PCS equipment operation and maintenance management system based on Springboot multiple tenants and multiple data sources

Abstract

The invention relates to the technical field of operation and maintenance of industrial internet of things and power electronic equipment, in particular to a multi-tenant and multi-data-source PCS equipment operation and maintenance management system based on Springboot, which comprises an edge layer consisting of a plurality of PCS equipment, a transmission layer comprising an MQTT proxy and an FRP intranet penetration component, a processing layer constructed based on a Springboot framework, and a control layer comprising an MQTT proxy and an FRP intranet penetration component. Comprising a multi-tenant management module, a data processing module, an alarm module, a remote control module, a work order system and an intelligent analysis module, the storage layer adopts a mixed database architecture, and the application layer comprises a front-end display module and a system monitoring module. Unification of high safety and high resource utilization rate is achieved, and through an innovative'dynamic data source 'multi-tenant framework, the contradiction between data isolation and the resource utilization rate in multi-customer service of a traditional system is successfully solved. Therefore, absolute security and independence of data of each tenant are guaranteed, hardware and operation and maintenance cost is greatly reduced by sharing the application instance and the database server, and the resource utilization rate is improved.

Description

Technical Field

[0001] This invention relates to the field of industrial Internet of Things and power electronic equipment operation and maintenance technology, specifically to a PCS equipment operation and maintenance management system based on Springboot multi-tenant and multi-data source. Background Technology

[0002] With the rapid development of the new energy industry and industrial automation, the application scenarios of PCS (Power Conversion System) equipment are becoming increasingly widespread, and the number is increasing dramatically. This places higher demands on the efficiency, safety, and intelligence of equipment operation and maintenance management. Currently, it is involved in the following areas: New energy power plant operation and maintenance: Applied to photovoltaic power plants, energy storage power plants and other scenarios, it enables centralized monitoring, alarm, remote maintenance and health assessment of PCS equipment such as energy storage converters and photovoltaic inverters in the station, thereby improving the efficiency of power plant operation and maintenance and reducing labor costs.

[0003] Industrial automation production line management: Applied to industrial automation production lines that use PCS equipment for power management or motor control, it enables real-time status monitoring, fault warning, and remote parameter configuration of key power equipment in the production line, ensuring production continuity.

[0004] Data center power management: Used for PCS equipment such as UPS (Uninterruptible Power Supply) or high-voltage DC systems in data centers to monitor their operating status and battery health, and perform remote maintenance and firmware upgrades to ensure the safety of power supply in data centers.

[0005] However, the PCS equipment operation and maintenance management system in the above application scenarios still needs improvement, as follows: Insufficient data isolation and security: In multi-customer, multi-site application scenarios, traditional SCADA or general cloud platforms struggle to achieve effective data isolation. Device data, configuration information, and maintenance records from different tenants (customers) are easily leaked to each other, posing serious data security risks and failing to meet enterprise-level data security requirements. Low system scalability and resource utilization: Monolithic systems are difficult to scale horizontally as the number of connected devices and tenants increases. Deploying a separate system for each new tenant not only results in long deployment cycles but also leads to a huge waste of hardware resources such as servers and databases, and high operation and maintenance costs. The automation and intelligence levels of operation and maintenance processes are low: Existing technologies focus mainly on data collection and display, lacking the ability to automate and intelligently link alarms, work orders, remote maintenance, and knowledge base accumulation. After an alarm occurs, manual notification and handling are required, work order processes are not standardized, and historical fault experience is difficult to reuse effectively, resulting in long problem-solving cycles and difficulty in improving operation and maintenance efficiency. Limited remote maintenance capabilities: For complex industrial equipment such as PCS devices, simple data reading and writing cannot meet the needs of in-depth operation and maintenance. Existing systems often lack secure and reliable remote channels, making it impossible for operation and maintenance engineers to easily access the device's web backend for in-depth diagnostics, firmware upgrades, and other operations, still requiring frequent on-site visits; Therefore, a PCS device operation and maintenance management system based on Spring Boot multi-tenant and multi-data source is needed to improve the above problems. Summary of the Invention

[0006] To address the issues of insufficient data isolation and security, low system scalability and resource utilization, and low levels of automation and intelligence in operation and maintenance processes, this invention provides a PCS device operation and maintenance management system based on Spring Boot with multiple tenants and multiple data sources, and addresses the limitation of remote maintenance capabilities.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a PCS device operation and maintenance management system based on Spring Boot multi-tenant multi-data source, comprising: The edge layer consists of several PCS devices and is used to collect device operation data.

[0008] The transport layer includes MQTT brokers and FRP NAT traversal components.

[0009] The processing layer, built on the Spring Boot framework, includes a multi-tenant management module, a data processing module, an alarm module, a remote control module, a work order system, and an intelligent analysis module.

[0010] The storage layer employs a hybrid database architecture, including relational databases, time-series databases, and cache databases.

[0011] The application layer includes the front-end display module and the system monitoring module.

[0012] The multi-tenant management module achieves tenant data isolation through a dynamic data source routing mechanism. Specifically, it creates an independent schema or database instance for each tenant in the relational database. When the processing layer receives a request, it parses the tenant identifier through a request interceptor and routes the database operation to the corresponding tenant data source based on the identifier using the dynamic data source framework.

[0013] As a preferred embodiment of the present invention, the dynamic data source routing mechanism includes: A tenant context holder stores the tenant identifier of the current request based on ThreadLocal; a request interceptor is used to extract and set the tenant identifier from the HTTP request to the tenant context holder.

[0014] The dynamic data source router, which inherits from AbstractRoutingDataSource, uses the determineCurrentLookupKey method to obtain the tenant identifier as the data source lookup key from the tenant context holder.

[0015] As a preferred embodiment of the present invention, the remote control module integrates FRP intranet penetration service, used for: Receive remote access requests and perform permission verification.

[0016] Based on the target PCS device's internal network address information, dynamically start the FRP client process to establish an encrypted proxy channel between the public network port and the device's internal web port.

[0017] Generate and return the corresponding remote access link.

[0018] As a preferred embodiment of the present invention, the alarm module, the work order system and the knowledge base module form an automated operation and maintenance closed loop. The specific process is as follows: after the alarm is triggered, the alarm module automatically calls the work order system interface to create a maintenance work order. After the work order is processed, the submitted solution is stored in a structured manner in the knowledge base module; When a new alarm triggers a work order, the system calls the large model interface to match and recommend solutions for similar historical cases from the knowledge base based on the alarm information.

[0019] As a preferred embodiment of the present invention, the intelligent analysis module includes a health assessment unit, used for: The key component runtime timing data of the device is extracted from the timing database according to a preset cycle.

[0020] Based on a predefined weighting model, the operating status of each component is scored and the overall health score is calculated by weighting.

[0021] Generate a health assessment report that includes health level, identification of abnormal components, and optimization suggestions.

[0022] As a preferred embodiment of the present invention, the data processing module is provided with out-of-order data processing logic to ensure that time-series data from the same device with inconsistent timestamp orders can be displayed in the correct time order at the application layer.

[0023] As a preferred embodiment of the present invention, the tenant identifier is derived from at least one of the following methods: tenant prefix information in the user login account, tenant identifier parameters carried in the HTTP request header, or tenant configuration information bound to the device serial number.

[0024] As a preferred embodiment of the present invention, the data processing module not only stores the parsed real-time device data into the time-series database, but also actively pushes the real-time data to the front-end display module through WebSocket or server event sending technology to achieve real-time refresh of the monitoring interface.

[0025] As a preferred embodiment of the present invention, the knowledge base module integrates an Elasticsearch search engine for indexing structured solutions to support fast matching and semantic retrieval.

[0026] As a preferred embodiment of the present invention, the intelligent analysis module further includes a fault prediction unit, which uses historical operating data and fault records of the equipment obtained from the relational database and the time-series database to train a machine learning model to predict the failure probability and potential risk components of the specified equipment in the future period.

[0027] Compared with existing technologies, this invention achieves a balance between high security and high resource utilization: through an innovative "dynamic data source" multi-tenant architecture, it successfully resolves the contradiction between data isolation and resource utilization in traditional systems serving multiple clients. It ensures the absolute security and independence of each tenant's data while significantly reducing hardware and maintenance costs and improving resource utilization by sharing application instances and database servers.

[0028] Compared with existing technologies, this invention constructs an automated and intelligent closed-loop operation and maintenance system: it organically connects equipment monitoring, alarms, work orders, remote maintenance, and knowledge base to form an automated operation and maintenance workflow. In particular, the design of automatic alarm triggering of work orders and intelligent recommendation solutions from the knowledge base significantly reduces manual intervention, shortens fault handling time, and upgrades the operation and maintenance mode from "passive response" to "proactive prediction and intelligent assistance".

[0029] Compared to existing technologies, this invention provides secure and efficient remote deep operation and maintenance capabilities: by integrating FRP intranet penetration technology, it provides maintenance personnel with a secure, convenient, and low-cost remote channel, enabling them to perform in-depth diagnostics, parameter configuration, and firmware upgrades on PCS devices as if they were on-site. This significantly reduces the need for on-site service, lowers travel and time costs, and improves customer satisfaction.

[0030] Compared with existing technologies, this invention has superior system scalability and flexibility: based on the microservice concept of Spring Boot and cloud-native architecture, the system can be easily horizontally scaled to support more device access and tenant onboarding. At the same time, the multi-tenant architecture makes it possible to provide services to new customers simply by opening an account, shortening the deployment cycle from weeks to minutes, and providing extremely fast business response speed. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of the overall structure of the present invention; Figure 2 This is a core business process diagram of the present invention; Figure 3 This is a flowchart illustrating the multi-tenant data isolation process of the present invention. Detailed Implementation

[0032] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0033] Example 1: A PCS device operation and maintenance management system based on Spring Boot multi-tenant and multi-data source includes: The edge layer consists of several PCS devices, which are responsible for collecting telemetry data (temperature, current) and remote signaling data (device status). Data is generated according to a preset sampling period (2 seconds) and supports MQTT protocol data transmission.

[0034] The transport layer includes the MQTT broker (MQTTBroker) and the FRP intranet penetration component. The MQTTBroker is responsible for encrypting and transmitting device data to the processing layer, while the FRP component allows maintenance engineers to remotely access the PCS device's web backend through the application layer, ensuring data transmission security and remote connection stability. The processing layer is built on the Spring Boot, Spring MVC, and Spring Security frameworks. It includes core modules such as multi-tenant management, data processing, alarm, remote control, intelligent analysis, and work order system. It is responsible for data parsing, permission verification, and business logic processing (such as alarm judgment and work order generation).

[0035] The storage layer adopts a hybrid storage solution of "time-series database (TDengine) + relational database (MySQL) + cache database (Redis) + local cache (Caffeine)". TDengine stores telemetry / telecommunications time-series data, MySQL stores basic device information, configuration data, and operation records, and Redis and Caffeine are used to cache tenant information, permission data, and real-time alarm data to improve system access speed.

[0036] The application layer includes a front-end display module (developed based on Vue3 + ElementPlus / Vben + Ant-Design-Vue) and a system monitoring module (SpringBootAdmin, SkyWalking). The front-end module provides a visual interface for device lists, details pages, alarm notifications, work order management, health reports, etc. The system monitoring module monitors the service running status, data transmission latency, and database performance in real time, and supports rapid troubleshooting. The multi-tenant management module achieves tenant data isolation through a dynamic data source routing mechanism, specifically: Tenant Registration and Initialization: Administrators create tenants through the multi-tenant management module. The system automatically creates an independent schema for each tenant in MySQL, initializes a tenant-specific data table, and caches tenant information (tenant ID, schema name, and permission configuration) in Redis. Tenant Identity Verification: When a user logs in, the system verifies the account and password using the Spring Security framework, extracts the tenant ID from the login information (such as account prefix and request header parameters), and retrieves the corresponding schema name and data source configuration from Redis based on the tenant ID. Dynamic Data Source Switching: The DynamicDatasource framework intercepts database operation requests and automatically switches to the corresponding schema data source based on the tenant ID, achieving tenant data isolation and preventing data access between different tenants. Simultaneously, the HibernateValidator component validates tenant request parameters to prevent illegal data operations.

[0037] In this embodiment, the dynamic data source routing mechanism includes: A tenant context holder stores the tenant identifier of the current request based on ThreadLocal; a request interceptor is used to extract and set the tenant identifier from the HTTP request to the tenant context holder.

[0038] The dynamic data source router, which inherits from AbstractRoutingDataSource, uses the determineCurrentLookupKey method to obtain the tenant identifier as the data source lookup key from the tenant context holder.

[0039] In this embodiment, the remote control module integrates FRP intranet penetration service for: Receive remote access requests and perform permission verification.

[0040] Based on the target PCS device's internal network address information, dynamically start the FRP client process to establish an encrypted proxy channel between the public network port and the device's internal web port.

[0041] Generate and return the corresponding remote access link.

[0042] In this embodiment, the alarm module, work order system, and knowledge base module form an automated operation and maintenance closed loop. The specific process is as follows: After an alarm is triggered, the alarm module automatically calls the work order system interface to create a maintenance work order. After the work order is processed, the submitted solution is stored in a structured manner in the knowledge base module; When a new alarm triggers a work order, the system calls the large model interface to match and recommend solutions for similar historical cases from the knowledge base based on the alarm information.

[0043] In this embodiment, the intelligent analysis module includes a health assessment unit, used for: The key component runtime timing data of the device is extracted from the timing database according to a preset cycle.

[0044] Based on a predefined weighting model, the operating status of each component is scored and the overall health score is calculated by weighting.

[0045] Generate a health assessment report that includes health level, identification of abnormal components, and optimization suggestions.

[0046] In this embodiment, the data processing module is equipped with out-of-order data processing logic to ensure that time-series data from the same device with inconsistent timestamp orders can be displayed in the correct time order at the application layer.

[0047] In this embodiment, the tenant identifier is derived from at least one of the following methods: tenant prefix information in the user login account, tenant identifier parameters carried in the HTTP request header, or tenant configuration information bound to the device serial number.

[0048] In this embodiment, while storing the parsed real-time device data into the time-series database, the data processing module also actively pushes the real-time data to the front-end display module via WebSocket or server event sending technology to achieve real-time refresh of the monitoring interface.

[0049] In this embodiment, the knowledge base module integrates an Elasticsearch search engine to index structured solutions, supporting fast matching and semantic retrieval.

[0050] In this embodiment, the intelligent analysis module further includes a fault prediction unit, which uses historical operating data and fault records of the equipment obtained from the relational database and the time-series database to train a machine learning model to predict the failure probability and potential risk components of the specified equipment in the future period.

[0051] Furthermore, the specific workflow of the system's collaborative operation is described in conjunction with five typical scenarios: Scenario 1: Equipment Data Acquisition and Real-time Monitoring Data Acquisition: The PCS device (edge ​​layer) collects telemetry and teleindication data at a configured 2-second sampling period, generating raw data containing timestamps, device serial numbers (SN), and data from the collection points. Data Transmission: The PCS device assembles the raw data into JSON messages, encrypts them via the MQTT protocol (using SSL / TLS encryption), and transmits them to the MQTTBroker. The transport layer ensures data integrity and security. Data processing: The data processing module in the processing layer listens to the message queue of MQTTBroker, receives JSON messages and parses them (extracting device SN, data type, and value), obtains the tenant ID through the device SN, and switches the dynamic data source based on the tenant ID; at the same time, the data processing module pushes real-time data to the front-end display module through WebSocket, and the front-end PCS list page displays data blocks and the details page refreshes data in real time. Data storage: The data processing module writes telemetry / remote signaling data into TDengine tables (partitioned by device SN and timestamp to optimize time-series data query speed), and updates device status (such as online / offline, abnormal operation) to the corresponding MySQL schema tables; at the same time, Redis caches the real-time data of the last 5 minutes to improve the efficiency of front-end data query.

[0052] Scenario 2: Alarm Triggering and Notification Response Alarm Judgment: The alarm module in the processing layer periodically (synchronized with the sampling period, every 2 seconds) reads real-time data from the TDengine table, obtains the alarm threshold (maximum value, minimum value) and "indicator alarm level" (3 - severe, 2 - moderate, 1 - mild) from the table of the corresponding tenant schema in MySQL, compares the real-time data with the alarm threshold, and determines whether an alarm is triggered and the alarm level. Notification Triggering: If an alarm is triggered, the alarm module retrieves the contact information (phone number, email address) of the associated "site manager" from the MySQL table, calls the SMS gateway (such as Alibaba Cloud SMS service) and email gateway (such as SMTP service), and sends the alarm information (including device SN, alarm content, alarm level, and trigger time). Simultaneously, the alarm information is stored in the Redis alarm cache queue for real-time display on the front end. Work Order Generation: The alarm module calls the work order system interface to automatically create a work order based on the alarm information (including work order ID, device SN, alarm level, creation time, and pending status), writes the work order data to the work order table in the corresponding tenant schema of MySQL, and adds the work order to the pending queue. The work order system supports sorting by alarm level, prioritizing critical alarm work orders. Scenario 3: Remote Maintenance and Control Permission verification: Operations engineers log in to the system through the front end. The Spring Security framework verifies user permissions based on a three-level permission model (visitor, engineer, administrator). It retrieves the range of devices that the user can operate on (associated device SN) from the permission table of the corresponding tenant schema in MySQL. If an engineer initiates a remote maintenance operation (such as parameter configuration or firmware upgrade), the system verifies the operation permission (ensuring that only authorized users can perform sensitive operations). Remote connection: After the permission verification is passed, the remote control module starts the FRP intranet penetration service, obtains the intranet IP and port of the PCS device from the MySQL table based on the device SN, and establishes an encrypted connection (using SSH encryption) between the cloud platform and the PCS device web backend. Engineers can access the device web backend through the front-end remote control interface to perform in-depth diagnostics (such as viewing device logs and operating parameters). Operation execution: Remote configuration: Engineers input the operating parameters to be modified (such as current threshold and sampling period) at the front end. The remote control module encapsulates the parameters into instructions (including instruction code, device SN, and parameter value) and sends them to the PCS device via the FRP connection. After the device executes the parameter modification, it returns the execution result. The system writes the operation record to the MySQL table and updates the device parameters synchronously. FOTA upgrade: Engineers upload firmware packages to the file storage module of the cloud platform (such as MinIO), the remote control module pushes upgrade instructions (including firmware package URL and upgrade version) to the PCS device, monitors the upgrade progress in real time (through the progress percentage reported by the device), and records the upgrade results (success / failure, reason for failure) in the table; after the upgrade is completed, the firmware version field of the table is updated. Status synchronization: After the remote operation is completed, the remote control module will synchronize the latest status of the device (parameters, firmware version) to the front-end display module and the storage layer to ensure data consistency. Scenario 4: Work Order Processing and Knowledge Base Accumulation Work order workflow: The work order system supports full-process management of "creation-assignment-processing-closure". Administrators assign work orders to maintenance engineers through the front end. After receiving the work order, the engineer processes it (such as on-site repair, remote debugging) and submits the processing result after processing. The system records the executor, processing time and operation content of each workflow node and stores them in the work order table of MySQL. Solution Entry: When an engineer closes a work order, they fill in the "Solution" (including the cause of the fault, handling steps, and preventive measures) on the front end. After the system verifies the completeness of the solution (using HibernateValidator), it associates it with the work order ID and stores it in the MySQL solution table. Knowledge Base Update: The knowledge base module periodically (e.g., every morning) extracts data from the solution table, performs structured processing according to "fault category" (e.g., hardware failure, parameter abnormality) and "root cause" (e.g., component aging, configuration error), generates knowledge base entries (including entry ID, fault category, solution, and associated work order ID), and stores them in the MySQL knowledge base table; simultaneously, a knowledge base index is built using Elasticsearch to optimize query speed. Intelligent Recommendation: When a new work order is generated, the intelligent recommendation module calls the interface of large models (such as Qianwen, Zhipu and other large domestic models) to parse the alarm information in the work order (such as device SN, fault phenomenon), and based on the Elasticsearch knowledge base to query similar cases (sorted by fault phenomenon matching degree), pushes the solutions of the top 3 similar cases to the processing engineer to improve the efficiency of work order processing.

[0053] Scenario 5: Health Assessment and Fault Prediction Data extraction: The intelligent analysis module extracts key component operating data (such as module temperature, capacitor switching count, and fan running time) from the TDengine table according to a preset cycle (such as early morning every day), obtains equipment model, operating years, and reserved fields for key parameters (such as component rated life and normal operating range) from the MySQL table. Health Assessment: The intelligent analysis module uses a weighted scoring method to calculate the equipment health status (SoH). Weights are based on component importance (e.g., module weight 0.3, capacitor weight 0.2, fan weight 0.1), with a scoring range of 0-100 (80-100 for Excellent, 60-79 for Good, 40-59 for Average, and 0-39 for Poor). A health assessment report is generated, including health score, abnormal components (e.g., modules with excessive temperature), and improvement suggestions, and stored in a MySQL health report table. Fault Prediction: The intelligent analysis module uses historical fault data from the same model of equipment (extracted from the "Fault Records" field and the work order table). It trains a fault prediction model using machine learning models (e.g., logistic regression, random forest). Inputting the current equipment's operating data and health status, it predicts the probability of failure (e.g., 30% probability of fan failure) and risk points (e.g., fan operating time approaching its rated life) within the next month. The prediction results are added to the health assessment report, supplementing fault prevention measures (e.g., replacing the fan in advance). Report Push: The intelligent analysis module pushes health assessment reports and fault prediction reports to the R&D personnel's email addresses (retrieved from the R&D personnel information table in MySQL) via an email gateway. Simultaneously, the reports are displayed on the front end, supporting the R&D team in product iteration (such as optimizing component selection and improving equipment design). Based on the above embodiments, a specific usage example of the system is as follows: A new energy operation and maintenance company (tenant A) manages 10 photovoltaic power stations with a total of 100 photovoltaic inverters (PCS devices). A power group (tenant B) manages 5 photovoltaic power stations with a total of 50 PCS devices of the same model. The two sets of devices need to be connected to the same operation and maintenance system to achieve data isolation, remote operation and maintenance, and intelligent alarm.

[0054] Specific steps: Step 1: Tenant Initialization and Device Access The system administrator creates tenant A (new energy operation and maintenance company) and tenant B (power group). The system automatically creates independent schemas (tenant_A, tenant_B) for the two tenants and initializes the device table, permission table, etc.

[0055] The maintenance personnel enter the SN, site information, and alarm threshold (such as the serious alarm triggered when the inverter temperature is ≥85℃) of 100 PCS devices of tenant A into the system front end. The system writes the device information into the device_info table of tenant_A. Similarly, the information of 50 devices of tenant B is entered into tenant_B.

[0056] All PCS devices are configured with MQTT parameters (connecting to the system MQTTBroker, SSL encrypted), and the sampling period is set to 2 seconds. They then begin collecting and uploading data such as temperature, current, and power generation.

[0057] Step 2: Data Acquisition and Isolated Storage Tenant A's PCS device collects data every 2 seconds, encapsulates it into a JSON message (containing tenant_id:A, device_sn:PCS-A-001, temperature:78℃), and transmits it to the MQTTBroker via MQTTTLS encryption.

[0058] After receiving the data, the processing layer extracts tenant_id=A, switches DynamicDatasource to the schema of tenant_A, writes the basic status of the device into MySQL, and writes the time series data (temperature, current) into TDengine by "month + day" partition (Tag includes tenant_id=A).

[0059] Tenant B's data flow is the same as A's, but the data is written to the tenant_B schema and the TDengine tenant_id=B tag partition to achieve physical / logical dual isolation.

[0060] Step 3: Alarm Triggering and Work Order Processing The temperature of PCS-A-056 equipment at a certain site of tenant A rose to 86℃, triggering a critical alarm when the alarm module polled TDengine data.

[0061] The system automatically generates a work order (Work Order ID: WO-A-20250501001) and pushes an alarm SMS to Zhang, the maintenance engineer of tenant A.

[0062] Xiao Zhang logs into the system (V2 privileges), views alarm details, and initiates a remote maintenance request. The system verifies Xiao Zhang's operating permissions (he can only operate on tenant A's devices). Once verified, an FRPSSH encrypted channel is established to connect to the web backend of PCS-A-056.

[0063] Xiao Zhang checked the equipment logs and found that the fan was faulty. He remotely issued a shutdown command and created a "Fan Replacement" sub-work order, which was then assigned to the site personnel.

[0064] After troubleshooting, Xiao Zhang filled in the solution in the system ("Fan bearing wear caused insufficient heat dissipation; replacing the fan restored normal operation") and closed the work order.

[0065] Step 4: Knowledge Accumulation and Health Assessment Every day at midnight, the system extracts nearly 30 days of operating data from 100 devices belonging to tenant A and calculates their health status using a weighted scoring method: 95 devices have a health status ≥ 80 points (excellent), and 5 devices (including PCS-A-056) have a health status of 65-79 points (good). The intelligent analysis module trains a random forest model based on historical fault data (including the current fan failure) to predict that tenant A has a ≥40% probability of fan failure in 3 devices in the next month, and generates a health assessment report which is pushed to tenant A's administrator.

[0066] The knowledge base module stores the solution to this fan failure in a structured manner. If tenant B experiences a similar alarm in the future, the system will automatically recommend this solution.

[0067] Data flow direction: PCS device (tenant A) → MQTTBroker (encryption) → Processing layer (tenant A schema switching) → MySQL (tenant_A) / TDengine (tenant_id=A) → Front-end display (visible to tenant A with permissions) Results: Data isolation: Tenant A cannot view the device data of Tenant B, and the same applies to Tenant B, ensuring zero data leakage.

[0068] Operational efficiency: From alarm triggering to remote handling completion, this fault took only 15 minutes, which is 87.5% more efficient than traditional on-site operation and maintenance (average 2 hours).

[0069] Resource utilization: With two tenants sharing a system instance, server resource utilization has increased from 30% in traditional independent deployments to 85%, and hardware costs have been reduced by 60%.

[0070] Intelligent assistance: Subsequent similar faults will be resolved by recommending solutions based on the knowledge base, reducing the processing time to 5 minutes.

[0071] Example 2: As a further replacement for the PCS device operation and maintenance management system based on Spring Boot multi-tenant and multi-data source provided in Embodiment 1, the specific solution includes: Multi-tenant isolation alternatives: In addition to the "shared database - independent schema" solution, the "independent database" solution can be adopted, which creates an independent MySQL database for each tenant and switches database connections through DynamicDatasource. This solution has stronger isolation and is suitable for scenarios with extremely high data security requirements (such as the financial industry), but it will increase the cost of database deployment and maintenance and is not convenient for cross-tenant data statistical analysis. Time-series data storage alternatives: In addition to TDengine, InfluxDB, iotDB, or Prometheus can be used as time-series databases. InfluxDB and iotDB support higher write throughput and are suitable for ultra-large-scale device data scenarios. Prometheus combined with Grafana can achieve richer time-series data visualization, but its long-term storage cost for historical data is higher than that of TDengine. The choice should be made based on the actual data scale and visualization needs. Remote maintenance technology alternatives: In addition to FRP intranet penetration, WebSocket or VPN technology can be used to achieve remote connection. WebSocket is suitable for lightweight remote operation (such as parameter viewing) and the connection establishment speed is fast, but it does not support complex device web backend access and requires an external IP address; VPN technology supports complete intranet access, but the configuration is complex and the tenant needs to install a VPN client, and its compatibility is lower than that of FRP. Fault prediction model alternative: In addition to machine learning models, rule engines (such as Drools) can be used to achieve fault prediction. Fault judgment rules are set based on expert experience (such as "fan running time > 5000 hours and temperature > 80℃ → fault probability 80%). This solution has a short development cycle and is easy to understand, but it has poor flexibility and cannot adapt to the dynamic changes in the operating status of the equipment. It is suitable for scenarios with a single type of equipment and fixed fault modes. Message queue alternatives: In addition to the MQTT protocol, Kafka can be used as a message queue. Kafka supports higher message throughput and partitioned storage, making it suitable for scenarios with millions of devices online simultaneously. However, it is less compatible with edge devices than MQTT (some small PCS devices only support the MQTT protocol), requiring the addition of a protocol conversion module (such as an MQTT-Kafka gateway) at the transport layer, which increases system complexity.

[0072] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A PCS device operation and maintenance management system based on Spring Boot multi-tenant multi-data source, characterized in that, include: The edge layer consists of several PCS devices used to collect device operation data; The transport layer includes MQTT brokers and FRP (Fiber Optic Regression Program) components; The processing layer, built on the Spring Boot framework, includes a multi-tenant management module, a data processing module, an alarm module, a remote control module, a work order system, and an intelligent analysis module. The storage layer adopts a hybrid database architecture, including relational databases, time-series databases, and cache databases; The application layer includes the front-end display module and the system monitoring module; The multi-tenant management module achieves tenant data isolation through a dynamic data source routing mechanism. Specifically, it creates an independent schema or database instance for each tenant in the relational database. When the processing layer receives a request, it parses the tenant identifier through a request interceptor and routes the database operation to the corresponding tenant data source based on the identifier using the dynamic data source framework.

2. The PCS equipment operation and maintenance management system based on Spring Boot multi-tenant multi-data source as described in claim 1, characterized in that, The dynamic data source routing mechanism includes: A tenant context holder stores the tenant identifier of the current request based on ThreadLocal; a request interceptor is used to extract and set the tenant identifier from the HTTP request to the tenant context holder. The dynamic data source router, which inherits from AbstractRoutingDataSource, uses the determineCurrentLookupKey method to obtain the tenant identifier as the data source lookup key from the tenant context holder.

3. The PCS equipment operation and maintenance management system based on Spring Boot multi-tenant multi-data source as described in claim 1, characterized in that, The remote control module integrates FRP intranet penetration service for: Receive remote access requests and perform permission verification; Based on the internal network address information of the target PCS device, dynamically start the FRP client process and establish an encrypted proxy channel between the public network port and the device's internal web port; Generate and return the corresponding remote access link.

4. The PCS equipment operation and maintenance management system based on Spring Boot multi-tenant multi-data source as described in claim 1, characterized in that, The alarm module, work order system, and knowledge base module form an automated operation and maintenance closed loop. The specific process is as follows: after the alarm is triggered, the alarm module automatically calls the work order system interface to create a maintenance work order. After the work order is processed, the submitted solution is stored in a structured manner in the knowledge base module; When a new alarm triggers a work order, the system calls the large model interface to match and recommend solutions for similar historical cases from the knowledge base based on the alarm information.

5. The PCS equipment operation and maintenance management system based on Spring Boot multi-tenant multi-data source as described in claim 1, characterized in that, The intelligent analysis module includes a health assessment unit, used for: Extract key component runtime timing data of the device from the time-series database according to a preset cycle; Based on a predefined weighting model, the operating status of each component is scored and the overall health score is calculated by weighting the scores. Generate a health assessment report that includes health level, identification of abnormal components, and optimization suggestions.

6. The PCS equipment operation and maintenance management system based on Spring Boot multi-tenant multi-data source as described in claim 1, characterized in that, The data processing module is equipped with out-of-order data processing logic to ensure that time-series data from the same device with inconsistent timestamp orders can be displayed in the correct time order at the application layer.

7. A PCS equipment operation and maintenance management system based on Spring Boot multi-tenant multi-data source as described in claim 1 or 2, characterized in that, The tenant identifier is derived from at least one of the following methods: tenant prefix information in the user login account, tenant identifier parameters carried in the HTTP request header, or tenant configuration information bound to the device serial number.

8. The PCS equipment operation and maintenance management system based on Spring Boot multi-tenant multi-data source as described in claim 1, characterized in that, While storing the parsed real-time device data into the time-series database, the data processing module also actively pushes the real-time data to the front-end display module via WebSocket or server event sending technology to achieve real-time refresh of the monitoring interface.

9. A PCS equipment operation and maintenance management system based on Spring Boot multi-tenant multi-data source as described in claim 4, characterized in that, The knowledge base module integrates the Elasticsearch search engine to index structured solutions, supporting fast matching and semantic retrieval.

10. A PCS equipment operation and maintenance management system based on Spring Boot multi-tenant multi-data source as described in claim 5, characterized in that, The intelligent analysis module also includes a fault prediction unit, which uses historical operating data and fault records of the equipment obtained from the relational database and time series database to train a machine learning model to predict the failure probability and potential risk components of the specified equipment in the future.

Images