CN-121981705-A - Unmanned vending equipment state monitoring method and system based on integration of flow batch

CN121981705ACN 121981705 ACN121981705 ACN 121981705ACN-121981705-A

Abstract

The invention discloses a method and a system for monitoring the state of unmanned vending equipment based on a flow batch integration, wherein the method collects multi-dimensional state data of the equipment through a data collection layer and performs unified structuring processing based on a data model, a flow processing layer obtains structured equipment state event flows from a Kafka data source, real-time processing is performed through timestamp distribution, data grouping and window aggregation strategies, a multi-dimensional health scoring algorithm is called to calculate real-time health scores, a storage layer adopts a flow batch integration architecture to store data into corresponding data tables respectively to meet real-time query and historical analysis requirements, an analysis query layer obtains analysis results based on the aggregation calculation of the data by the aggregation tables, and an intelligent decision layer calls an anomaly detection algorithm to perform anomaly judgment to realize equipment health assessment and predictive maintenance. The invention effectively solves the problems of insufficient real-time processing, poor storage adaptability, intelligent prediction deficiency, weak multidimensional analysis and limited expansibility of the existing monitoring system.

Inventors

WANG LEI
HUANG AIHUA
YIN JUEHUI

Assignees

上海趣致网络科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260126

Claims (10)

1. The method for monitoring the state of the unmanned vending equipment based on the integration of the flow batch is characterized by comprising the following steps: s1, acquiring multi-dimensional state data of unmanned vending equipment through a data acquisition layer, and carrying out unified structural processing on the multi-dimensional state data based on a preset data model, wherein the data model comprises equipment ID, a time stamp, temperature, humidity, door opening times, voltage, position information and a sensor data mapping field; s2, the stream processing layer utilizes the device state event stream after the S1 unified structuring processing from the Kafka data source to process the device state data in real time through timestamp distribution, data grouping and window aggregation strategies, and invokes a multi-dimensional health scoring algorithm to calculate the real-time health score of the device; S3, the storage layer adopts a stream batch integrated storage architecture, and the real-time equipment state data and the health grading data processed in the S2 are respectively stored into corresponding data tables so as to meet the requirements of real-time inquiry and historical data analysis; S4, the analysis query layer carries out aggregation calculation on the real-time stored data and the historical stored data of the S3 based on the constructed aggregation table to obtain an aggregation analysis result; S5, the intelligent decision layer calls an abnormality detection algorithm to perform abnormality judgment on the real-time equipment state data output by the S2 and the aggregation analysis result obtained by the S4, so that equipment health assessment and predictive maintenance are realized.
2. The method for monitoring the state of the vending machine based on the integration of the flow batch according to claim 1, wherein in the step S2, the flow processing layer processes the data with deviation in time through a preset out-of-order data compatibility strategy based on a flow computing engine, groups the data according to a unique identifier of the equipment, aggregates the same equipment data by adopting a sliding window mechanism, and completes the calculation of the health score of the equipment through customized aggregation logic; in step S2, the calculation formula of the multidimensional health scoring algorithm is as follows: In the formula, Is the first The weight coefficients of the individual dimensions comprise weights corresponding to hardware states, environmental factors, use frequencies and maintenance histories; Is the first Normalized scores of the dimensions are obtained by normalizing the original data of each dimension; Is the first And the importance coefficients of the individual dimensions are dynamically adjusted according to the equipment operation scene.
3. The method for monitoring the status of a vending machine based on the integration of a batch as claimed in claim 1, wherein in step S3, the data table includes a real-time table of the status of the machine and a history table of the health score of the machine; The equipment state real-time table is provided with equipment identification, time information, environment parameters, use parameters, position information, health scores and equipment state fields, and is used for carrying out partition storage according to the time dimension and configuring the corresponding storage partition number and data writing mode; The equipment health score history table is provided with equipment identification, time information, score of each dimension, comprehensive health score and predictive score fields, and is used for carrying out partition storage according to the dimension of the date, and configuring the corresponding storage partition number and data writing mode.
4. The method for monitoring the state of the vending machine based on the integration of the flow batch according to claim 1, wherein in the step S4, an aggregation table is constructed by analyzing a database in real time, the equipment identification, the position information, the date information and the hour information are used as aggregation dimensions, and a hash distribution mode is adopted to store data and configure the corresponding storage partition number; The indicators stored in the aggregation table comprise average health scores, lowest health scores, highest health scores, abnormal times, maintenance alarm times and total event numbers, and the multidimensional aggregation query is supported according to equipment, positions and time dimensions.
5. The method for monitoring the status of a vending machine based on the integration of a batch as claimed in claim 1, wherein in step S5, the abnormality determination condition of the abnormality detection algorithm is: In the formula, A standardized deviation value for the current equipment status index; As the current index value of the index, As a mean value of the historical data, Standard deviation of historical data; Is that Is an anomaly threshold value for (2); is 25% of the quantiles of the historical data, 75% Quantile of historical data; the four bit distances are calculated according to the formula ; And the actual value of the current equipment state index is obtained.
6. Unmanned sales equipment state monitored control system based on it is integrative that class, its characterized in that includes: The system comprises a data acquisition module, a data processing module and a data processing module, wherein the data acquisition module is used for acquiring multi-dimensional state data of unmanned vending equipment through a data acquisition layer and carrying out unified structuring processing on the multi-dimensional state data based on a preset data model, and the data model comprises an equipment ID, a time stamp, temperature, humidity, door opening times, voltage, position information and a sensor data mapping field; The stream processing module is used for the stream processing layer to utilize the uniformly structured device state event stream from the Kafka data source, process the device state data in real time through timestamp distribution, data grouping and window aggregation strategies, and call a multidimensional health scoring algorithm to calculate the real-time health score of the device; the data storage module is used for storing the processed real-time equipment state data and health scoring data to corresponding data tables respectively by adopting a stream batch integrated storage architecture in the storage layer so as to meet the requirements of real-time query and historical data analysis; the aggregation analysis module is used for analyzing the aggregation calculation of the real-time storage data and the historical storage data based on the constructed aggregation table by the query layer to obtain an aggregation analysis result; The abnormality judgment module is used for calling an abnormality detection algorithm by the intelligent decision layer to carry out abnormality judgment on the output real-time equipment state data and the obtained aggregation analysis result, so as to realize equipment health assessment and predictive maintenance.
7. The system for monitoring the state of the vending machine based on the integration of the flow batch according to claim 6, wherein in the flow processing module, a flow processing layer processes data with deviation in time through a preset out-of-order data compatibility strategy based on a flow computing engine, groups the data according to a unique identifier of the equipment, performs aggregation processing on the same equipment data by adopting a sliding window mechanism, and completes calculation of the health score of the equipment through customized aggregation logic; In the stream processing module, the calculation formula of the multidimensional health scoring algorithm is as follows: In the formula, Is the first The weight coefficients of the individual dimensions comprise weights corresponding to hardware states, environmental factors, use frequencies and maintenance histories; Is the first Normalized scores of the dimensions are obtained by normalizing the original data of each dimension; Is the first And the importance coefficients of the individual dimensions are dynamically adjusted according to the equipment operation scene.
8. The system for monitoring the status of the vending machine based on the integration of the fluid batch according to claim 6, wherein the data table comprises a real-time table of the status of the machine and a history table of the health score of the machine in the data storage module; The equipment state real-time table is provided with equipment identification, time information, environment parameters, use parameters, position information, health scores and equipment state fields, and is used for carrying out partition storage according to the time dimension and configuring the corresponding storage partition number and data writing mode; The equipment health score history table is provided with equipment identification, time information, score of each dimension, comprehensive health score and predictive score fields, and is used for carrying out partition storage according to the dimension of the date, and configuring the corresponding storage partition number and data writing mode.
9. The system for monitoring the state of the unmanned vending machine based on the integration of the flow batch according to claim 6, wherein in the aggregation analysis module, an aggregation table is constructed through real-time analysis of a database, and the data is stored and the corresponding storage partition number is configured in a hash distribution mode by taking the equipment identifier, the position information, the date information and the hour information as aggregation dimensions; The indicators stored in the aggregation table comprise average health scores, lowest health scores, highest health scores, abnormal times, maintenance alarm times and total event numbers, and the multidimensional aggregation query is supported according to equipment, positions and time dimensions.
10. The system for monitoring the status of a vending machine based on the integration of a batch of fluid according to claim 6, wherein in the anomaly determination module, the anomaly determination conditions of the anomaly detection algorithm are: In the formula, A standardized deviation value for the current equipment status index; As the current index value of the index, As a mean value of the historical data, Standard deviation of historical data; Is that Is an anomaly threshold value for (2); is 25% of the quantiles of the historical data, 75% Quantile of historical data; the four bit distances are calculated according to the formula ; And the actual value of the current equipment state index is obtained.

Description

Unmanned vending equipment state monitoring method and system based on integration of flow batch Technical Field The invention relates to the technical field of unmanned vending equipment, in particular to a method and a system for monitoring the state of unmanned vending equipment based on integration of fluid batches. Background Along with popularization of mobile payment technology, diversified expansion of consumption scenes and rapid rise of unmanned retail industry, deployment scale of unmanned vending equipment (such as intelligent vending machines, unmanned shelves, automatic vending cabinets and the like) is exponentially increased, various places such as transportation hubs, office parks, communities and campuses are widely covered, and the unmanned vending equipment is an important carrier for convenient consumption. The stable operation of the equipment is directly related to user consumption experience, commodity loss control and operation income, so that the equipment state monitoring becomes a core link of unmanned retail operation management. However, the existing state monitoring system for the unmanned vending equipment has a plurality of limitations in technical architecture and function realization, and is difficult to meet the large-scale, high-precision and intelligent monitoring requirements. The traditional system adopts a batch processing mode, the data processing period is in the order of hours, equipment fault discovery is delayed, commodity loss is easy to cause, user experience is reduced, and the operation efficiency is seriously affected. The existing system mostly adopts a single data storage architecture, and either focuses on historical analysis but has high real-time query delay, or focuses on data storage but has low historical analysis efficiency, so that the dual requirements of real-time monitoring and historical data analysis cannot be met at the same time. Most of the existing systems are in a mode of 'post alarm', lack of an effective equipment health assessment model, incapability of predicting potential faults through multi-dimensional data fusion, difficulty in realizing predictive maintenance, long unplanned downtime and high maintenance cost. The equipment operation is influenced by multiple factors such as hardware working condition, environmental condition, use frequency, maintenance record and the like, the existing system lacks multidimensional data integration analysis capability, cannot provide accurate basis for operation decision, and is difficult to realize fine management. In the face of millions of equipment deployment scale, the traditional architecture is difficult to support concurrent data acquisition, processing and storage, the performance is easy to bottleneck, the expansion cost is high, and the requirement of industrial scale development cannot be met. Disclosure of Invention Therefore, the invention provides a method and a system for monitoring the state of unmanned vending equipment based on integration of flow batch, which solve the problems of insufficient real-time processing, poor suitability of a storage architecture, intelligent prediction loss, weak multidimensional analysis and limited expansibility of the existing unmanned vending equipment monitoring system. In order to achieve the above purpose, the invention provides a method for monitoring the state of the unmanned vending equipment based on the integration of a flow batch, which comprises the following steps: s1, acquiring multi-dimensional state data of unmanned vending equipment through a data acquisition layer, and carrying out unified structural processing on the multi-dimensional state data based on a preset data model, wherein the data model comprises equipment ID, a time stamp, temperature, humidity, door opening times, voltage, position information and a sensor data mapping field; s2, the stream processing layer utilizes the device state event stream after the S1 unified structuring processing from the Kafka data source to process the device state data in real time through timestamp distribution, data grouping and window aggregation strategies, and invokes a multi-dimensional health scoring algorithm to calculate the real-time health score of the device; S3, the storage layer adopts a stream batch integrated storage architecture, and the real-time equipment state data and the health grading data processed in the S2 are respectively stored into corresponding data tables so as to meet the requirements of real-time inquiry and historical data analysis; S4, the analysis query layer carries out aggregation calculation on the real-time stored data and the historical stored data of the S3 based on the constructed aggregation table to obtain an aggregation analysis result; S5, the intelligent decision layer calls an abnormality detection algorithm to perform abnormality judgment on the real-time equipment state data output by the S2 and the aggregation analysis result obtaine