0098] Through real-time monitoring and dynamic resource adjustment algorithm, the system can automatically optimize resource allocation and data storage level according to real-time feedback, improve the intelligent and automation degree of the system, and reduce manual intervention; through hierarchical storage architecture and thermal region division, different types of data can be clearly distinguished, facilitating data management and maintenance, and improving data management efficiency and operation simplicity.

[0099] In one embodiment of the application, the S23 comprises:

[0100] Comprehensively scan the network topology involved in the distributed data storage architecture, identify all available network links and connection relationships between storage nodes;

[0101] Based on the network topology information obtained by scanning, use a multi-path generation algorithm to generate multiple potential transmission paths from the data source to each target storage node;

[0102] Comprehensively evaluate each generated potential transmission path, use a weighted scoring method to give different weights according to different indicators of the path, and calculate the comprehensive score of each path; select the highest score path as the actual data transmission path;

[0103] According to the characteristics and size of the preprocessed shelf monitoring data, use intelligent data sharding strategy to divide the data into multiple moderately sized data pieces; generate a certain number of redundant data pieces for each data piece, and distribute the generated data pieces and redundant data pieces to different storage nodes on the multiple transmission paths selected previously;

[0104] Through a dynamic transmission scheduling algorithm, dynamically adjust the transmission order and transmission rate of data pieces on different transmission paths according to the real-time monitored network link state and storage node load condition;

[0105] Establish a coordination mechanism between paths, so that data transmission on different transmission paths can cooperate with each other; when the data transmission on a certain path is delayed or fails, the data transmission on other paths can automatically speed up or take on more transmission tasks; during data transmission, the data transmission on each transmission path is monitored in real time, and the monitored information is transmitted to the transmission scheduling algorithm in real time through a real-time feedback mechanism;

[0106] When all data pieces and redundant data pieces are successfully transmitted to the target storage nodes, each storage node sends a transmission completion confirmation message to the data source; after receiving the confirmation messages from all storage nodes, the data source confirms that the data transmission task is completed; at the same time, a timeout retransmission mechanism is used, if the confirmation message from a certain storage node is not received within a specified time, the data source will resend the data piece that the storage node has not received;

[0107] After the data transmission is completed, a consistency check is performed on the data stored on each storage node; a hash algorithm is used to calculate the check value for each data slice and redundant data slice, and the check value is compared with the check value pre-calculated by the data source;

[0108] If the checksums are found to be inconsistent, it indicates that the data may have been damaged or lost during transmission. The data recovery mechanism is then activated to recover the original data using redundant data fragments, and the transmission and verification are repeated until the data on all storage nodes are consistent.

[0109] The working principle of the above technical solution is as follows: a comprehensive scan of the network topology involved in the distributed data storage architecture is performed to identify all available network links and the connection relationships between storage nodes; a detailed network topology map is obtained using professional network topology discovery tools, in which the location of each storage node, the bandwidth of the network link, latency, and reliability and other key parameters should be clearly marked.

[0110] Based on the network topology information obtained from the scan, a multi-path generation algorithm is used to generate multiple potential transmission paths from the data source to each target storage node. These paths not only need to consider the shortest distance, but also need to take into account factors such as network link load, bandwidth utilization, and historical transmission success rate to ensure that the generated paths are diverse and reliable.

[0111] Each potential transmission path generated is comprehensively evaluated using a weighted scoring method. Different weights are assigned to different indicators such as bandwidth, latency, and reliability of the path to calculate the comprehensive score of each path. The path with the highest score is selected as the actual data transmission path to ensure that network resources can be fully utilized during transmission and to improve the efficiency and reliability of data transmission.

[0112] Based on the characteristics and size of the pre-processed shelf monitoring data, an intelligent data fragmentation strategy is adopted to divide the data into multiple appropriately sized data fragments. Considering the correlation and access patterns of the data, data with strong correlations are grouped into the same or adjacent data fragments as much as possible to improve the efficiency of subsequent data processing. For example, for shelf monitoring data generated according to time series, fragmentation can be performed according to a certain time interval to ensure that each data fragment contains complete monitoring information for a period of time.

[0113] For each data slice, generate a certain number of redundant data slices, and rationally distribute the generated data slices and redundant data slices to different storage nodes on the multiple transmission paths previously selected; ensure that the data slices and redundant data slices stored on each storage node have a certain degree of dispersion to avoid large amounts of data loss due to the failure of a single storage node; at the same time, consider the load balancing of storage nodes to make the data transmission and storage tasks undertaken by each storage node relatively balanced.

[0114] Through the dynamic transmission scheduling algorithm, the transmission order and transmission rate of data pieces on different transmission paths are dynamically adjusted according to the real-time monitored network link state and the load condition of the storage node; for example, when the network bandwidth of a certain transmission path suddenly decreases or congestion occurs, the scheduling algorithm can automatically transfer part of the data pieces to other paths with sufficient bandwidth for transmission, ensuring that the data can arrive at the target storage node on time and accurately;

[0115] A coordination mechanism between paths is established to enable data transmission on different transmission paths to cooperate with each other; when the data transmission on a certain path is delayed or fails, the data transmission on other paths can automatically speed up or take on more transmission tasks to ensure that the progress of the overall data transmission is not affected; for example, a distributed lock mechanism or a message queue mechanism is used to realize the coordination and synchronization of transmission tasks on different paths; during data transmission, the data transmission on each transmission path is monitored in real time, including transmission rate, transmission delay, data packet loss rate and other indicators; through a real-time feedback mechanism, the monitored information is timely transmitted to the transmission scheduling algorithm, so that the algorithm can be dynamically adjusted and optimized according to the actual situation, ensuring the stability and reliability of data transmission;

[0116] When all data pieces and redundant data pieces are successfully transmitted to the target storage node, each storage node sends a transmission completion confirmation message to the data source; after receiving the confirmation messages from all storage nodes, the data source confirms that the data transmission task is completed; at the same time, a timeout retransmission mechanism is used, if the confirmation message from a certain storage node is not received within a specified time, the data source will resend the data pieces that have not been received by the storage node, ensuring the complete transmission of data;

[0117] After data transmission is completed, consistency check is performed on the data stored in each storage node; a hash algorithm is used to calculate the check value of each data piece and redundant data piece, and the check value is compared with the check value calculated by the data source in advance;

[0118] If the check values are found to be inconsistent, it indicates that the data may have been damaged or lost during transmission, the data recovery mechanism is started, the original data is recovered using the redundant data pieces, and transmission and checking are performed again until the data on all storage nodes are consistent.

[0119] The effect of the above technical scheme is that: by using the multi-path generation algorithm and the intelligent data fragmentation strategy, not only the transmission delay is reduced, but also the fast and accurate transmission of data is ensured through the dynamic transmission scheduling algorithm and the inter-path coordination mechanism, and the efficiency of data transmission is greatly improved; by generating and reasonably distributing the redundant data pieces, data loss caused by storage node failure is avoided, and the complete transmission of data is ensured through the timeout retransmission mechanism during data transmission;

[0120] By using multiple transmission paths, redundant data pieces and dynamic scheduling algorithm, even if part of the path or storage node fails, data transmission can still proceed smoothly, enhancing the reliability and fault tolerance of the system when facing failures; by optimizing the network topology, selecting the optimal path and balancing the network load, the bandwidth of each path is fully utilized, and the overall utilization of network resources is improved;

[0121] When a certain path is congested or the bandwidth is reduced, the dynamic transmission scheduling algorithm can automatically adjust the transmission path of the data piece, avoiding the influence of network congestion and ensuring the stable transmission of data; by monitoring the rate, delay and packet loss rate and other indicators on the data transmission path in real time, and feeding back the monitoring data to the scheduling algorithm in time, the data transmission process is monitored and dynamically optimized in all directions, and the stability of data transmission is improved;

[0122] The hash algorithm is used for data consistency check, ensuring that the data on all storage nodes is not damaged or lost during transmission, improving the consistency and integrity of the data; by introducing intelligent data fragmentation, redundant storage and automatic recovery mechanism, the need for manual intervention is reduced, the maintenance cost of the system is reduced, and the self-repairing ability of the system is improved;

[0123] By introducing the multi-path generation algorithm, dynamic scheduling and path coordination mechanism, the system can optimize itself according to real-time conditions, improving the intelligent level of the data transmission process; by reasonably allocating data pieces and redundant data pieces, it is ensured that the storage and transmission tasks borne by each storage node are balanced, avoiding overload or resource waste of some nodes, and improving the efficiency of the overall storage system.

[0124] An embodiment of the present application, the S3, comprises:

[0125] S31, a unique upload credential information is generated for each redundant monitoring upload data packet, and the upload credential information is stored in association with the redundant monitoring upload data packet;

[0126] S32, based on the upload credential information, the redundant monitoring upload data received by different storage nodes is analyzed; the data part with difference is found out; the data with difference is recorded and analyzed in detail, and a monitoring upload difference data report is generated;

[0127] S33, adopt data consistency negotiation algorithm and intelligent voting decision mechanism, according to the monitoring upload difference data report, the redundant monitoring upload data is carried out difference conflict resolution;In the negotiation process, data interaction and information sharing between each storage node, send their own data and certificate information to other nodes for comparison and verification;

[0128] S34, generate accurate shelf unit monitoring data, update the data in each storage node, and record and audit the difference conflict resolution process.

[0129] The working principle of the above technical solution is: a unique upload certificate information is generated for each redundant monitoring upload data packet, which contains data source identification, upload timestamp, data checksum, digital signature and encryption key and other elements;Data source identification is used to identify the data generation device and location, which is convenient for tracing the source of data;Upload timestamp records the upload time of data, ensuring the timeliness of data;Data checksum is used to verify the integrity and accuracy of data, which is calculated by a specific algorithm;Comparison is made at the receiving end;Digital signature uses asymmetric encryption algorithm to sign data, ensuring data authenticity and non-repudiation;Encryption key is used to encrypt and decrypt data, ensuring data confidentiality;And the upload certificate information is stored in association with the redundant monitoring upload data packet, so that the certificate information can be quickly and accurately obtained in subsequent data processing and analysis, verifying the legality and integrity of the data;

[0130] Based on the upload certificate information, the redundant monitoring upload data received by different storage nodes is analyzed for differences;By comparing the checksum, timestamp and specific content of the data, the data part with differences is found out;For example, if the checksum of the same data packet received by two storage nodes is inconsistent, it means that the data may have been wrong or lost during transmission;If the timestamp is significantly different, there may be a problem of data transmission delay or data update out of sync;The data with differences is recorded and analyzed in detail, and a monitoring upload difference data report is generated;The report contains information such as the source of the difference data, the difference type, the difference degree and the possible reasons, which provides a basis for subsequent difference conflict resolution;

[0131] The data consistency negotiation algorithm and the intelligent voting decision mechanism are adopted, and the redundant monitoring upload data is differentiated and conflict is eliminated according to the monitoring upload difference data report. In the negotiation process, data interaction and information sharing are carried out between the storage nodes, and the data and credential information of each storage node are sent to other nodes for comparison and verification. The final correct data version is determined by comprehensively considering the data source reliability, data update time and other factors. For example, if the data of a storage node is reliable and the update time is recent, the data of this node is preferred. If the data of multiple nodes is consistent and reliable, the data of these nodes is used as the reference.

[0132] The accurate shelf unit monitoring data is generated, the data in each storage node is updated, and the consistency and accuracy of the data stored in each storage node are ensured. The difference conflict elimination process is recorded and audited for subsequent tracing and analysis.

[0133] The effects of the above technical solutions are: by generating a unique upload credential information for each redundant monitoring upload data packet, the source, timeliness, integrity and authenticity of the data are effectively verified, thereby improving the security of the data upload process; through the redundant data storage and difference analysis mechanism, errors or losses occurring during data transmission or storage can be discovered and repaired in time, enhancing the reliability of data transmission;

[0134] Through the data consistency negotiation algorithm and the intelligent voting decision mechanism, the conflicts between redundant data can be effectively eliminated, ensuring that the final data version is correct and reducing problems caused by data conflicts; through hash verification, digital signature and other technologies, the legality and integrity of the data can be quickly verified, so that in the process of large data transmission, abnormal situations can be efficiently identified and handled;

[0135] The redundant data storage and difference analysis mechanism ensures that even if some storage nodes fail, the system can still ensure the consistency and accuracy of the data, thereby enhancing the fault tolerance and recovery ability of the system; by continuously monitoring the checksum and timestamp information of the upload data, problems in data transmission can be detected in real time, and transmission strategies can be adjusted in time to ensure the stability and reliability of the data;

[0136] By monitoring and recording the timestamp of data upload, it is ensured that the data of each storage node can be updated synchronously, reducing the risk of asynchronous data update. The intelligent voting decision mechanism and dynamic decision algorithm are adopted, and the system can automatically optimize according to real-time conditions, enhancing the intelligence and flexibility of data storage management;

[0137] Through automatic data difference analysis, conflict resolution and audit records, the need for manual intervention is reduced, and the maintainability and management efficiency of the system are improved; the introduction of intelligent data sharding, redundant storage and automatic recovery mechanism not only improves the efficiency of data transmission and storage, but also greatly reduces the operation and maintenance cost of the system.

[0138] In one embodiment of the present application, the S4 comprises:

[0139] S41, according to the characteristics and management requirements of the shelf unit monitoring data, the data is divided according to multiple dimensions by using dynamic data sharding strategy; monitoring sharding characteristic data is generated, and the characteristic information of each data shard is recorded;

[0140] S42, based on the monitoring sharding characteristic data, the commonly used data shards are cached in the cache device by using intelligent cache synchronization algorithm; the intelligent cache synchronization algorithm dynamically adjusts the cache strategy according to the access mode and update frequency of the data;

[0141] S43, according to the update of the data, the cache data is synchronized in real time, when the data in the underlying storage is updated, the cache system is notified to update the corresponding cache data; when the data in the cache is modified, the data is written back to the underlying storage according to the set synchronization strategy;

[0142] S44, cache management log data is generated, and the information of the cache data is recorded; through the analysis of the cache management log data, the running condition of the cache system is obtained.

[0143] The working principle of the above technical solution is: according to the characteristics and management requirements of the shelf unit monitoring data, the data is divided according to multiple dimensions such as product category, time period, access frequency and data importance by using dynamic data sharding strategy; for example, for the data of hot-selling goods, fine-grained sharding is performed according to product category and time period, so as to quickly query and analyze; for cold data which is not frequently accessed, coarse-grained sharding is performed according to a larger time period, so as to reduce the storage space occupation; the dynamic data sharding strategy can be dynamically adjusted according to the real-time changes of the data and the management requirements; for example, when the sales volume of a certain product suddenly increases, the sharding granularity of the data of the product is automatically adjusted to improve the access efficiency of the data; when the display layout of the shelf changes, the sharding mode of the data related to the display is adjusted accordingly to ensure the consistency of the data and the actual business; monitoring sharding characteristic data is generated, and the characteristic information of each data shard is recorded, such as shard identifier, shard dimension, shard size, data range, etc., which provides a basis for subsequent intelligent cache synchronization and management;

[0144] Based on the monitoring of the characteristics of the data slices, the commonly used data slices are cached in the cache device using an intelligent cache synchronization algorithm. The intelligent cache synchronization algorithm dynamically adjusts the cache strategy according to the access mode and update frequency of the data. For example, for frequently accessed data slices, a prefetch strategy is used to cache them in the cache in advance, reducing data access delay. For data slices with high update frequency, a write-back strategy is used to update the data in the cache first, and then write it back to the underlying storage at an appropriate time, improving data write efficiency.

[0145] The cache data is synchronized in real time according to the data update, ensuring the consistency of the cache data and the underlying storage data. When the data in the underlying storage is updated, the cache system is notified to update the corresponding cache data. When the data in the cache is modified, the data is written back to the underlying storage according to the set synchronization strategy.

[0146] Cache management log data is generated to record the storage location, update time, access frequency, hit rate, and cache state of the cache data. Through analysis of the cache management log data, the running status of the cache system is obtained, and problems in the cache strategy are found, providing a basis for cache state evaluation and optimization.

[0147] The effect of the above technical solution is that through the dynamic data slicing strategy, the data is sliced according to different dimensions and optimized according to access frequency and importance, ensuring that hot-selling goods or frequently accessed data can be quickly queried and analyzed, thereby improving data access efficiency.

[0148] For cold data that is not frequently accessed, coarse-grained slicing reduces the storage space occupation, avoids waste of storage resources, and improves the utilization rate of storage space. Through the intelligent cache synchronization algorithm, frequently accessed data slices are cached in the cache device, and the cache strategy is dynamically adjusted according to the access mode and update frequency, thereby significantly improving the cache hit rate and reducing data access delay.

[0149] By synchronizing the cache data and the underlying storage data in real time, the data in the cache and the underlying storage always remains consistent, avoiding data inconsistency and improving data accuracy and reliability. Through the combination of the prefetch strategy and the write-back strategy, frequently accessed and updated data can be responded to in a timely manner, reducing data access delay and improving system response speed.

[0150] Dynamic adjustment of data slicing granularity and cache strategy enables the system to flexibly adjust according to real-time changes in business requirements and data characteristics, enhancing the scalability of the system. By recording cache management log data, the system can monitor the state, hit rate, update frequency, and other information of the cache in real time, thereby optimizing the cache strategy and improving cache management efficiency.

[0151] By means of the intelligent caching and dynamic data sharding strategy, the burden of underlying storage is effectively reduced, the use of storage resources is optimized, and the storage and management costs are reduced; cache management log data is generated and analyzed, so that the running status of the cache system can be identified and optimized in time, and the maintainability of the system is enhanced; through the intelligent caching synchronization and sharding strategy based on data characteristics, the intelligent level of the data processing process is improved, so that the system can be autonomously adjusted and optimized, and manual intervention is reduced.

[0152] In an embodiment of the present application, the S42 comprises:

[0153] By means of data mining technology, the access records of each data shard in the past period of time are deeply analyzed; the access association relationship between different data shards is found out, the access behavior of the current data shard is tracked in real time, and the key information of each data shard is recorded; by analyzing these information in real time, the sudden change of data access is found out;

[0154] Based on the results of historical access data mining and real-time access behavior monitoring, a data access pattern model is constructed; considering multiple characteristic dimensions of data shards, a multi-dimensional caching strategy is designed; through an adaptive mechanism, the parameters of the caching strategy are dynamically adjusted according to the change of the data access pattern and the system running state;

[0155] According to the prediction result of the data access pattern model, the data shards that may be accessed are prefetched into the cache in advance; by means of the association relationship of data access, when a data shard is accessed, other data shards associated with it are also prefetched;

[0156] For data shards with high access frequency and strong timeliness, a higher priority is given to try to keep them in the cache; for data shards with low access frequency and weak timeliness, they are preferentially replaced out of the cache; according to the running state of the system and the use of the cache, the threshold value of cache replacement is dynamically adjusted.

[0157] The working principle of the above technical solution is: through data mining technology, the access records of each data shard in the past period of time are deeply analyzed; the access association relationship between different data shards is found out, for example, the data shards of some goods are often accessed at the same time in a certain time period, which may imply that they are closely related in the business scenario; at the same time, using time series analysis method, the time law of data shard access is studied, such as the access volume of some goods data is obviously higher on weekends than on weekdays, or the access volume will rise sharply during a certain promotion activity; through real-time monitoring system, the access behavior of the current data shard is tracked in real time; record the access request source, access time, access frequency and other key information of each data shard; through real-time analysis of these information, the sudden change of data access is found, for example, a data shard with low access volume is suddenly accessed frequently in a short time, which may be caused by sudden changes in market demand or adjustment of business operation;

[0158] Based on the results of historical access data mining and real-time access behavior monitoring, a data access pattern model is constructed; considering multiple characteristic dimensions of data shards, such as access frequency, update frequency, data importance, etc., a multi-dimensional cache strategy is designed; for data shards with high access frequency and low update frequency, long-term cache strategy is adopted, which is kept in cache for a long time to reduce data access delay; for data shards with high access frequency but also high update frequency, short-term cache combined with prefetching strategy is adopted, which is kept in cache for a short time and pre-fetches the data that may be accessed according to the access pattern prediction; for data shards with high data importance but relatively low access frequency, on-demand cache strategy is adopted, which is quickly cached to cache when receiving access request; through adaptive mechanism, the parameters of cache strategy are dynamically adjusted according to the change of data access pattern and system running state; for example, when the system load is high, the capacity of cache is appropriately reduced, and the data shards with the highest access frequency are preferentially cached to ensure the overall performance of the system; when the data access pattern changes greatly, the prefetching strategy and replacement strategy of cache are automatically adjusted to make the cache adapt to the new access pattern faster; and different cache strategies are set with priority to ensure that in the case of limited resources, the cache strategy with the greatest impact on system performance is preferentially executed; for example, the cache strategy to ensure data consistency is set as the highest priority to ensure that the cache data is always consistent with the underlying storage data; the cache strategy to improve data access speed is set as a higher priority to meet the real-time requirements of business;

[0159] According to the result predicted by the data access mode model, the data fragments that are likely to be accessed are prefetched into the cache in advance; for the data fragments with a high predicted access probability, they are read from the underlying storage and cached in the cache before they are actually accessed, reducing the time for the user to wait for data loading; for example, if it is predicted that a certain commodity will be queried a lot in the next few hours, the related data fragments of the commodity are prefetched into the cache in advance; by using the correlation of data access, when a certain data fragment is accessed, other data fragments associated with it are also prefetched; for example, when a user queries the sales data of a certain commodity, in addition to prefetching the sales data fragment of the commodity, the inventory data fragment of the commodity, the supplier information fragment and other associated data can also be prefetched, improving the efficiency and continuity of data query; for data fragments with small data volume and frequent updates, an incremental prefetching method is used to prefetch only the changed part of the data, reducing network transmission and cache space occupation; for data fragments with large data volume and relatively less updates, a batch prefetching method is used to prefetch the entire data fragment into the cache at one time, improving the prefetching efficiency;

[0160] For data fragments with high access frequency and strong timeliness, a higher priority is given to try to keep them in the cache; for data fragments with low access frequency and weak timeliness, they are preferentially replaced out of the cache; for example, a variant of the LRU (Least Recently Used) algorithm is used, combined with the timeliness factor of the data fragments, to sort the data fragments in the cache, and when the cache space is insufficient, the data fragments at the back of the sorting are preferentially replaced out; in the cache replacement process, the importance of the data fragments is fully considered; for data fragments with high importance, even if their access frequency is low, they are also kept in the cache as much as possible to avoid data loss or affect the normal operation of the business due to replacement; different weights can be set for data fragments with different importance, and a comprehensive evaluation and decision is made according to the weights during replacement; according to the running state of the system and the usage of the cache, the threshold for cache replacement is dynamically adjusted; for example, when the system load is low and the cache space is sufficient, the replacement threshold is appropriately increased to reduce the replacement frequency of the data fragments and improve the cache hit rate; when the system load is high or the cache space is tight, the replacement threshold is reduced to release the cache space in time and ensure the performance and stability of the system.

[0161] The effect of the above technical solution is that by using the dynamic data fragment strategy and the intelligent cache algorithm, the query speed of hot-selling commodities and frequently accessed data is optimized, ensuring fast response to user requests and greatly improving the data access efficiency; by using the coarse-grained data fragment and cold data optimization strategy, the storage space occupation of infrequently accessed data is reduced, unnecessary storage resource waste is avoided, and the utilization rate of the storage space is improved;

[0162] By intelligent cache synchronization and dynamic adjustment of cache strategy, frequently accessed data is effectively cached to high-speed storage devices, significantly improving cache hit rate and reducing data access delay; by real-time synchronization of cache and underlying storage data, cache data and actual data are kept consistent, avoiding data inconsistency and improving data accuracy and reliability;

[0163] By combining prefetch and write-back strategies, frequently accessed and updated data can be responded in time, reducing the delay caused by waiting for data loading and improving the response speed of the system; dynamic adjustment of cache and data sharding strategies enables the system to make real-time adjustments according to business needs and data characteristics, increasing the flexibility and scalability of the system;

[0164] By recording detailed cache management log data, the system can monitor cache status, hit rate and update frequency in real time, and optimize cache strategy based on this information, improving cache management efficiency; through intelligent cache and sharding strategies, the pressure on the underlying storage system is reduced, storage resource usage is optimized, and overall data storage and management costs are reduced;

[0165] By generating cache management logs and analyzing them, real-time optimization of cache strategy is achieved, ensuring efficient operation and maintainability of the system and reducing the risk of system failure; based on data characteristics, intelligent cache and sharding strategies improve the automation and intelligence level of data processing, reduce human intervention, and enable the system to make autonomous adjustments based on real-time data flow.

[0166] One embodiment of the present application, the S5, comprises:

[0167] S51, according to the cache management log data, use the cache state evaluation model, comprehensively consider multiple indexes, comprehensively evaluate the cache state; through machine learning algorithm, the above-mentioned index is analyzed comprehensively, and the monitoring cache state data is generated;

[0168] S52, based on the monitoring cache state data, combined with the business goals and key performance indicators of the shelf management, use the comprehensive analysis algorithm to analyze the management state of the monitoring sharding feature data; generate a comprehensive data management index.

[0169] The working principle of the above technical solution is: according to the cache management log data, using a cache state evaluation model, and comprehensively considering multiple indicators including cache hit rate, cache utilization rate, cache update frequency, data access delay and cache error rate, the cache state is comprehensively evaluated; the cache hit rate reflects the satisfaction degree of the cache system to the data request, the higher the hit rate, the better the cache effect; the cache utilization rate represents the use of cache space, too high utilization rate may cause cache overflow, and too low utilization rate wastes storage resources; the cache update frequency reflects the timeliness of cache data, too high update frequency will increase system overhead, and too low update frequency may cause cache data to be out of date; data access delay is an important indicator to measure the performance of the cache system, the lower the delay, the faster the data access speed; the cache error rate reflects the probability of errors occurring in the running process of the cache system, too high error rate will affect the stability and reliability of the system; the above indicators are analyzed by machine learning algorithm to generate monitoring cache state data; the data can directly reflect the performance and health status of the cache system, and provide decision basis for the optimization and management of the cache; for example, if the cache hit rate is low, the cache strategy may need to be adjusted or the cache capacity may need to be increased; if the cache error rate is high, the hardware and software configuration of the cache system needs to be checked to find out the fault reason;

[0170] Based on the monitoring cache state data, combined with the business objectives and key performance indicators of shelf management, a comprehensive analysis algorithm is used to analyze the management state of the monitoring fragmented feature data; factors such as product inventory accuracy, display layout rationality, operation risk level, sales performance influence and customer satisfaction are comprehensively considered to evaluate the overall efficiency of shelf management; the product inventory accuracy reflects the consistency between the actual inventory of the shelf and the system recorded inventory, the higher the accuracy, the more accurate the inventory management; the display layout rationality evaluates whether the product display on the shelf meets the purchasing habits of consumers and market demand, and reasonable display layout can improve the sales volume of products; the operation risk level includes the occurrence probability and influence degree of risks such as product shortage, expiration and equipment failure, the lower the risk level, the more stable the shelf operation; the sales performance influence analyzes the influence of shelf management strategy on product sales volume, and evaluates the effectiveness of the management strategy by comparing the sales data under different time periods or different display layouts; the customer satisfaction is obtained through questionnaires, user feedback and other ways, reflecting the satisfaction of consumers on shelf display, product supply and other aspects; a comprehensive data management index is generated, which is a comprehensive index to measure the overall efficiency of shelf management, obtained by weighted calculation of the above factors; the comprehensive data management index can provide scientific and comprehensive basis for the decision of shelf management, help managers to find out the problems in shelf management in time, formulate targeted improvement measures, and improve the intelligent level and operation efficiency of shelf management.

[0171] The effect of the above technical solution is: through comprehensive evaluation of the cache state, including cache hit rate, cache utilization rate, cache update frequency, data access delay and cache error rate, the monitoring accuracy of the cache system is improved, so that the manager can timely discover potential performance bottlenecks or problems, and optimize the cache management strategy;

[0172] Considering the cache utilization rate, the situation of too high or too low cache space is avoided, thereby reducing the risk of cache overflow, reducing the waste of storage resources, and improving the utilization efficiency of cache space; by monitoring and analyzing the cache error rate, the hardware and software problems of the cache system are found and checked in time, the stability and reliability of the system are enhanced, and the health status of the cache system in long-term operation is ensured;

[0173] By evaluating the access delay of the cache system, taking corresponding measures to optimize the cache strategy, reducing the delay of data access, improving the response speed of user requests, and enhancing the real-time performance of the system; combined with the accuracy of goods inventory, ensuring that the actual inventory of the shelf is consistent with the system record, reducing the error in inventory management, and improving the accuracy of inventory management;

[0174] By evaluating the operating risk level, identifying and warning possible risks such as goods shortage, expiration or equipment failure in time, reducing the potential risks in operation, and ensuring the stability of shelf management; by analyzing the sales data under different time periods or different display layouts, evaluating the effectiveness of the management strategy, thereby optimizing the sales strategy, and improving the predictability and stability of sales performance;

[0175] Through user feedback and questionnaires, the satisfaction of consumers with shelf display and goods supply is reflected, the trust and satisfaction of customers with products are enhanced, and the sales growth is promoted; by generating a comprehensive data management index, the comprehensive evaluation result based on weighted calculation provides a more scientific and comprehensive decision basis for shelf management, helping managers to better develop improvement measures;

[0176] Using comprehensive analysis algorithm, intelligent analysis and optimization of shelf management are carried out, the efficiency and intelligent degree of shelf management are improved, and the innovation and upgrading of management mode are promoted.

[0177] An embodiment of the present application, the S6, comprises:

[0178] S61, based on the comprehensive data management index, using autonomous intelligent decision algorithm to intelligently analyze and decide the goods control, display layout and operating risk of the shelf;

[0179] S62, according to the autonomous intelligent decision result, the shelf is adjusted and optimized in real time through the intelligent control system; and the decision execution result is fed back to the system to form a closed-loop control cycle.

[0180] The working principle of the above technical solution is: based on the comprehensive data management index, the self-intelligent decision algorithm is used to intelligently analyze and decide the goods control, display layout and operation risk of the shelf; the self-intelligent decision algorithm combines the preset rule base, expert knowledge model and machine learning algorithm, automatically generates decision suggestions such as goods replenishment strategy, goods allocation scheme, display adjustment scheme and operation risk early warning measures according to real-time data and market dynamics; the preset rule base contains basic rules and business processes of shelf management, such as inventory upper and lower limit rules, goods display specifications, etc.; the expert knowledge model summarizes the experience and knowledge of industry experts, providing reference for decision-making; the machine learning algorithm learns and analyzes historical data to mine potential rules and patterns in the data, providing intelligent support for decision-making; for example, when the comprehensive data management index shows that the inventory accuracy of a certain type of goods is low, the self-intelligent decision algorithm will analyze the reasons for the inventory difference according to the preset rules and historical data, and generate corresponding replenishment strategies or inventory check suggestions; when the operation risk level is high, the algorithm will issue early warning information in time and propose risk response measures;

[0181] According to the self-intelligent decision result, the shelf is adjusted and optimized in real time through the intelligent control system; for example, the goods replenishment process is automatically triggered, the replenishment order is sent to the supplier; the display layout of the shelf is adjusted, the best-selling goods are placed in a more eye-catching position; the operation risk response mechanism is started, the expired goods are removed, the equipment failure is repaired in time, etc.; and the decision execution result is fed back to the system to form a closed-loop control cycle; through monitoring and evaluation of the decision execution effect, the shelf management strategy and method are continuously optimized; for example, if the sales volume of goods does not increase significantly after the implementation of a display adjustment scheme, the system will analyze the reasons and adjust the display strategy; if the replenishment strategy leads to inventory accumulation, the system will optimize the replenishment parameters to improve the accuracy of inventory management; through the closed-loop control cycle, the intelligent level and operation efficiency of shelf management are continuously improved, realizing the automation, intelligence and refinement of shelf management.

[0182] The effect of the above technical solution is: through the combination of self-intelligent decision algorithm and comprehensive data management index, the system can automatically generate replenishment strategies, allocation schemes, display adjustment schemes, etc. according to real-time data and market dynamics, improving the intelligence and accuracy of shelf management; through the real-time operation risk early warning mechanism, the system can discover potential risks in time and propose corresponding response measures, avoiding inventory shortage, expired goods accumulation and other problems, reducing operation risk;

[0183] With the help of expert knowledge model, machine learning algorithm and historical data analysis, the system can intelligently analyze inventory differences and generate replenishment strategies and inventory count suggestions, avoiding inaccurate inventory and overstocking; according to real-time sales data and shelf layout, the system can automatically adjust the display position of goods, placing hot-selling goods in a prominent position, improving the exposure and sales of goods;

[0184] Through the automatic adjustment and optimization of the intelligent control system, the system can realize the automation of replenishment, allocation, display, risk response and other tasks, reducing manual intervention and improving operational efficiency; intelligent decision-making and automated execution reduce manual intervention and management costs, improve operational efficiency and reduce human resource investment;

[0185] The system can flexibly adjust management strategies according to real-time market changes and data analysis, ensuring timely optimization of shelf layout and product configuration, enhancing the flexibility and adaptability of shelf management; through intelligent replenishment and inventory management strategies, the system can ensure that goods are replenished at the right time, avoiding overstocking and improving inventory turnover;

[0186] Intelligent monitoring and risk response mechanisms help to detect expired or slow-selling goods early, reducing product waste and overstocking of slow-moving goods; by feeding back the decision execution results to the system and continuously optimizing management strategies based on actual results, a perfect closed-loop control mechanism is formed, ensuring continuous improvement and system optimization.

[0187] One embodiment of the present application is an intelligent shelf system, comprising:

[0188] One or more processors;

[0189] Memory for storing one or more programs,

[0190] When the one or more programs are executed by the one or more processors, the one or more processors implement the method of any one of the above.

[0191] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application; thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application also intends to include these modifications and variations.

Claims

1. A method for managing intelligent shelves, characterized in that, The method includes: S1. Design the topology architecture of multiple types of sensing devices for the intelligent shelf unit and construct the topology architecture data of the multi-type sensing signal fusion network; based on the multi-type sensing signal fusion network topology architecture data, collect relevant information of the shelf in real time to form the time-series raw shelf monitoring data stream; use data preprocessing algorithms to preprocess the time-series raw shelf monitoring data stream to generate preprocessed shelf monitoring data. S2. Based on the multi-type sensing signal fusion network topology data, a distributed data storage node optimization algorithm is adopted to dynamically lay out and allocate resources for storage nodes to construct a distributed data storage architecture. Using the distributed data storage architecture, a multi-path redundant transmission strategy is adopted to transmit the pre-processed shelf monitoring data to multiple storage nodes simultaneously, generating redundant monitoring upload data packets. S3. Generate a unique upload credential for each redundant monitoring upload data packet. Based on the upload credential, perform a difference analysis on the redundant monitoring upload data received by different storage nodes; generate a monitoring upload difference data report; resolve the difference conflicts of redundant monitoring upload data according to the monitoring upload difference data report; and generate accurate shelf unit monitoring data. S4. Based on the characteristics of the shelving unit monitoring data and management needs, a dynamic data sharding strategy is used to flexibly shard the data according to multiple dimensions to generate monitoring shard feature data. Based on the monitoring shard feature data, an intelligent cache synchronization algorithm is adopted to cache commonly used data shards in a high-speed cache device and synchronize the cached data in real time according to the data update. S5. Based on cache management log data, use the cache status evaluation model to comprehensively evaluate the cache status and generate monitoring cache status data; based on the monitoring cache status data, combined with the business objectives and key performance indicators of shelf management, use a comprehensive analysis algorithm to conduct a comprehensive analysis of the management status of the monitoring shard feature data and generate a comprehensive data management index. S6. Based on the comprehensive data management index, the system uses autonomous intelligent decision-making algorithms to intelligently analyze and make decisions regarding the control of goods on the shelves, the layout of the shelves, and operational risks. Based on the results of the autonomous intelligent decision-making, the system adjusts and optimizes the shelves in real time through the intelligent control system. At the same time, the results of the decision execution are fed back into the system to form a closed-loop control cycle. S2 includes: S21. Based on the multi-type sensing signal fusion network topology architecture data, collect relevant information from each storage node, and organize and analyze it. S22. A distributed data storage node optimization algorithm is adopted to dynamically deploy and allocate resources for storage nodes; a distributed data storage architecture is constructed, dividing storage nodes into different levels and regions for hierarchical data storage and distributed management. S23. Using the constructed distributed data storage architecture, a multi-path redundant transmission strategy is adopted to transmit the pre-processed shelf monitoring data to multiple storage nodes simultaneously. S24. During the transmission process, the data is encrypted, redundant monitoring and uploading data packets are generated, and the transmission process is monitored and logged in real time to promptly detect and handle transmission anomalies. S23 includes: A comprehensive scan of the network topology involved in the distributed data storage architecture is performed to identify all available network links and the connections between storage nodes; Based on the network topology information obtained from the scan, a multi-path generation algorithm is used to generate multiple potential transmission paths from the data source to each target storage node. Each potential transmission path generated is comprehensively evaluated using a weighted scoring method, which assigns different weights to different indicators of the path to calculate the comprehensive score of each path; the path with the highest score is selected as the actual data transmission path. Based on the characteristics and size of the pre-processed shelf monitoring data, an intelligent data fragmentation strategy is adopted to divide the data into multiple appropriately sized data fragments; a certain number of redundant data fragments are generated for each data fragment, and the generated data fragments and redundant data fragments are allocated to different storage nodes on multiple previously selected transmission paths. The dynamic transmission scheduling algorithm dynamically adjusts the transmission order and rate of data slices on different transmission paths based on the real-time monitored network link status and storage node load. Establish a collaborative mechanism between paths to enable data transmission on different transmission paths to cooperate with each other; when data transmission on a certain path is delayed or fails, data transmission on other paths can automatically speed up or take on more transmission tasks. During the data transmission process, the data transmission status on each transmission path is monitored in real time, and the monitored information is promptly transmitted to the transmission scheduling algorithm through a real-time feedback mechanism. After all data slices and redundant data slices have been successfully transmitted to the target storage node, each storage node sends a transmission completion confirmation message to the data source. After receiving confirmation messages from all storage nodes, the data source confirms that the data transmission task is complete. At the same time, a timeout retransmission mechanism is adopted. If no confirmation message is received from a storage node within the specified time, the data source will retransmit the data slices that the storage node did not receive. After the data transmission is completed, a consistency check is performed on the data stored on each storage node; a hash algorithm is used to calculate the check value for each data slice and redundant data slice, and the check value is compared with the check value pre-calculated by the data source; If the checksums are found to be inconsistent, it indicates that the data has been damaged or lost during transmission. The data recovery mechanism is then activated to recover the original data using redundant data fragments, and the transmission and verification are repeated until the data on all storage nodes are consistent.

2. The intelligent shelf management method according to claim 1, characterized in that, S1 includes: S11. Conduct a comprehensive spatial analysis and functional planning for the intelligent shelf unit, and determine the installation locations of various sensing devices based on the shelf size, number of layers, product display method, and management needs. S12. Clarify the interconnection relationships between various sensing devices and construct the network topology using a combination of wired and wireless communication methods; S13. Record the sensing range of various sensing devices, and determine the area and boundary that each device can effectively monitor through actual testing and simulation analysis; set the working parameters according to the working principle and performance characteristics of different types of sensing devices. S14. Based on the constructed multi-type sensing signal fusion network topology data, start various sensing devices and collect relevant information about the shelf in real time according to the set sampling frequency; S15. Initially integrate the collected data and generate a time-series raw shelving monitoring data stream in chronological order; process the time-series raw shelving monitoring data stream using data cleaning algorithms and convert data of different dimensions into a unified standard range using data normalization methods; further reduce noise interference in the data using data denoising algorithms; after preprocessing, generate preprocessed shelving monitoring data.

3. The intelligent shelf management method according to claim 1, characterized in that, S22 includes: Based on the collected data access frequency information of each storage node, the number of times each storage node is accessed in different time periods is counted using the time axis as a benchmark. According to the statistical results of access frequency, the storage nodes are divided into different hot zones. A threshold for access frequency is set, and the area with access frequency higher than the threshold is defined as a high-heat area, the area with access frequency lower than the threshold but with a certain amount of access is defined as a medium-heat area, and the area with extremely low access frequency is defined as a low-heat area. For high-traffic areas, prioritize allocating more storage resources and processing power; for medium-traffic areas, allocate storage resources and processing power reasonably based on their access frequency and business needs; for low-traffic areas, allocate basic storage resources. Analyze network latency and bandwidth across different geographical locations; optimize network topology based on geographical location analysis and network performance test results; conduct a heat assessment on pre-processed shelf monitoring data; and categorize the data into three types: hot data, warm data, and cold data. Based on the data popularity assessment results, a hierarchical storage architecture is designed: hot data is stored in the high-speed cache layer, warm data is stored in the middle layer, and cold data is stored in the low-speed storage layer. Establish a real-time monitoring system to monitor the resource usage, data access, and network performance of storage nodes in real time; based on the feedback information from the real-time monitoring, use a dynamic resource adjustment algorithm to dynamically adjust the resource allocation of storage nodes.

4. The intelligent shelf management method according to claim 1, characterized in that, The S3 includes: S31. Generate a unique upload credential information for each redundant monitoring upload data packet, and associate the upload credential information with the redundant monitoring upload data packet for storage; S32. Based on the uploaded credential information, perform a difference analysis on the redundant monitoring uploaded data received by different storage nodes; identify the data parts that differ; record and analyze the data that differs in detail, and generate a monitoring uploaded difference data report. S33. A data consistency negotiation algorithm and intelligent voting decision-making mechanism are adopted to resolve the differences and conflicts of redundant monitoring and uploaded data based on the monitoring and uploaded difference data reports. During the negotiation process, each storage node interacts with data and shares information, sending its own data and credential information to other nodes for comparison and verification. S34. Generate accurate shelf unit monitoring data, update the data in each storage node, and record and audit the process of resolving discrepancies and conflicts.

5. The intelligent shelf management method according to claim 1, characterized in that, The S4 includes: S41. Based on the characteristics of the shelving unit monitoring data and management needs, use a dynamic data segmentation strategy to segment the data according to multiple dimensions; generate monitoring segment feature data, and record the feature information of each data segment; S42. Based on monitoring fragmentation characteristic data, a smart cache synchronization algorithm is used to cache frequently used data fragments into a high-speed cache device; the smart cache synchronization algorithm dynamically adjusts the caching strategy according to the data access pattern and update frequency; S43. Synchronize cached data in real time according to data updates. When data in the underlying storage is updated, notify the caching system to update the corresponding cached data. When data in the cache is modified, write the data back to the underlying storage according to the set synchronization strategy. S44. Generate cache management log data to record cache data information; obtain the operating status of the cache system by analyzing the cache management log data.

6. An intelligent shelving system, comprising: One or more processors; Memory, used to store one or more programs. Wherein, when the one or more programs are executed by the one or more processors, the one or more processors implement the method of any one of claims 1 to 5.

Images