Intelligent water meter reading platform and method based on high-precision internet map landmark positioning

Abstract

The invention belongs to the technical field of Internet of Things and geographic information, and particularly relates to an intelligent water meter reading platform and method based on high-precision Internet map landmark positioning. The data acquisition module is used for acquiring water meter reading and equipment running state data and acquiring original position data according to a preset positioning strategy; the data transmission network is used for providing a bidirectional wireless data communication link between the intelligent water meter terminal and the intelligent water meter reading platform; the invention discloses an intelligent water meter reading platform. The position information of the intelligent water meter can be accurately and reliably associated with a specific geographic landmark by transferring the positioning precision challenge to the powerful calculation and data processing capability of a platform side and combining a feedback mechanism of manual calibration; therefore, the core contradiction that reliable, accurate and automatic association between the water meter and the high-precision internet map landmarks cannot be achieved in the prior art is fundamentally solved, and a solid technical support is provided for refined operation of intelligent water affairs.

Description

Technical Field

[0001] This invention belongs to the fields of Internet of Things (IoT) technology and geographic information technology. Specifically, it is a smart water meter reading platform and method based on high-precision Internet map landmark positioning. Background Technology

[0002] With the development of global urbanization and smart cities, smart water meters and their supporting platforms have become an important support for the digital transformation of water management. They integrate multiple technologies to achieve automatic data collection, remote transmission and centralized management of water meters, and have achieved remarkable results in improving efficiency, reducing costs, improving services and assisting decision-making.

[0003] To achieve remote monitoring and asset management, existing smart water meters employ various methods to obtain location information. These include using a built-in GPS module to receive satellite signals, determine coordinates, and upload the data; or utilizing cellular network base stations and Wi-Fi fingerprint positioning to obtain coarser location information. These methods represent a significant improvement over traditional manual meter reading, enhancing the automation of data collection and management efficiency, and resolving the problems of time-consuming, labor-intensive, and error-prone manual meter reading.

[0004] However, with the changing requirements of urban development and management, existing technological solutions have gradually revealed their limitations. Satellite positioning or communication network-based positioning technologies face challenges in practical applications regarding accuracy, reliability, power consumption, and environmental adaptability. These problems are particularly pronounced when it is necessary to accurately correlate water meter data with specific landmarks on high-precision internet maps. Specifically: Satellite positioning suffers from decreased accuracy in areas with signal obstruction, and its high power consumption impacts the battery life of smart water meters. Low positioning accuracy in communication networks makes it difficult to accurately match specific landmarks, leading to positioning difficulties in management and hindering efficiency improvements. Furthermore, existing coordinate data is difficult to integrate into sophisticated geospatial information systems, impeding the deep fusion of smart water meter data with modern geospatial information systems and limiting landmark-based intelligent analysis and decision-making applications. This has become a bottleneck for the intelligent and sophisticated development of smart water meter reading platforms. Therefore, this paper proposes a smart water meter reading platform and method based on high-precision internet map landmark positioning. Summary of the Invention

[0005] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0006] The technical solution adopted by the present invention to solve its technical problem is: a smart water meter reading platform based on high-precision Internet map landmark positioning, comprising: a smart water meter terminal, a data transmission network, and a smart water meter reading platform.

[0007] The smart water meter terminal is configured on the water meter body and includes a data acquisition module, a multimodal positioning module, a wireless communication module, a local storage and processing module, and an energy management module. The data acquisition module periodically collects water meter readings and the smart water meter terminal's operational status data. The multimodal positioning module is configured to selectively or concurrently activate and collect at least one type of raw location data according to a preset positioning strategy, within a preset period or when a specific event is triggered. The wireless communication module is configured to upload water meter readings collected by the data acquisition module, equipment operating status data, and raw location data collected by the multimodal positioning module to the smart water meter reading platform via a data transmission network. The local storage and processing module is configured to temporarily store the collected data and perform preliminary filtering or preprocessing of the raw location data according to a preset algorithm. The energy management module is configured to precisely control the power supply and operating mode of the multimodal positioning module according to the positioning strategy, thereby achieving low-power operation of the smart water meter terminal.

[0008] The data transmission network configuration is used to provide a two-way wireless data communication link between the smart water meter terminal and the smart water meter reading platform. The link adopts a low-power wide area network technology protocol to ensure wide coverage, low power consumption and high reliability of data transmission.

[0009] The smart water meter reading platform is configured in a cloud computing environment or a local server cluster, and includes a data access layer, a data storage and management layer, a high-precision landmark data integration module, a multi-source positioning data fusion module, a landmark semantic association and optimization module, a visualization and management module, and an application service interface.

[0010] The data access layer is configured to receive water meter readings, equipment operating status data, and raw location data uploaded from the data transmission network, and to perform protocol parsing, data format conversion, and preliminary data verification.

[0011] The data storage and management layer is configured to persistently store water meter readings, equipment operating status data, raw location data, and a high-precision landmark knowledge graph constructed by the high-precision landmark data integration module. The storage structure includes, but is not limited to: a time-series database for time-series water meter readings and equipment status data; a relational database for raw location data and fused location data; and a graph database for complex topological relationships and semantic associations within the landmark knowledge graph.

[0012] The high-precision landmark data integration module is configured to: communicate with at least one external high-precision internet map service provider via a standard application programming interface to obtain and synchronize the latest geospatial data and landmark information; The system receives data from building information modeling systems, geographic information system databases, or manual on-site surveys via a data import interface. The data includes the precise coordinates and semantic descriptions of the building's internal structure, room layout, pipe routing, and preset water meter installation points. Based on the acquired and received data, a hierarchical, multi-dimensional landmark knowledge graph is constructed and maintained. The landmark knowledge graph associates physical geographic coordinates with semantically meaningful geographic entities and stores the hierarchical relationships, topological relationships, attribute information, and spatial geometric information between these entities.

[0013] The multi-source positioning data fusion module is configured to analyze, process, and fuse various raw location data from smart water meter terminals to generate a more accurate and reliable estimate of the water meter's physical location. The fusion module includes a navigation satellite system data processing unit, configured to receive and parse the NMEA protocol data stream provided by the GNSS receiver module, extract raw latitude and longitude, altitude, positioning accuracy factor (DOP value), number of satellites, and signal strength, and perform preliminary error correction and data smoothing. When the GNSS signal quality meets a preset threshold, it is used as the primary positioning input. The Wi-Fi fingerprint positioning unit is configured to receive and parse Wi-Fi AP identification information and its corresponding RSSI data uploaded by the Wi-Fi transceiver module. The unit pre-builds a Wi-Fi fingerprint database, which is established through offline surveying in a specific geographical area and stores the MAC addresses and corresponding RSSI distribution characteristics of multiple Wi-Fi APs at known geographical coordinates. The unit estimates the location of the smart water meter by comparing the real-time uploaded Wi-Fi fingerprints with the fingerprints in the database using a pattern matching algorithm. The Bluetooth Low Energy (BLE) beacon positioning unit is configured to receive and parse BLE beacon identification information and its corresponding RSSI data uploaded by the BLE transceiver module. The unit has a pre-built BLE beacon database storing identification information and signal propagation model parameters of BLE beacons with known deployment locations. The unit estimates the location of the smart water meter by using trilateration or a ranging algorithm based on the signal propagation model, combined with RSSI data from multiple beacons. The data fusion engine is configured to comprehensively consider the absolute accuracy of GNSS positioning, the local accuracy of Wi-Fi fingerprint positioning, and the indoor accuracy of BLE beacon positioning, along with their respective confidence levels, based on a preset fusion algorithm. It outputs a fused location estimate with higher accuracy and confidence, represented by geographic coordinates and associated error ellipses or covariance matrices. The fusion engine can dynamically adjust the weights of each positioning source according to different environmental conditions.

[0014] The landmark semantic association and optimization module is configured to match and associate the water meter physical location estimate output by the multi-source positioning data fusion module with a high-precision landmark knowledge graph, and continuously optimize it to generate a high-precision landmark positioning result with semantic meaning. The module includes a preliminary spatial matching unit, configured to receive the fused location estimate and use a spatial index structure to quickly retrieve and identify candidate landmark entities in the landmark knowledge graph that are spatially closest to or contain the fused location estimate. The matching process involves determining the geometric relationship between geographic coordinates and the spatial geometry of the landmark entity. The landmark constraint inference unit is configured to apply a series of predefined geographical and logical constraint rules to filter, verify, and optimize initially matched candidate landmarks. Geographical constraint rules include, but are not limited to: water meters are typically installed inside buildings, in pipe shafts, or in specific outdoor areas; there is a minimum physical distance between adjacent water meters; specific types of water meters typically correspond to specific installation location semantics. Logical constraint rules include, but are not limited to: each residential unit typically corresponds to only one main water meter; a landmark entity can only be associated with one unique smart water meter within a specific time period. Based on these rules, the inference unit eliminates unreasonable landmark associations and increases the confidence of the most likely landmarks. The historical trajectory and behavior pattern analysis unit is configured to analyze historical location data, historical meter reading data, and historical maintenance records of smart water meters. The unit uses machine learning algorithms to learn typical location state transition patterns, resident landmark characteristics, and abnormal location jump patterns of the water meters. When there is a deviation between the new fused location estimate and the historical landmark association, the unit will combine historical information to make trend judgments and corrections. The unit can also assist in judging whether the location has been moved or the data has been tampered with based on abnormal fluctuations in water meter readings, and trigger a reassessment of the landmark association. The manual calibration and feedback unit is configured to provide a human-computer interaction interface, allowing on-site maintenance personnel to accurately collect the real physical location of the water meter on-site using a handheld high-precision positioning device, and submit the accurate water meter ID-real coordinates-landmark semantic ID mapping relationship to the platform by intuitively selecting or inputting landmark semantic information on a high-precision internet map.

[0015] The unit prioritizes manually calibrated data as the highest-priority information, updating the water meter landmark association records in the landmark knowledge graph. It also utilizes this data to continuously optimize and retrain the parameters of the landmark constraint inference unit and the historical trajectory and behavior pattern analysis unit, forming a closed-loop feedback mechanism. The landmark semantic association and optimization module ultimately outputs a high-precision landmark location result, including the water meter's unique identifier, accurate geographic coordinates, and the most reliable semantic landmark description.

[0016] Visualization and Management Module Configuration: This module provides a graphical and intuitive display of the real-time location, device operating status, latest meter readings, and historical data curves of all smart water meters on a high-precision internet map. The display supports map zooming and panning operations. It supports multi-layer map overlay, including but not limited to: building floor plans, pipeline network diagrams, water meter installation point maps, and service area division maps. Users can choose to show or hide different layers as needed for multi-dimensional information analysis; It provides alarm information display and management functions. When the water meter positioning accuracy is lower than the preset threshold, there is ambiguity in the landmark association, or the equipment status is abnormal, the module generates a visual alarm and prompts the management personnel to perform manual verification or calibration. It provides asset management functions, supporting lifecycle management of smart water meter assets, including registration, deregistration, maintenance records, and replacement records, and associates them with high-precision landmark positioning results.

[0017] The application service interface configuration provides a standardized application programming interface, allowing external water management systems, water billing systems, fault reporting systems, and customer service platforms to access water meter data and high-precision landmark positioning results through the interface, thereby achieving data sharing and business collaboration.

[0018] To achieve the above-mentioned objectives, the present invention further provides a smart water meter reading method based on high-precision internet map landmark positioning, the method comprising the following steps: Step 1: Smart water meter initialization and landmark pre-binding; During the on-site installation of smart water meters, on-site staff use high-precision handheld terminals equipped with high-precision GNSS receivers, laser rangefinders, or visual recognition modules to accurately measure and collect the initial geographic coordinates of the smart water meter terminals. On-site staff can intuitively select or scan an NFC / QR code pre-installed on the water meter or its installation location using the interactive interface of a high-precision handheld terminal on a high-precision internet map. This binds the smart water meter's unique identifier to a specific semantic landmark entity in the high-precision internet map's landmark knowledge graph. The NFC / QR code pre-stores a landmark semantic ID that uniquely corresponds to the landmark entity in the landmark knowledge graph. The high-precision handheld terminal uploads the water meter ID, precise initial geographic coordinates, landmark semantic ID, and other installation context information to the data access layer of the smart water meter reading platform via a wireless communication network. The landmark semantic association and optimization module of the smart water meter reading platform receives and processes the initial binding data, stores it as high-priority information in the landmark knowledge graph, establishes a mapping relationship between smart water meter ID, initial precise coordinates and landmark semantic ID, and updates the landmark knowledge graph. The smart water meter terminal locally stores the landmark semantic ID and related initial configuration parameters.

[0019] Step 2: Periodic Data Collection and Transmission of Smart Water Meters: The data collection module of the smart water meter terminal automatically collects the current water meter reading and the device operation status data of the smart water meter terminal itself within the preset meter reading cycle; The multimodal positioning module of the smart water meter terminal selectively or in parallel activates corresponding positioning units to collect raw location data based on a preset low-power positioning strategy and environmental conditions. The strategy includes: a. When external GNSS signals are available, the GNSS receiver module is activated first to receive satellite signals and calculate the current geographic coordinates. To save energy, the GNSS receiver module can be activated for a short time, acquire positioning data, and then immediately enter a sleep state; b. When the GNSS signal is unavailable or its quality is below a preset threshold, the multimodal positioning module switches to Wi-Fi scanning mode, activates the Wi-Fi transceiver module, scans the surrounding Wi-Fi access points, and collects the MAC address and corresponding RSSI value of each access point. c. When Wi-Fi signal is unavailable or the assisted positioning effect is poor, or as a supplement to Wi-Fi positioning, the multimodal positioning module activates the BLE transceiver module, scans the surrounding BLE beacons, and collects the UUID / Major / Minor identifier of each beacon and its corresponding RSSI value. d. The multimodal positioning module can dynamically adjust the activation frequency and priority of different positioning modes according to the actual environment and energy budget. For example, it can only perform Wi-Fi / BLE scanning when the battery is low and try GNSS positioning when the battery is high. The wireless communication module of the smart water meter terminal will collect water meter readings, device operating status data, and raw location data, including but not limited to GNSS coordinates, Wi-Fi fingerprint data, and BLE beacon data, and upload them to the data access layer of the smart water meter reading platform through the data transmission network.

[0020] Step 3: Multi-source positioning data fusion and landmark association on the platform: The data access layer of the smart water meter reading platform receives and parses various types of data uploaded from smart water meter terminals; The multi-source positioning data fusion module processes and fuses the received raw location data.

[0021] a. If GNSS data is received, the navigation satellite system data processing unit performs preliminary verification and smoothing processing on it, and calculates its position accuracy factor; b. If Wi-Fi fingerprint data is received, the Wi-Fi fingerprint positioning unit uses the Wi-Fi fingerprint database and a pattern matching algorithm to estimate the approximate location of the smart water meter; c. If BLE beacon data is received, the Bluetooth Low Energy beacon positioning unit uses the BLE beacon database and ranging algorithm to estimate the approximate location of the smart water meter; d. The data fusion engine comprehensively analyzes positioning data from different positioning sources, along with their accuracy and confidence levels, and uses a fusion algorithm to generate an optimized water meter physical location estimate with higher accuracy and confidence.

[0022] The landmark semantic association and optimization module receives the fused physical location estimate of the water meter and performs landmark semantic association and optimization operations: a. The preliminary spatial matching unit uses the spatial index of the landmark knowledge graph to quickly identify multiple candidate landmark entities that are spatially closest to or contain the fused location estimate; b. The landmark constraint reasoning unit calls the preset geographical constraint rules and logical constraint rules to evaluate the candidate landmark entities, exclude landmarks that do not meet the rules, and adjust the confidence of the remaining candidate landmarks according to the degree of compliance with the rules. c. The historical trajectory and behavior pattern analysis unit combines the water meter's historical location records, historical landmark associations, and behavior patterns in the water meter reading data to further revise and optimize the current candidate landmarks; d. If, after the above steps, there are still multiple candidate landmarks or the landmark confidence level is lower than the preset threshold, the landmark semantic association and optimization module can further request manual assistance for calibration.

[0023] The landmark semantic association and optimization module ultimately outputs a precise and semantically accurate landmark positioning result, which includes the smart water meter's ID, the determined geographic coordinates, and the corresponding semantic landmark description.

[0024] Step 4: Platform Data Management and Application: The data storage and management layer of the smart water meter reading platform will associate and store and update the smart water meter ID, the latest water meter reading, equipment operation status data, and high-precision landmark positioning results; The visualization and management module displays the location, real-time data, and device status of smart water meters in an intuitive graphical way on a high-precision internet map, based on high-precision landmark positioning results. When the landmark semantic association and optimization module determines that the positioning accuracy is insufficient, the landmark association is ambiguous, or it is seriously inconsistent with historical data, the visualization and management module generates alarm information and sends it to the management personnel terminal, prompting manual verification or triggering the manual auxiliary calibration and feedback unit. The platform receives calibration data submitted by the manual calibration and feedback unit, and uses this data to update the water meter landmark association information in the landmark knowledge graph, and further optimizes the model parameters of the landmark constraint reasoning unit and the historical trajectory and behavior pattern analysis unit, forming a continuous optimization cycle; The application service interface provides water meter readings, device status, and historical data that are accurately linked to the landmark based on requests from external systems.

[0025] The beneficial effects of this invention are as follows: Through the above technical solution, this invention integrates multimodal positioning technology, high-precision internet map data, and intelligent semantic association algorithms at the system level, achieving precise perception and landmark-level management of smart water meter locations. This solution avoids the problems of insufficient terminal positioning accuracy and susceptibility to environmental influences inherent in existing technologies. By transferring the challenge of positioning accuracy to the powerful computing and data processing capabilities of the platform, and combining this with a feedback mechanism for manual calibration, the location information of smart water meters can be accurately and reliably associated with specific geographical landmarks. This fundamentally solves the core contradiction of existing technologies in achieving reliable, accurate, and automated association between water meters and high-precision internet map landmarks, providing solid technical support for the refined operation of smart water management. Attached Figure Description

[0026] The invention will now be further described with reference to the accompanying drawings.

[0027] Figure 1 This is a structural block diagram of the smart water meter reading platform in this invention; Figure 2 This is a schematic diagram of the smart water meter reading method in this invention.

[0028] In the diagram: 10. Smart water meter terminal; 11. Data acquisition module; 12. Multimodal positioning module; 13. Wireless communication module; 14. Local storage and processing module; 15. Energy management module; 20. Data transmission network; 30. Smart water meter reading platform; 31. Data access layer; 32. Data storage and management layer; 33. High-precision landmark data integration module; 34. Multi-source positioning data fusion module; 341. Navigation satellite system data processing unit; 342. Wi-Fi fingerprint positioning unit; 343. Bluetooth low-power beacon positioning unit; 344. Data fusion engine; 35. Landmark semantic association and optimization module; 351. Preliminary spatial matching unit; 352. Landmark constraint reasoning unit; 353. Historical trajectory and behavior pattern analysis unit; 354. Manual assisted calibration and feedback unit; 36. Visualization and management module; 37. Application service interface. Detailed Implementation

[0029] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0030] This invention provides a smart water meter reading platform based on high-precision internet map landmark positioning, including: a smart water meter terminal 10, a data transmission network 20, and a smart water meter reading platform 30; The smart water meter terminal 10 is the core edge-side device of this invention. Its physical design is intended to adapt to harsh outdoor and semi-outdoor installation environments. It typically employs an integrated encapsulation structure that meets IP68 protection standards, ensuring stable operation under long-term immersion in water, dust, and extreme temperature and humidity conditions. Its outer shell can be made of high-performance engineering plastics, such as polycarbonate (PC) or modified polyphenylene ether (PPE), while the interior features multiple integrally molded reinforcing ribs to enhance structural rigidity and a platform providing mounting surfaces for internal electronic modules. The terminal is powered by an integrated lithium thionyl chloride battery or a multi-cell series-parallel lithium-ion battery pack, with battery capacity typically designed to support maintenance-free operation for more than ten years.

[0031] The smart water meter terminal 10 is internally equipped with a data acquisition module 11, a multimodal positioning module 12, a wireless communication module 13, a local storage and processing module 14, and an energy management module 15. The core of the data acquisition module 11 is an integrated high-precision flow sensor, such as a metering sensor based on ultrasonic principles or the Hall effect. This sensor can periodically collect water meter readings, including cumulative flow and instantaneous flow values, and its metering accuracy meets relevant national metrological standards (such as OIMLR49 or GB / T778). Simultaneously, this module also integrates multiple auxiliary sensors for collecting equipment operating status data of the smart water meter terminal. These include monitoring ambient and water flow temperatures via a built-in temperature sensor, monitoring battery power via a voltage detection circuit, detecting abnormal disassembly or strong magnetic interference (i.e., tamper alarm status) via a magnetic sensor or accelerometer, and providing feedback on valve status (such as open / closed status, abnormal jamming) via a valve control circuit. All collected data is sampled at a preset frequency (such as every 15 minutes, every hour, or at fixed times daily).

[0032] The multimodal positioning module 12 is a key component for achieving high-precision positioning in the smart water meter terminal 10. It is configured to selectively or concurrently activate and collect at least one type of raw location data according to a preset positioning strategy within a preset period or when triggered by a specific event. This module integrates a Global Navigation Satellite System (GNSS) receiver module, such as the U-blox M8N series high-precision positioning chip, supporting multiple constellations including GPS, GLONASS, BeiDou (BDS), and Galileo. Its cold start time is typically less than 25 seconds, and its warm start time is less than 1 second. It also supports A-GNSS functionality to accelerate the time to first positioning (TTFF) and improve positioning performance in weak signal environments. Simultaneously, this module also includes a Wi-Fi transceiver module for capturing Wi-Fi signals, such as the ESP32 series module, which has low-power Wi-Fi scanning capabilities and can identify the BSSID / SSID of Wi-Fi access points (APs) in the surrounding environment, along with their corresponding Received Signal Strength Indication (RSSI) data. In addition, to enhance positioning capabilities in indoor and underground environments, this module also integrates a Bluetooth Low Energy (BLE) transceiver module, such as one based on the Nordic NRF52 series chip, supporting the BLE 5.0 standard. This module can scan and identify the UUID, Major, and Minor identifiers of surrounding Bluetooth beacons (such as iBeacon or Eddystone protocol beacons) and collect their corresponding RSSI data. The multimodal positioning module's positioning strategy is carefully designed to balance positioning accuracy and power consumption requirements. In a typical implementation, the terminal is in deep sleep mode by default, waking up every 2 hours. Upon waking, it first attempts to activate the GNSS receiver module for a short period (e.g., 5-10 seconds) of satellite signal acquisition and positioning. If the GNSS signal strength (C / N0) and positioning accuracy factor (DOP value, e.g., HDOP < 2.5) meet preset thresholds, the GNSS positioning result is used as the primary location data. When GNSS signals are unavailable, of poor quality, or cannot be located for an extended period (e.g., inside buildings, underground pipe shafts, or in urban canyons where GNSS signals are limited), the positioning module will automatically switch to Wi-Fi scanning mode, quickly scanning for surrounding Wi-Fi access points (APs) and collecting fingerprint data within seconds. If the Wi-Fi scan fails to obtain sufficient AP information (e.g., fewer than 3 independent APs), or if BLE beacons are pre-deployed in a specific indoor area, the BLE transceiver module will be further activated to perform beacon scanning and collect beacon data. This strategy can be dynamically adjusted based on battery power and environmental conditions; when the battery power is below a certain threshold, Wi-Fi / BLE scanning, which consumes less power, will be prioritized over GNSS positioning.

[0033] The wireless communication module 13 is responsible for uploading the water meter readings collected by the data acquisition module 11, the equipment operating status data, and the raw location data collected by the multimodal positioning module 12 to the smart water meter reading platform 30 via the data transmission network 20. This module typically integrates a low-power wide-area network (LPWAN) communication chip, such as the Huawei Boudica or Quectel BC95 module to support the Narrowband Internet of Things (NB-IoT) protocol, or the STMicroelectronics STM32WL series chip to support the LoRaWAN protocol. NB-IoT technology utilizes licensed frequency bands in cellular networks, featuring wide coverage, low power consumption, large connectivity, and low cost. Its communication rate is typically tens to hundreds of Kbps, sufficient to meet the small data packet and low-frequency data transmission needs of smart water meters. LoRaWAN technology uses unlicensed frequency bands and achieves long-distance communication and anti-interference capabilities through frequency hopping spread spectrum technology, also suitable for IoT applications. Both technologies support bidirectional communication links, enabling the platform to issue commands to the smart water meter terminal, such as remote valve control, meter reading cycle adjustment, or firmware upgrades. To ensure communication reliability, this module is typically equipped with a high-gain antenna and supports data encryption and integrity verification mechanisms.

[0034] The local storage and processing module 14 integrates a high-performance, low-power microcontroller (MCU), such as the STM32 series or NXP Pinetis series chips based on the ARM Cortex-M4 core. It is equipped with sufficient flash memory (e.g., 256KB-1MB) for storing firmware, configuration parameters, historical data, and a local cache for the Wi-Fi / BLE fingerprint database, while also having sufficient RAM (e.g., 64KB-256KB) for runtime data processing. This module is configured to temporarily store the raw data acquired by the data acquisition module 11 and the multimodal positioning module 12, and to perform preliminary filtering or preprocessing of the raw location data according to a preset algorithm. For example, for GNSS data, a simple moving average filter can be performed to remove abnormal jump points; for Wi-Fi / BLE SCI data, averaging can be performed to reduce the impact of signal fluctuations. These preliminary processing steps aim to reduce the amount of data and improve data quality, laying the foundation for subsequent complex processing on the platform side.

[0035] The energy management module 15 is the core guarantee for the low-power operation of the smart water meter terminal 10. It maximizes battery life by precisely controlling the power supply and operating mode of the multimodal positioning module 12 and other energy-consuming components. This module includes a high-efficiency DC-DC converter to convert the battery voltage to the operating voltage required by each module, and a low-power power management IC (PMIC) for smooth switching between different power consumption modes (such as normal operation, idle, sleep, and deep sleep). It precisely manages the power supply sequence of each module based on positioning strategies, communication requirements, and system events (such as timed wake-up and alarm triggering), ensuring that power is supplied to the corresponding modules only when needed and that they immediately enter a low-power state after the task is completed. For example, the GNSS receiver will immediately power off or enter the lowest power mode after completing positioning, and the Wi-Fi / BLE module will also quickly enter sleep mode after scanning.

[0036] The data transmission network 20 provides a two-way wireless data communication link between the smart water meter terminal 10 and the smart water meter reading platform 30. This network employs Low Power Wide Area Network (LPWAN) technology protocols, such as the aforementioned NB-IoT or LoRaWAN protocols. The NB-IoT network relies on existing cellular base stations for coverage, offering good penetration and wide coverage in urban areas. A single base station can support tens of thousands of terminal connections, with data transmission latency typically in the order of seconds. The LoRaWAN network aggregates and forwards data through LoRa gateways. Its long-distance communication capability (up to several kilometers in urban environments and tens of kilometers in open areas) makes it suitable for deployment in remote or poorly covered areas. Both technologies achieve extremely low terminal power consumption through optimized protocol stack and physical layer design, enabling the smart water meter terminal to operate for extended periods without charging or battery replacement.

[0037] The smart water meter reading platform 30 is configured in a cloud computing environment or a local server cluster, possessing powerful data processing, storage, and analysis capabilities. In a typical implementation, the platform is deployed on mainstream cloud service providers (such as Alibaba Cloud, Tencent Cloud, or Huawei Cloud), employing a microservice architecture and containerization technologies (such as Docker and Kubernetes) to achieve high availability, scalability, and elasticity. The overall platform structure includes a data access layer 31, a data storage and management layer 32, a high-precision landmark data integration module 33, a multi-source positioning data fusion module 34, a landmark semantic association and optimization module 35, a visualization and management module 36, and an application service interface 37.

[0038] The data access layer 31 serves as the platform's entry point, configured to receive water meter readings, device operating status data, and raw location data uploaded from the data transmission network 20. This layer typically consists of an API Gateway and a message queue service. The API Gateway handles security authentication, flow control, and request routing from the endpoints. The received raw data undergoes protocol parsing, converting LPWAN-specific binary data packets into standardized data formats (such as JSON or Protobuf) and performing data format conversion. A preliminary data verification unit performs rapid checks on data integrity (such as checksums) and validity (such as timestamps, data types, and numerical ranges), filtering out obviously erroneous or corrupted data packets to ensure the quality of data processed subsequently.

[0039] The data storage and management layer 32 is responsible for storing all data received and processed by the persistent storage platform. To optimize storage performance and query efficiency, this layer employs a hybrid storage strategy using multiple database types. For time-series water meter readings and device status data (such as hourly flow data and 15-minute battery voltage data), a time-series database (such as TimescaleDB or InfluxDB) is configured. These databases are optimized for writing and querying time-series data, enabling efficient storage and retrieval of large-scale continuous data. For raw positioning data (such as raw GNSS ephemeris and Wi-Fi / BLE scan logs) and fused location data generated by the multi-source positioning data fusion module 34 (such as precise latitude, longitude, altitude, and location accuracy estimates), a relational database (such as PostgreSQL or MySQL) is used for structured querying and management. For the high-precision landmark knowledge graph constructed by the high-precision landmark data integration module 33, this layer configures a graph database (such as Neo4j or JanusGraph) to store and query complex topological relationships and semantic associations, capable of representing and manipulating hierarchical, proximity, and attribute relationships between landmark entities. All databases are configured with redundant backup and disaster recovery mechanisms to ensure data security and availability.

[0040] The high-precision landmark data integration module 33 is the foundation for building landmark-level positioning capabilities in this invention. The configuration of this module consists of three main aspects. First, it communicates with at least one external high-precision internet map service provider via a standard application programming interface (API) to obtain and synchronize the latest geospatial data and landmark information. These service providers can be mainstream commercial map service providers, such as Gaode Maps, Baidu Maps, or Google Maps Platform, which provide rich geospatial information, including but not limited to: accurate geocoding and reverse geocoding services, massive amounts of Points of Interest (POI) data (such as shops and public facilities), detailed building outlines (including polygon geometry), floor plans (partially provided), and standardized house number information. By regularly calling these APIs, the platform can obtain the latest changes in urban geographic information and keep the landmark data up-to-date. Second, it receives data from Building Information Modeling (BIM) systems, Geographic Information System (GIS) databases, or manual field survey data through a data import interface. BIM systems (such as Revit and ArchiCAD) can provide extremely detailed 3D model data of building interiors, including room layouts, wall structures, pipe routing (such as water supply and drainage pipes, HVAC pipes), and precise 3D coordinates and rich semantic descriptions of preset water meter installation points (e.g., "cold water meter under the kitchen sink," "water meter next to the fire hydrant in the corridor"). GIS databases (such as ESR and ArcGIS) provide more macroscopic spatial data on urban infrastructure. Manual field survey data, collected using handheld high-precision RTK devices, can supplement and correct the deficiencies of the aforementioned automated data sources, providing the most realistic installation point data on site. Third, based on the acquired and received data, a hierarchical, multi-dimensional landmark knowledge graph is constructed and maintained. The landmark knowledge graph is based on ontology, deeply associating physical geographic coordinates with semantically meaningful geographic entities. These entities include, but are not limited to: country, province, city, district, street, community, building, floor, unit, room number, and ultimately, the specific smart water meter installation point. The landmark knowledge graph stores hierarchical relationships (e.g., "unit" belongs to "floor", "floor" belongs to "building"), topological relationships (e.g., "room A" is adjacent to "room B"), attribute information (e.g., the year the building was constructed, the purpose of the room, and the characteristic description of the location), and spatial geometric information (e.g., boundary polygons, center point coordinates, and elevation information). To improve query efficiency, this spatial geometric information is usually stored in conjunction with spatial indexes (e.g., R-trees or Quadtrees). Furthermore, the landmark knowledge graph further stores historical water meter installation records, maintenance logs, and calibration data related to each landmark entity, forming a rich, structured, and spatial knowledge base.

[0041] The multi-source positioning data fusion module 34 is the core intelligent component for achieving high-precision water meter positioning on the platform side. It is configured to analyze, process, and fuse various raw location data from the smart water meter terminal 10 to generate a more accurate and reliable estimate of the water meter's physical location. The fusion module includes a navigation satellite system data processing unit 341, a Wi-Fi fingerprint positioning unit 342, a Bluetooth Low Energy beacon positioning unit 343, and a data fusion engine 344. The navigation satellite system data processing unit 341 is configured to receive and parse the NMEA protocol data stream provided by the GNSS receiver module, extracting the raw latitude and longitude, altitude, positioning accuracy factor (DOP value, such as HDOP, VDOP, PDOP), the number of captured satellites, and the signal strength (C / N0) of each satellite. This unit performs preliminary error correction and data smoothing, such as fitting multiple sets of consecutive GNSS readings using the least squares method or applying Kalman filtering to remove measurement noise. Simultaneously, it also assesses the current confidence level of the GNSS positioning based on the DOP value and the number of satellites. When the GNSS signal quality meets the preset threshold (e.g., HDOP < 2.0, number of visible satellites > 6), the GNSS positioning result will be given a higher weight and used as the main positioning input.

[0042] The Wi-Fi fingerprint positioning unit 342 is configured to receive and parse the Wi-Fi AP identification information (MAC address / BSSID) and its corresponding RSSI data uploaded by the Wi-Fi transceiver module. This unit pre-builds a large Wi-Fi fingerprint database, established through offline surveys in specific geographical areas (such as inside high-rise buildings, underground parking lots, or areas with high concentrations of water meters). During the survey, personnel use specialized equipment to scan and record the MAC addresses of all detectable Wi-Fi APs and their corresponding RSSI distribution characteristics (such as mean and standard deviation) at numerous known, precise geographical coordinate points (determined via RTK-GNSS or laser rangefinders). This database stores these "location-AP fingerprint" mappings. During real-time positioning, the unit compares the real-time Wi-Fi fingerprints uploaded by the smart water meter terminal (i.e., the currently scanned AP list and their RSSI values) with the fingerprints in the database, using a pattern matching algorithm to estimate the smart water meter's location. Commonly used algorithms include K-Nearest Neighbors (K-NN), Support Vector Machine (SVM), or deep learning-based neural network models. For example, the K-NN algorithm calculates the Euclidean distance between the real-time fingerprint and all fingerprints in the database, selects the K closest fingerprints, and performs a weighted average of their corresponding geographic coordinates. The weights are typically related to the reciprocal of the distance. This achieves line-of-sight-independent propagation and is suitable for indoor environments.

[0043] The Bluetooth Low Energy (BLE) beacon positioning unit 343 is configured to receive and parse BLE beacon identification information (UUID / Major / Minor) and its corresponding RSSI data uploaded by the BLE transceiver module. Similar to a Wi-Fi fingerprint database, this unit pre-builds a BLE beacon database, storing identification information of BLE beacons with known deployment locations and calibrated signal propagation model parameters (such as path loss index and environmental factors). Beacons are typically deployed near the water meter installation point, such as inside the pipe well or inside the equipment box. The unit estimates the location of the smart water meter by combining RSSI data from multiple beacons using trilateration or a ranging algorithm based on the signal propagation model. For example, the Log-DistancePathLoss model can convert RSSI values ​​into estimated distances, and then cross-reference the distance information from at least three beacons to determine the water meter location. This method provides high positioning accuracy in small indoor environments (such as rooms or floors).

[0044] The data fusion engine 344 is the core of the multi-source positioning data fusion module 34. It is configured to comprehensively consider the absolute accuracy of GNSS positioning, the local accuracy of Wi-Fi fingerprint positioning, and the indoor accuracy and confidence level of BLE beacon positioning based on a preset fusion algorithm, outputting a fused location estimate with higher accuracy and confidence. Commonly used fusion algorithms include Extended Kalman Filter (EKF), Particle Filter, or Bayesian Network. Taking EKF as an example, its state vector is typically defined as the geographic coordinates (longitude, latitude, and altitude) and velocity of the water meter, predicted through a nonlinear motion model. When new GNSS, Wi-Fi, or BLE positioning measurements are received, the fusion engine updates the state vector based on the independent error covariance matrix of each measurement source (reflecting its accuracy and confidence level). This fusion engine can dynamically adjust the weights of each positioning source according to different environmental conditions (indoor / outdoor, signal obstruction, satellite visibility, Wi-Fi AP density). In open outdoor areas, GNSS positioning is given the highest weight; indoors, the weight of Wi-Fi fingerprint positioning or BLE beacon positioning will be significantly increased. The final output fused location estimate is represented by geographic coordinates and associated error ellipses or covariance matrices, used to quantify the uncertainty of the location estimate.

[0045] The landmark semantic association and optimization module 35 is a key innovation of this invention in achieving accurate association between water meters and landmarks. Its configuration is used to match and associate the water meter physical location estimate output by the multi-source positioning data fusion module 34 with a high-precision landmark knowledge graph, and to continuously optimize it to generate a high-precision landmark positioning result with semantic meaning. This module includes a preliminary spatial matching unit 351, a landmark constraint reasoning unit 352, a historical trajectory and behavior pattern analysis unit 353, and a manual-assisted calibration and feedback unit 354. The preliminary spatial matching unit 351 is configured to receive the fused location estimate and, using a preset spatial index structure (R-tree, KD-tree) in the landmark knowledge graph, quickly retrieve and identify candidate landmark entities that are spatially closest to or contain the fused location estimate from a massive number of landmark entities. This matching process involves complex judgments of the geometric relationship between geographic coordinates and the spatial geometry of landmark entities (building polygons, unit areas, room polygons), such as point-in-polygon testing and nearest neighbor search. The output is a set of candidate landmark entities and their distances or overlap with the merged location.

[0046] The Landmark Constraint Inference Unit 352 is configured to apply a series of predefined geographical and logical constraint rules to screen, verify, and optimize initially matched candidate landmarks. These rules aim to simulate common-sense human judgment regarding water meter installation locations. Geographical constraint rules include, but are not limited to: water meters are typically installed inside buildings (pipe shafts, kitchens, bathrooms), in common areas on floors, or in specific outdoor areas (water meter boxes at the main entrance of the community); there is a minimum physical distance between adjacent water meters (less than 50cm indicates that the same water meter has been detected multiple times or that the installation locations are very close); specific types of water meters (master meters, sub-meters) typically correspond to specific installation location semantics (master meters should be installed before entering the household, sub-meters at different water usage points). Logical constraint rules include, but are not limited to: each residential unit typically corresponds to only one master water meter at a time; a landmark entity (a room) can only be associated with one unique smart water meter within a specific time period; the physical location of a water meter is typically fixed after installation. Based on these rules, the inference unit calculates a confidence score for each candidate landmark entity through fuzzy logic inference or a rule-based expert system, eliminating unreasonable landmark associations and increasing the confidence of the most likely landmark. If a fusion location falls within the geometry of room A, but the water meter's history shows it belongs to room B, and room A is occupied by other water meters, then the confidence level of room A as a candidate landmark will be significantly reduced.

[0047] The historical trajectory and behavior pattern analysis unit 353 is configured to analyze the historical location data, historical meter reading data, and historical maintenance records of smart water meters. This unit uses machine learning algorithms (Hidden Markov Model (HMM), Recurrent Neural Network (RNN), and especially Long Short-Term Memory (LSTM) networks) to learn typical location state transition patterns, resident landmark characteristics, and abnormal location jump patterns of the water meter. The HMM model defines the water meter's location state as a series of discrete landmark entities, predicting the most likely landmark state by learning state transition probabilities and observation probabilities (i.e., the probability of observing a specific location result at a specific landmark). The RNN model can handle long-term series data, learning the water meter's location drift, dwell patterns, and historical associations with other landmarks over different time periods. When a new fused location estimate deviates from the historical landmark association, the unit combines historical information for trend judgment and correction. For example, if a water meter has been stably associated with a landmark ("Unit XX, Room XX") for a long time (months), and suddenly experiences a brief location drift (the fused location jumps to the next room), but its range is small and it recovers quickly, the unit tends to maintain the original historical landmark association and considers the current deviation as noise. Conversely, if persistent and widespread location drift occurs, it will trigger a reassessment of landmark associations. The unit can also help determine whether there is a possibility of location relocation or data tampering based on abnormal fluctuations in water meter readings (long-term zero readings or sudden high readings, which may be related to pipe rupture, water theft, or water meter movement), and trigger a reassessment of landmark associations, thereby achieving cross-validation of multi-dimensional data.

[0048] The manual calibration and feedback unit 354 is an important component of the closed-loop optimization mechanism of this invention. It is configured to provide a human-computer interaction interface, allowing on-site maintenance personnel to accurately collect the real physical location of the water meter on-site (up to centimeter-level accuracy) using a handheld high-precision positioning device (a Trimble R12i or Leica GS18 mobile station supporting RTK-GNSS technology, a laser rangefinder, or a mobile terminal with visual positioning / SLAM function). By intuitively selecting or inputting landmark semantic information on a high-precision internet map (scanning an NFC / QR code preset on the water meter or installation location, which is pre-bound to a landmark semantic ID in the landmark knowledge graph, such as the UUID of the landmark entity), the precise mapping relationship of "water meter ID - real coordinates - landmark semantic ID" and contextual information such as on-site photos and text descriptions are submitted to the platform. The unit prioritizes manually calibrated data as the highest priority information, immediately updates the water meter landmark association records in the landmark knowledge graph, and uses the manually calibrated data to continuously optimize the parameters of the landmark constraint reasoning unit 352 and the historical trajectory and behavior pattern analysis unit 353 and retrain the model. For example, through supervised learning methods, it uses manually calibrated real landmarks as label data to continuously improve the model's ability to identify and judge complex scenes, forming a positive closed-loop feedback mechanism.

[0049] The landmark semantic association and optimization module 35 ultimately outputs a high-precision landmark location result, which includes the water meter's unique identifier (ID), precise geographic coordinates (usually the optimal location after fusion, if available), and the most reliable semantic landmark description ("the water meter in apartment XX, unit XX, building XX, XX community"). This result is more business-valued and readable than simple geographic coordinates.

[0050] The Visualization and Management Module 36 is configured to provide water management personnel with an intuitive and efficient operational view. First, this module graphically displays the real-time location of all smart water meters, their operating status (battery percentage, signal strength in dBm, valve status, alarm type), latest meter readings, and historical data curves (daily, weekly, and monthly water consumption) on a high-precision internet map. The display supports stepless zooming and panning of the map and can differentiate water meters based on their status using different colors or icons. Second, this module supports multi-layered map overlays, including but not limited to: detailed building floor plans (showing walls and room layouts), underground pipeline network maps (showing main and branch water supply pipes), water meter installation point maps (precisely marking the location of each water meter), and service area division maps. Users can choose to show or hide different layers as needed for multi-dimensional information analysis; for example, overlaying the pipeline network map can quickly locate the upstream and downstream relationships of water meter anomalies within a specific area. Third, this module provides alarm information display and management functions. When the water meter's positioning accuracy falls below a preset threshold (e.g., the area of ​​the error ellipse in the fused location estimation is too large), there is ambiguity in landmark association (e.g., multiple candidate landmarks have similar confidence scores), or the device status is abnormal (e.g., low battery power, signal interruption, water leakage alarm), the module generates a visual alarm, prominently displayed on the map (blinking icon, red marker), and can automatically send SMS, email, or App push notifications to management personnel, prompting them to conduct manual verification or calibration. Fourth, the module provides asset management functions, supporting full lifecycle management of smart water meter assets, including registration, deregistration, maintenance records, replacement records, and inspection records, and closely linking them with high-precision landmark positioning results. All asset information is traceable and can be quickly located and retrieved through landmarks.

[0051] Application Service Interface 37 is configured to provide standardized Application Programming Interfaces (APIs), allowing external water management systems, water billing systems, fault reporting systems, and customer service platforms to access water meter data and high-precision landmark location results through these interfaces, enabling data sharing and business collaboration. These APIs adhere to the RESTful design style, support JSON data format, and employ industry standards such as OAuth2 for authentication and authorization, ensuring the security and controllability of data access. The water billing system can obtain historical water consumption and current water meter readings for specified users through the APIs; the fault reporting system can obtain the precise location and landmark information of abnormal water meters, guiding maintenance personnel to quickly reach the site; and the customer service platform can provide users with personalized water consumption reports and water meter status query services.

[0052] Furthermore, this invention also provides a smart water meter reading method based on high-precision internet map landmark positioning, the process of which is as follows: Figure 2 As shown, it includes the following steps: Smart water meter initialization and landmark pre-binding: This step marks the beginning of the water meter's lifecycle, aiming to establish a precise initial location and landmark association for each smart water meter. First, during on-site installation, staff use a high-precision handheld terminal equipped with a high-precision GNSS receiver (supporting RTK / PPP technology, with sub-meter or even centimeter-level positioning accuracy), a laser rangefinder, or a visual recognition module (combined with AR / SLAM technology). This handheld terminal accurately measures and collects the initial geographic coordinates of the smart water meter terminal, including longitude, latitude, and altitude, typically achieving sub-meter or even centimeter-level accuracy, through the GNSS receiver or laser rangefinder function. Second, through the high-precision handheld terminal's interface, staff can intuitively select on a high-precision internet map (such as Gaode Maps) or scan a pre-set NFC / QR code on the water meter or its installation location to bind the smart water meter's unique identifier (ID, such as the water meter number) to a specific semantic landmark entity in the high-precision internet map's landmark knowledge graph (such as "XX Community, Building XX, Unit XX, Room XX," or more specifically, the installation location "below the kitchen sink"). The NFC / QR code pre-stores a landmark semantic ID that uniquely corresponds to a landmark entity in the landmark knowledge graph. Third, the high-precision handheld terminal uploads the water meter ID, precise initial geographic coordinates, landmark semantic ID, and other installation context information (such as water meter type, installation environment photos, installer ID, and installation timestamp) to the data access layer 31 of the smart water meter reading platform 30 via its own wireless communication network (such as 4G / 5G). Fourth, the landmark semantic association and optimization module 35 of the smart water meter reading platform 30 receives and processes the initial binding data, storing it as the highest priority information in the landmark knowledge graph, establishing a mapping relationship of "smart water meter ID - initial precise coordinates - landmark semantic ID," and updating the landmark knowledge graph, including adding the associated record of the water meter to the relevant landmark entities. Fifth, the smart water meter terminal 10 locally stores the landmark semantic ID and related initial configuration parameters, such as reporting period and positioning strategy thresholds, for reference in case of platform communication interruption or positioning ambiguity. The smart water meter data is collected and transmitted periodically. This step describes the data collection and reporting process of the smart water meter terminal during daily operation. First, the data collection module 11 of the smart water meter terminal 10 automatically collects the current water meter reading (such as cumulative flow value, instantaneous flow value, remote valve status) and the smart water meter terminal's own equipment operating status data, such as battery level, signal strength, valve status, and leakage alarm status, within a preset meter reading cycle (such as at a fixed time every day, every hour, or triggered according to changes in water consumption). This data undergoes preliminary verification and packaging by the local storage and processing module 14. Second, the multimodal positioning module 12 of the smart water meter terminal 10 selectively or in parallel activates the corresponding positioning units to collect raw location data according to a preset low-power positioning strategy and environmental conditions. Specifically: a. When external GNSS signals are available, the GNSS receiver module is activated first to receive satellite signals and calculate the current geographic coordinates. To save energy, the GNSS receiver module can be activated for a short time (e.g., 5-10 seconds), and immediately enters a sleep state after acquiring positioning data (NMEA data stream). b. When GNSS signals are unavailable or their quality is below a preset threshold (e.g., indoors, underground, or in urban canyon environments), the multimodal positioning module switches to Wi-Fi scanning mode, activates the Wi-Fi transceiver module, and scans for surrounding Wi-Fi access points within a preset scanning time (e.g., 3-5 seconds), collecting the MAC address (BSSID / SSID) and corresponding RSSI value of each access point to form Wi-Fi fingerprint data. c. When Wi-Fi signals are unavailable or the assisted positioning effect is poor, or as a supplement to Wi-Fi positioning, the multimodal positioning module activates the BLE transceiver module, scans for surrounding BLE beacons, and collects the UUID / Major / Minor identifier and corresponding RSSI value of each beacon to form BLE beacon data. d. The multimodal positioning module can dynamically adjust the activation frequency and priority of different positioning modes according to the actual environment (e.g., whether it is in a known GNSS blind zone), battery power, and preset energy budget. For example, when the battery level is below 30%, only the lower-power Wi-Fi / BLE scan is performed, while when the battery level is high, the more energy-intensive but potentially more accurate GNSS positioning is attempted. Third, the wireless communication module 13 of the smart water meter terminal 10 uploads the water meter readings, device operating status data, and raw location data (including but not limited to GNSS coordinates, Wi-Fi fingerprint data, and BLE beacon data) processed by the local storage and processing module 14 to the data access layer 31 of the smart water meter reading platform 30 via the data transmission network 20 (e.g., NB-IoT or LoRaWAN protocol). To save power, the data is usually compressed and batch-processed to minimize wireless transmission time. The platform-side multi-source positioning data fusion and landmark association is the core computational process for achieving high-precision landmark positioning in this invention. First, the data access layer 31 of the smart water meter reading platform 30 receives and parses various types of data uploaded from the smart water meter terminal 10, completing protocol conversion and preliminary verification. Second, the multi-source positioning data fusion module 34 processes and fuses the received raw location data. a. If GNSS data is received, the navigation satellite system data processing unit 341 performs preliminary verification and smoothing, and calculates its position accuracy factor (DOP value) and signal quality assessment. b. If Wi-Fi fingerprint data is received, the Wi-Fi fingerprint positioning unit 342 uses a pre-built Wi-Fi fingerprint database and pattern matching algorithms such as K-NN, SVM, or neural networks to estimate the approximate location of the smart water meter and outputs its estimated accuracy. c. If BLE beacon data is received, the Bluetooth Low Energy beacon positioning unit 343 uses a pre-built BLE beacon database and a ranging algorithm based on a signal propagation model to estimate the approximate location of the smart water meter and outputs its estimated accuracy. d. The data fusion engine 344 comprehensively analyzes positioning data from different positioning sources (GNSS, Wi-Fi, BLE) and their accuracy and confidence, and uses fusion algorithms (such as extended Kalman filter, particle filter) to generate an optimized water meter physical location estimate with higher accuracy and confidence. This estimate includes accurate geographic coordinates (longitude, latitude, altitude) and a covariance matrix representing uncertainty. The fusion engine dynamically adjusts the weights of each positioning source based on the real-time signal quality and environmental context. Third, the landmark semantic association and optimization module 35 receives the fused water meter physical location estimate and performs landmark semantic association and optimization operations: a. The preliminary spatial matching unit 351 uses the spatial index of the landmark knowledge graph (such as R-tree) to quickly retrieve and identify multiple candidate landmark entities that are spatially closest to or contain the fused location estimate, such as "X Community, Building X, Unit X" and its subordinate "Room X". b. The landmark constraint reasoning unit 352 invokes preset geographical and logical constraint rules (such as "water meter is located inside a building" and "one meter per unit") to evaluate candidate landmark entities, exclude landmarks that do not conform to the rules, and adjust the confidence level of the remaining candidate landmarks according to the degree of conformity to the rules. c. The historical trajectory and behavior pattern analysis unit 353 combines the historical location records of the water meter, historical landmark associations, and behavior patterns (such as water usage patterns) in the water meter reading data to further correct and optimize the current candidate landmarks. For example, if there is a small deviation between the current fusion location and the historical landmark, but it is still within a reasonable range and conforms to the historical water usage pattern, it tends to maintain or fine-tune to the historical landmark; if there is a large and persistent deviation, it will trigger a landmark re-evaluation or alarm.d. If, after the above steps, multiple candidate landmarks still exist or the landmark confidence level is lower than the preset threshold, the landmark semantic association and optimization module 35 can further request manual assistance for calibration to obtain final confirmation. Fourth, the landmark semantic association and optimization module 35 finally outputs a precise and semantically accurate landmark positioning result, which includes the smart water meter's ID, the determined geographical coordinates (the optimal location after fusion), and the corresponding semantic landmark description (such as "the water meter in apartment XX, unit XX, building XX, XX community"). Platform data management and application: This step covers platform-side data storage, visualization, and external services. First, the data storage and management layer 32 of the smart water meter reading platform 30 associates and updates the smart water meter ID, the latest water meter reading, device operating status data, and high-precision landmark positioning results. For example, it associates the landmark ID as a foreign key with the water meter data table and updates the latest location of the water meter entity in the landmark knowledge graph. Second, the visualization and management module 36 displays the location, real-time data, and device status of the smart water meter in an intuitive graphical way on a high-precision internet map based on the high-precision landmark positioning results. Detailed information such as the floor and room where the water meter is located can be displayed on the map. Third, when the landmark semantic association and optimization module 35 determines that the positioning accuracy is insufficient, the landmark association is ambiguous (e.g., the best landmark confidence and the second-best landmark confidence are very close), or it is seriously inconsistent with historical data, the visualization and management module 36 generates an alarm message and sends it to the management personnel terminal (such as a web interface or mobile app), prompting manual verification or triggering the manual auxiliary calibration and feedback unit 354. Fourth, platform 30 receives calibration data submitted by the manual calibration and feedback unit 354 and uses this data to update the water meter landmark association information in the landmark knowledge graph. Simultaneously, it further optimizes the model parameters of the landmark constraint reasoning unit 352 and the historical trajectory and behavior pattern analysis unit 353 (e.g., by retraining the machine learning model), forming a continuous optimization loop to continuously improve the accuracy of the system's automated landmark association. Fifth, application service interface 37 provides water meter readings, equipment status, and historical data accurately associated with landmarks based on requests from external systems. This supports various business applications such as water management, billing, fault diagnosis, and on-site work assignment, greatly enhancing the refinement and intelligence of water operations.

[0053] Example: This example aims to verify the actual performance of the smart water meter reading platform and its reading method based on high-precision internet map landmark positioning proposed in this invention in a complex urban residential environment. We selected a typical urban high-rise residential community as the test site. The community consists of five 20-story buildings, each building containing four units, with two households per floor in each unit, and each household equipped with a main water meter. In addition, the community also has public area water meters, such as water meters for greening and fire fighting, which may be installed in outdoor boxes or underground pipe shafts.

[0054] We deployed 100 smart water meter terminals 10 of this invention, model "AquaSensePro-M", of which 90 were installed inside residences (one per household) and 10 were installed in public areas. The configuration of the smart water meter terminal 10 is as follows: Main control chip: equipped with an STM32L476RG ultra-low power microcontroller, with a main frequency of 80MHz, and built-in 1MB Flash and 128KBSRAM; Flow sensor: ultrasonic flow sensor from SICKAG, Germany, with an accuracy class of OIMLR49Class2 and an operating temperature range of -20℃ to 70℃; Multimodal positioning module 12: GNSS receiver: U-bloxZED-F9P high-precision module, supporting GPS, GLONASS, BDS, Galileo L1 / L2 multi-band, with RTK function, which can achieve centimeter-level positioning accuracy under ideal conditions. To save power, it is configured to activate once every 2 hours, with each positioning lasting 10 seconds; Wi-Fi transceiver module: ESP32-WROOM-32E module, supporting IEEE802.11b / g / n, with a scanning range of -90dBm to -30dBm. Configured to activate when GNSS positioning fails or HDOP>2.5, scanning surrounding APs for 5 seconds; BLE transceiver module: Nordic NRF52832 chip, supports BLE5.0, configured to activate when the number of Wi-Fi scanning APs is less than 3, scanning iBeacon / Eddystone beacons for 5 seconds; Wireless communication module 13: Quectel BC95-B5NB-IoT module, supports China Telecom Band 5 band, maximum transmit power 23dBm, data rate uplink up to 62.5kbps, downlink up to 26.15kbps; Energy management module 15: uses Linear Technology LTC3331 nanowatt-level PMIC, integrates coulomb meter, and is powered by a 19Ah lithium thionyl chloride battery; Data acquisition frequency: water meter readings are collected once per hour, and device status data is collected once every 30 minutes. Location data is triggered according to the policy, and the size of each uploaded data packet is about 150-300 bytes; the smart water meter reading platform 30 is deployed on Alibaba Cloud and uses Kubernetes cluster management microservices. The core component configuration is as follows: Data access layer 31: Nginx is used as API Gateway and Kafka is used as message queue, with a peak throughput of up to 10,000 messages / second.

[0055] Data storage and management layer 32: Time series database: Alibaba Cloud TSDB for water meter readings and status data; Relational database: Alibaba Cloud RDS PostgreSQL 12 for raw location data and fused location data; Graph database: Neo4j Enterprise Edition, deployed on a dedicated server cluster for landmark knowledge graphs.

[0056] The high-precision landmark data integration module 33: It acquires POI data, building outline data, and road data by calling the Gaode Map API; imports the community BIM model (Revit format), extracting precise 3D geometric information and semantic descriptions of each room and pipe shaft; manually surveys the precise RTK-GNSS coordinates of 50 known water meter locations within the community and corrects them against the BIM data; the constructed landmark knowledge graph contains over 2000 landmark entities (community, building, unit, floor, room, pipe shaft), connected by relationships such as "belong to," "contains," and "adjacent to." Each landmark entity stores its geometric boundaries in GeoJSON format.

[0057] Multi-source positioning data fusion module 34: Navigation satellite system data processing unit 341: Performs differential correction on NMEA data (via RTK base station data or cloud PPP service), and applies extended Kalman filtering for smoothing; Wi-Fi fingerprint positioning unit 342: Offline surveys were conducted on the Wi-Fi AP fingerprints of each floor of each building and some typical apartment types within the community, constructing a fingerprint database containing 500 reference points. The weighted K-nearest neighbor (WKNN) algorithm was used, with K=5; Bluetooth Low Energy beacon positioning unit 343: Deployed a total of 150 BLEiBeacon beacons in the stairwells, elevator lobbies, and some areas with dense water meters in each unit of each building, pre-calibrating their transmission power and path loss models. Distance estimation was performed using trilateration combined with the Log-DistancePathLoss model; Data fusion engine 344: Employs adaptive extended Kalman filtering (AdaptiveEKF), with the state vector including longitude, latitude, elevation, and velocity. The weights of GNSS, Wi-Fi, and BLE are dynamically adjusted based on the HDOP value of GNSS, the number of valid APs / beacons scanned by Wi-Fi / BLE, and the signal strength.

[0058] Landmark semantic association and optimization module 35: Preliminary spatial matching unit 351: Use the R-tree spatial index based on PostGIS extension to perform point within polygon and nearest neighbor queries.

[0059] Landmark Constraint Reasoning Unit 352: The rule engine is implemented based on the Drools framework and contains more than 20 geographical and logical rules, such as "the water meter entering the household must be located within the residential unit", "the main meter cannot be in a public area", and "adjacent water meters less than 2 meters apart are considered to be the same installation point". Historical Trajectory and Behavior Pattern Analysis Unit 353: It adopts an LSTM-based neural network model. The input features include the fused location sequence of the past 7 days, daily water consumption changes, and location stability during specific time periods (such as midnight). The model is pre-trained on historical data to predict the rationality of landmarks at the current location. Human-Assisted Calibration and Feedback Unit 354: A high-precision handheld terminal App based on Android has been developed, integrating the Gaode Map SDK and RTK-GNSS module, and supporting NFC / QR code scanning and binding.

[0060] Experimental Procedure and Data Acquisition: After deploying 100 smart water meter terminals, initialization and landmark pre-binding were performed using handheld terminals. During the one-month testing period, the platform received and processed data from the terminals hourly. To evaluate positioning accuracy and landmark association accuracy, we randomly selected 20 water meters (15 indoors and 5 in public areas) as verification samples and invited a third-party surveying agency to conduct on-site measurements using professional RTK equipment to obtain their true physical coordinates and landmark attributes as ground truth values.

[0061] Experimental Results: For these 20 verification samples, the platform of this invention performed 24-hour fusion localization and landmark association on their locations. The daily average localization error and landmark association accuracy were recorded.

[0062] Indoor water meters (15 units): Due to severe GNSS signal attenuation indoors, positioning primarily relies on Wi-Fi fingerprinting and BLE beacons. The fusion module, combining historical data and landmark constraints, achieved significant results. The average positioning error was controlled within 2.5 meters, and the landmark association accuracy reached 98.5%. Even in areas with weak Wi-Fi / BLE signals, historical pattern analysis and landmark constraint inference can accurately associate water meters with the correct room or pipe shaft.

[0063] Public area water meters (5 units): These water meters are partially installed outdoors or semi-outdoors, where GNSS signals are relatively good. The fusion module fully utilizes the high precision advantage of RTK-GNSS, while also incorporating Wi-Fi / BLE and landmark knowledge graphs for assistance. The average positioning error is controlled within 0.8 meters, and the landmark association accuracy reaches 100% (accurate to the specific installation box or manhole cover).

[0064] The following table shows the accuracy of landmark association for five randomly selected representative water meters during the testing period: Water Meter ID Installation location (actual landmark) Main positioning environment Average fusion positioning error (m) Landmark association accuracy (%) M-001 Kitchen of Unit 1, Room 101, Building 1 Indoors, weak GNSS 2.15 99.2 M-002 Bathroom in Unit 3, Building 2, Room 502 Indoors, GNSS unavailable 2.88 97.8 M-003 3 buildings, 2 units, 12-story pipe shaft In semi-indoor, GNSS is extremely weak. 1.83 100.0 M-004 Water meter boxes in the green belt at the entrance of Building 4 Outdoors, GNSS is superior 0.67 100.0 M-005 Fire hydrants in 5 underground parking garages Underground, GNSS unavailable 3.52 96.5 As can be seen from the above data, even in complex indoor and underground environments, this invention can still achieve high positioning accuracy and reliable landmark association through multimodal data fusion and landmark semantic association, especially in terms of landmark association accuracy, which is of great significance for the refinement of water management.

[0065] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A smart water meter reading platform based on high-precision internet map landmark positioning, characterized in that: include: Smart water meter terminal (10): configured on the water meter body, used to collect water meter readings, equipment operation status data and collect raw location data according to preset positioning strategy; Data transmission network (20): Provides a two-way wireless data communication link between the smart water meter terminal (10) and the smart water meter reading platform (30); Smart water meter reading platform (30): Configured on cloud or local server cluster, it receives and processes relevant data and generates high-precision landmark positioning results. It includes data access layer (31), data storage and management layer (32), high-precision landmark data integration module (33), multi-source positioning data fusion module (34), landmark semantic association and optimization module (35), visualization and management module (36), and application service interface (37).

2. The smart water meter reading platform based on high-precision internet map landmark positioning according to claim 1, characterized in that: The smart water meter terminal (10) includes: Data acquisition module (11): Periodically collects water meter readings and equipment operating status data; Multimodal positioning module (12): According to the preset positioning strategy, the corresponding module is activated under different conditions to collect raw location data; Wireless communication module (13): Uploads the collected data to the smart water meter reading platform (30) through the data transmission network (20); Local storage and processing module (14): Temporarily stores the collected data and performs preliminary filtering or preprocessing of the raw location data; Energy Management Module (15): Controls the power supply and working mode of the multimodal positioning module (12) to achieve low-power operation.

3. The smart water meter reading platform based on high-precision internet map landmark positioning according to claim 2, characterized in that: The positioning strategy of the multimodal positioning module (12) includes: When external navigation satellite system signals are available, the navigation satellite system receiver module is activated first to obtain satellite signal positioning. When the navigation satellite system signal is unavailable or of low quality, switch to Wi-Fi scanning mode to collect relevant data from Wi-Fi access points to assist in positioning; When the Wi-Fi signal is weak or used as a supplement, the Bluetooth Low Energy transceiver module is activated to collect Bluetooth beacon-related data for positioning, and the activation frequency and priority of different positioning modes can be dynamically adjusted.

4. The smart water meter reading platform based on high-precision internet map landmark positioning according to claim 1, characterized in that: The high-precision landmark data integration module (33) communicates with external high-precision Internet map service providers through API to obtain geospatial data and landmark information, receives data from other related systems, constructs and maintains a hierarchical, multi-dimensional landmark knowledge graph, associates geographic coordinates with geographic entities and stores their various relationships and information.

5. The smart water meter reading platform based on high-precision internet map landmark positioning according to claim 1, characterized in that: The multi-source positioning data fusion module (34) includes: Navigation satellite system data processing unit (341): parses relevant data streams and performs error correction processing; Wi-Fi fingerprint positioning unit (342): estimates location using a pre-built Wi-Fi fingerprint database and pattern matching algorithm; Bluetooth Low Energy Beacon Positioning Unit (343): Estimates location based on Bluetooth Low Energy Beacon database and related algorithms; Data fusion engine (344): integrates various positioning sources and outputs high-precision fused position estimates according to a preset fusion algorithm, and can dynamically adjust the weights.

6. The smart water meter reading platform based on high-precision internet map landmark positioning according to claim 5, characterized in that: The Wi-Fi fingerprint database is established through offline surveying and stores relevant features. The pattern matching algorithm includes K-nearest neighbor algorithm, support vector machine or neural network. The Bluetooth Low Energy beacon database stores corresponding identification information and signal propagation model parameters.

7. The smart water meter reading platform based on high-precision internet map landmark positioning according to claim 1, characterized in that: The landmark semantic association and optimization module (35) includes: Preliminary spatial matching unit (351): Retrieves and identifies candidate landmark entities; Landmark Constraint Reasoning Unit (352): Applying geographical and logical constraint rules to filter, verify, and optimize candidate landmarks; Historical Trajectory and Behavioral Pattern Analysis Unit (353): Analyzes historical data and combines it with machine learning algorithms to assist in judgment and correction; Human-assisted calibration and feedback unit (354): Collects human-computer interaction data to update the landmark knowledge graph and optimize relevant parameters and models through human-computer interaction interface.

8. The smart water meter reading platform based on high-precision internet map landmark positioning according to claim 7, characterized in that: The logical constraint rules involve limiting the installation location of water meters, and include limiting the correspondence between water meters and landmark entities.

9. The smart water meter reading platform based on high-precision internet map landmark positioning according to claim 1, characterized in that: The visualization and management module (36) supports multi-level map layer overlay, has alarm information display and management functions as well as asset management functions, and is associated with high-precision landmark positioning results.

10. A smart water meter reading method based on high-precision internet map landmark positioning, applicable to the smart water meter reading platform described in any one of claims 1-9, characterized in that: Includes the following steps: S1. Smart water meter initialization and landmark pre-binding: During installation, staff use a high-precision handheld terminal to collect the initial geographic coordinates of the smart water meter terminal (10) and bind them with the semantic landmark entity in the landmark knowledge graph. They upload the relevant information to the platform, and the platform establishes and updates the mapping relationship stored in the landmark knowledge graph. S2. Periodic collection and transmission of smart water meter data: Each module of the smart water meter terminal (10) collects and transmits data to the data access layer (31) of the platform according to the preset. S3, Platform-side multi-source positioning data fusion and landmark association: After the platform data access layer (31) receives the data, the multi-source positioning data fusion module (34) fuses the data to generate a location estimate, and the landmark semantic association and optimization module (35) performs association operations to generate a high-precision landmark positioning result; S4. Platform Data Management and Application: The platform data storage and management layer (32) associates and updates related data, and the visualization and management module (36) displays the location of smart water meters, real-time data and device status and provides alarm management functions.

Images