Portable multi-parameter water quality analyzer and intelligent calibration and remote transmission method

Abstract

The invention discloses a portable multi-parameter water quality analyzer and an intelligent calibration and remote transmission method. The analyzer comprises an autonomous navigation and positioning unit, a propelling and maneuvering unit, a sensing and sensor unit, a sampling and analyzing unit and a calculation and communication unit which jointly form an integrated mobile monitoring platform capable of autonomous cruising, multi-source sensing, sampling analysis and intelligent processing. According to the method, accurate navigation is realized by fusing positioning and real-time obstacle avoidance, and online anomaly detection and alarm are carried out by adopting a multi-threshold and machine learning algorithm; after an alarm is triggered, the system starts a multi-node cooperation mechanism, and data cross validation and autonomous diagnosis and calibration of fault nodes are carried out based on a Byzantine fault-tolerant consensus algorithm; and finally, performing space-time alignment and confidence weighted fusion on the calibrated multi-source data, and filling a monitoring gap through Kriging spatial interpolation to generate a high-resolution and high-credibility comprehensive water quality field analysis result.

Description

Technical Field

[0001] This invention relates to the field of environmental monitoring and calibration, and more specifically, to a portable multi-parameter water quality analyzer and an intelligent calibration and remote transmission method. Background Technology

[0002] With the increasing severity of global water scarcity and water pollution, high-precision, high-spatiotemporal resolution real-time monitoring and intelligent early warning of the water environment have become crucial for ensuring drinking water safety, maintaining the health of aquatic ecosystems, and achieving scientific water resource management. Traditional static monitoring methods, relying on fixed stations and manual inspections, struggle to capture the instantaneous dynamic evolution and spatial heterogeneity of water pollution events, exhibiting significant shortcomings in rapid response to sudden pollution incidents, accurate source tracing, and the routine, refined management of large-scale water areas. Therefore, developing a novel water quality monitoring technology system capable of autonomous movement, collaborative operation, intelligent sensing, and self-calibration and data fusion capabilities is of significant scientific value and urgent practical importance for building a smart water environment monitoring network and enhancing environmental risk early warning and emergency management capabilities.

[0003] Existing technologies suffer from the following key limitations: First, in terms of monitoring paradigms, they generally employ discrete and static point-based monitoring or rely on periodic manual sampling by manpower and boats. This results in numerous spatial and temporal "blind spots" and "gaps" in the monitoring network, failing to achieve continuous and seamless coverage of target water areas and offering weak characterization of pollutant diffusion paths and spatial distribution patterns. Second, regarding data quality and reliability, each sensor unit typically operates independently, lacking effective multi-source data cross-validation and collaborative calibration mechanisms. Abnormal data caused by sensor drift, biofouling, sudden malfunctions, or environmental interference are difficult to identify and eliminate in real time and accurately, easily leading to false alarms or missed alarms, affecting the long-term consistency and reliability of monitoring data. Finally, in terms of intelligence, existing systems primarily focus on data collection and simple data transmission, lacking the ability to make autonomous decisions during the monitoring process and the capacity for deep fusion and intelligent mining of massive amounts of heterogeneous monitoring data. This makes it difficult to provide in-depth, high-value-added information products for the causal analysis and situation assessment of pollution events. Summary of the Invention

[0004] The main objective of this invention is to provide a portable multi-parameter water quality analyzer and an intelligent calibration and remote transmission method, so as to at least solve the problems of large monitoring blind spots and delayed data updates in the prior art, and improve the long-term reliability of environmental monitoring data.

[0005] To achieve the above objectives, a portable multi-parameter water quality analyzer and an intelligent calibration and remote transmission method are provided.

[0006] In a first aspect, the present invention provides a portable multi-parameter water quality analyzer, the portable multi-parameter water quality analyzer comprising:

[0007] The main structure is a streamlined shell structure modeled after a tuna;

[0008] The autonomous navigation and positioning unit is located at the head of the main structure. The autonomous navigation and positioning unit is used to obtain the real-time position of the water quality analyzer, correct the real-time position and plan the preset route.

[0009] The propulsion and maneuvering unit is located at the tail and two fins of the main structure and is connected to the autonomous navigation and positioning unit. The propulsion and maneuvering unit is used to drive the water quality analyzer to navigate precisely according to the preset route, and at the same time, it performs real-time obstacle avoidance based on the real-time position.

[0010] The sensing and sensor unit is located at the head and two gills of the main structure and is connected to the computing and communication unit. The sensing and sensor unit is used to collect multiple water quality data and multiple water body data of the target water area and send the water quality data and water body data to the computing and communication unit.

[0011] The sampling and analysis unit is located in the abdomen of the main structure and is connected to the sensing and sensor unit. The sampling and analysis unit is used to collect multiple water samples from the target water area and perform preprocessing and data analysis on the multiple water samples to obtain multiple first analysis results of the target water area, and send the multiple first analysis results to the computing and communication unit.

[0012] The computing and communication unit, located inside the abdominal cavity of the main structure, is connected to the sensing and sensor unit and the sampling and analysis unit. The computing and communication unit is used to preset abnormal event thresholds, process multiple water quality data and multiple water body data to obtain processing results, and compare the processing results with the abnormal event thresholds to issue corresponding abnormal event alarms. The computing and communication unit is also used to compare and analyze multiple water quality data, multiple water body data, and multiple first analysis results to identify fault nodes and perform autonomous calibration on the fault nodes. Furthermore, the computing and communication unit is used to obtain a first comprehensive water quality field based on multiple water quality data, multiple water body data, and multiple first analysis results, fill the spatial monitoring gaps in the first comprehensive water quality field using interpolation to obtain a second comprehensive water quality field, and process the second comprehensive water quality field to obtain a second analysis result.

[0013] In a second aspect, the present invention provides an intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer. The intelligent calibration and remote transmission method is applied to the portable multi-parameter water quality analyzer described in the first aspect, and includes:

[0014] The real-time position of the water quality analyzer is obtained by the autonomous navigation and positioning unit, the real-time position is corrected and a preset route is planned, and the water quality analyzer is driven to navigate accurately according to the preset route by the propulsion and maneuvering unit, while real-time obstacle avoidance is performed based on the real-time position.

[0015] The system uses a sensing and sensor unit to collect multiple water quality data and multiple water body data of the target water area and sends them to the computing and communication unit. The system uses a sampling and analysis unit to collect multiple water body samples of the target water area and performs preprocessing and data analysis on the multiple water body samples to obtain multiple first analysis results of the target water area. The system then sends the multiple first analysis results to the computing and communication unit.

[0016] The system uses computing and communication units to process multiple water quality data and multiple water body data to obtain processing results. It presets anomaly event thresholds and compares the processing results with the anomaly event thresholds to issue corresponding anomaly event alarms.

[0017] When an abnormal event alarm is triggered, multiple sensing and sensor units and multiple sampling and analysis units exchange multiple water quality data, multiple water body data, and multiple first analysis results respectively; the computing and communication unit compares and analyzes the multiple water quality data, multiple water body data, and multiple first analysis results to obtain the fault node, and performs autonomous calibration on the fault node;

[0018] The computing and communication unit is used to perform spatiotemporal alignment and data fusion on multiple water quality data, multiple water body data and multiple first analysis results to obtain a first comprehensive water quality field. The spatial monitoring gaps of the first comprehensive water quality field are filled by interpolation to obtain a second comprehensive water quality field. The second comprehensive water quality field is then processed to obtain a second analysis result.

[0019] Specifically, the real-time position of the water quality analyzer is obtained using an autonomous navigation and positioning unit, the real-time position is corrected, and a preset route is planned. The propulsion and maneuvering unit then drives the water quality analyzer to navigate precisely according to the preset route, while simultaneously performing real-time obstacle avoidance based on the real-time position, including:

[0020] The autonomous navigation and positioning unit collects GPS signals, acoustic positioning signals, diving depth, acceleration, and heading angle. The extended Kalman filter is used to fuse the GPS signals, acoustic positioning signals, diving depth, acceleration, and heading angle to obtain fused positioning data.

[0021] The autonomous navigation and positioning unit plans the route based on the starting and ending points of the route, while the propulsion and maneuvering unit performs real-time obstacle avoidance based on the fused positioning data and the route.

[0022] Specifically, the system utilizes sensing and sensor units to collect multiple water quality data and multiple water body data of the target water area, and utilizes computing and communication units to process the multiple water quality data and multiple water body data to obtain processing results, including:

[0023] The sensing and sensor unit collects multiple water quality data and corresponding multiple water body data in the target water area. The water quality data includes pH value, dissolved oxygen, conductivity and chlorophyll value, and the water body data includes temperature and turbidity.

[0024] The computing and communication unit performs sliding window feature expansion on multiple water quality data and corresponding multiple water body data to obtain water quality feature vectors;

[0025] The isolated forest algorithm is used to process water quality feature vectors to obtain global anomaly scores and local contribution vectors.

[0026] Specifically, a preset abnormal event threshold is established, and the processing result is compared with the abnormal event threshold to issue a corresponding abnormal event alarm, including:

[0027] Preset thresholds for pH, dissolved oxygen, conductivity, chlorophyll, temperature, and turbidity. Compare these thresholds with the following values ​​to obtain multiple first comparison results: pH, dissolved oxygen, conductivity, chlorophyll, temperature, turbidity, and the corresponding thresholds for pH, dissolved oxygen, conductivity, chlorophyll, temperature, and turbidity.

[0028] A preset abnormal score threshold is set, and a second comparison result is obtained by comparing the global abnormal score with the abnormal score threshold. Based on multiple first comparison results, second comparison results and local contribution vectors, corresponding abnormal event alarms are issued.

[0029] Specifically, multiple water samples from the target water area are collected through a sampling and analysis unit, and preprocessing and data analysis are performed on these samples to obtain multiple first analysis results for the target water area, including:

[0030] The sampling and analysis unit collects water samples from the area in the target water body where an abnormal event alarm was generated.

[0031] The water sample is filtered, digested online, and enriched to obtain the treated water sample;

[0032] The first analytical results were obtained by colorimetric, electrochemical and optical detection of the treated water sample.

[0033] Specifically, when an abnormal event alarm is triggered, multiple sensing and sensor units and multiple sampling and analysis units exchange multiple water quality data, multiple water body data, and multiple first analysis results, including:

[0034] When an abnormal event alarm is triggered, the autonomous navigation and positioning unit at the trigger point broadcasts a request signal to the target water area;

[0035] Upon receiving the request signal, multiple sensing and sensor units and multiple sampling and analysis units exchange multiple water quality data, multiple water body data, and multiple first analysis results in a unified data format.

[0036] Specifically, the fault nodes are identified by comparing and analyzing multiple water quality data, multiple water body data, and multiple first analysis results using computing and communication units. Autonomous calibration of the fault nodes is then performed, including:

[0037] The computing and communication unit uses the Byzantine fault-tolerant consensus algorithm to calculate the water quality consensus result, water body consensus result, and first analysis consensus result of multiple water quality data, multiple water body data, and multiple first analysis results.

[0038] Preset water quality consensus threshold, water body consensus threshold and first analysis consensus threshold, and compare the water quality consensus result, water body consensus result and first analysis consensus result with the water quality consensus threshold, water body consensus threshold and first analysis consensus threshold respectively to obtain the fault node;

[0039] Perform soft and hard calibration on the faulty nodes.

[0040] Specifically, the computing and communication unit is used to perform spatiotemporal alignment and data fusion of multiple water quality data, multiple water body data, and multiple first analysis results to obtain a first comprehensive water quality field, including:

[0041] The computing and communication unit establishes a unified time reference and performs cubic spline interpolation on multiple water quality data, multiple water body data, and asynchronous data from multiple first analysis results;

[0042] Convert multiple water quality data, multiple water body data, and location data from multiple first analysis results into the UTM coordinate system;

[0043] The first comprehensive water quality field is obtained by using confidence-weighted fusion of multiple water quality data, multiple water body data, and multiple first analysis results.

[0044] Specifically, the second comprehensive water quality field is obtained by using interpolation to fill the spatial monitoring gaps in the first comprehensive water quality field, including:

[0045] Multiple constraints are defined, and a second integrated water quality field is generated based on these constraints using the Kriging space interpolation method.

[0046] This application provides a portable multi-parameter water quality analyzer and an intelligent calibration and remote transmission method. This method, through the collaborative operation of multiple sensor units and sampling analysis units, first achieves precise navigation along a preset route based on fusion positioning and real-time obstacle avoidance. Then, it utilizes multi-dimensional threshold judgment and an isolated forest algorithm to perform online anomaly detection and alarm on the real-time collected water quality data. When an alarm is triggered, the system activates a distributed collaborative mechanism, where multiple nodes exchange and compare data through a Byzantine fault-tolerant consensus algorithm, identifying and autonomously calibrating potentially faulty sensors to ensure the reliability of the data source. Subsequently, the system performs spatiotemporal alignment and confidence-weighted fusion of all valid data to construct a preliminary comprehensive water quality field. Furthermore, it fills monitoring gaps using Kriging spatial interpolation, ultimately generating a high-resolution, high-reliability comprehensive water quality analysis result (second analysis result), realizing intelligent processing across the entire chain from data acquisition and quality control to spatial modeling. Attached Figure Description

[0047] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0048] Figure 1 A connection diagram of a portable multi-parameter water quality analyzer provided in this application;

[0049] Figure 2 A schematic diagram of the structure of a portable multi-parameter water quality analyzer provided in this application;

[0050] Figure 3 This is a flowchart illustrating an intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer provided in this application.

[0051] The above figures include the following reference numerals:

[0052] 10. Forward guide and sensor compartment; 11. Multispectral lidar; 12. High-frequency obstacle avoidance sonar; 13. High-definition underwater camera; 14. Three-dimensional electric field sensor; 20. Central processing and sampling compartment; 21. Fiber optic dissolved oxygen sensor; 22. Four-ring conductivity probe; 23. Combined pH / ORP electrode; 24. Turbidity sensor; 25. Chlorophyll / cyanobacteria fluorometer; 26. Miniature CTD; 27. Retractable vertical profile sensor chain; 30. Energy and propulsion compartment; 40. Attitude control compartment; 50. Tail fin propulsion compartment. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0054] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein.

[0055] In this invention, the terms "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0056] This application provides a portable multi-parameter water quality analyzer and an intelligent calibration and remote transmission method. This method achieves reliable acquisition and in-depth analysis of water quality data through a closed-loop process of "perception-decision-coordination-fusion." Based on autonomous navigation and real-time obstacle avoidance, it cruises along a preset route and uses multi-threshold comparison and machine learning algorithms to perform real-time anomaly detection and alarm on in-situ sensor data. Once an anomaly is detected, a distributed collaborative diagnostic mechanism is triggered: multiple monitoring nodes exchange data and use a Byzantine fault-tolerant consensus algorithm to identify possible faulty nodes, followed by hardware and software calibration to ensure data source reliability. Subsequently, the system performs spatiotemporal standardization and confidence-weighted fusion on the calibrated multi-source heterogeneous data to generate a preliminary comprehensive water quality field. It then uses spatial modeling techniques such as Kriging interpolation to fill geographical gaps, ultimately outputting a high spatial resolution and high reliability comprehensive water quality assessment for the entire region, completing the entire process from dynamic monitoring and fault self-correction to spatial intelligent diagnosis.

[0057] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will be described below with reference to the accompanying drawings.

[0058] Figure 1A connection diagram of a portable multi-parameter water quality analyzer provided in this application is shown below. Figure 1 As shown, this embodiment provides a portable multi-parameter water quality analyzer, which includes:

[0059] The main structure is a streamlined shell structure modeled after a tuna;

[0060] The autonomous navigation and positioning unit is located at the head of the main structure. The autonomous navigation and positioning unit is used to obtain the real-time position of the water quality analyzer, correct the real-time position and plan the preset route.

[0061] The propulsion and maneuvering unit is located at the tail and two fins of the main structure and is connected to the autonomous navigation and positioning unit. The propulsion and maneuvering unit is used to drive the water quality analyzer to navigate precisely according to the preset route, and at the same time, it performs real-time obstacle avoidance based on the real-time position.

[0062] The sensing and sensor unit is located at the head and two gills of the main structure and is connected to the computing and communication unit. The sensing and sensor unit is used to collect multiple water quality data and multiple water body data of the target water area and send the water quality data and water body data to the computing and communication unit.

[0063] The sampling and analysis unit is located in the abdomen of the main structure and is connected to the sensing and sensor unit. The sampling and analysis unit is used to collect multiple water samples from the target water area and perform preprocessing and data analysis on the multiple water samples to obtain multiple first analysis results of the target water area, and send the multiple first analysis results to the computing and communication unit.

[0064] The computing and communication unit, located inside the abdominal cavity of the main structure, is connected to the sensing and sensor unit and the sampling and analysis unit. The computing and communication unit is used to preset abnormal event thresholds, process multiple water quality data and multiple water body data to obtain processing results, and compare the processing results with the abnormal event thresholds to issue corresponding abnormal event alarms. The computing and communication unit is also used to compare and analyze multiple water quality data, multiple water body data, and multiple first analysis results to identify fault nodes and perform autonomous calibration on the fault nodes. Furthermore, the computing and communication unit is used to obtain a first comprehensive water quality field based on multiple water quality data, multiple water body data, and multiple first analysis results, fill the spatial monitoring gaps in the first comprehensive water quality field using interpolation to obtain a second comprehensive water quality field, and process the second comprehensive water quality field to obtain a second analysis result.

[0065] This application provides a portable multi-parameter water quality analyzer. Firstly, the analyzer achieves precise positioning and route planning by fusing multi-source signals through an autonomous navigation and positioning unit, controlling the propulsion unit to navigate along a preset path and avoid obstacles in real time. During navigation, the sensing unit continuously collects multi-parameter water body data, and the computing unit uses an anomaly detection algorithm to determine water quality anomalies in real time and trigger alarms. When an alarm is triggered, the system activates a multi-node collaborative mechanism, performing data cross-validation and fault diagnosis based on Byzantine fault-tolerant consensus to achieve autonomous calibration of abnormal nodes. Finally, all valid data undergoes spatiotemporal alignment, multi-source fusion, and spatial interpolation to generate a high-resolution comprehensive water quality field covering the entire region, completing a closed-loop process from dynamic monitoring and fault self-correction to spatial analysis.

[0066] Figure 2 A schematic diagram of a portable multi-parameter water quality analyzer provided in this application is shown below. Figure 2 As shown, this embodiment provides a portable multi-parameter water quality analyzer. The main structure of the portable multi-parameter water quality analyzer includes:

[0067] Adopting a streamlined shape inspired by a tuna, it measures 80 cm in length, 20 cm in maximum diameter, and weighs 12 kg in air. The main structure utilizes a carbon fiber composite frame, balancing high strength and lightweight design. The outer shell features a double-layer design: an inner waterproof and sealed polycarbonate hard shell provides structural support and protection; the outer layer is covered with a silicon-based elastic biomimetic skin. This skin's surface replicates the microgrooved structure of shark dermal fins, composed of flexible silicone and high-strength polyester mesh, significantly reducing water flow resistance (more than 45% lower than traditional cylinders) and providing self-cleaning and anti-biofouling functions. An integrated piezoelectric sensor array detects water pressure distribution on the skin.

[0068] The fuselage is divided into five rapidly separable, sealed compartments from front to back: the forward guidance and sensor compartment 10, housing forward detection sensors; the central processing and sampling compartment 20, containing the core computing unit and microfluidics laboratory; the energy and propulsion compartment 30, housing the battery pack and main propulsion motor; the attitude control compartment 40, housing the pectoral fin drive mechanism; and the tail fin propulsion compartment 50, housing the tail fin drive mechanism and communication antenna. The compartments are connected by a double-sealed connection: the first layer is a static seal with rubber O-rings, and the second layer is a magnetically coupled waterproof quick-connect electrical connector, ensuring long-term reliable operation at a depth of 100 meters.

[0069] The tail fin propulsion chamber 50 serves as the drive mechanism: it employs a brushless DC servo motor paired with a harmonic reducer, converting the motor's high speed into the low-speed, high-torque oscillation required by the tailstock. The reducer's output shaft connects to a four-bar linkage, converting rotational motion into reciprocating oscillation of the tailstock. Flexible tailstock and tail fin: The tailstock is made of a carbon fiber / shape memory alloy composite layer, and its local stiffness can be electronically controlled to optimize passive deformation during oscillation. The tail fin is made of glass fiber reinforced silicone, with an embedded PVDF piezoelectric film, capable of sensing water pressure and converting vibrational energy into electrical energy during gliding.

[0070] The attitude control cabin 40 utilizes independent dual-fin drive: each pectoral fin is independently driven by two miniature DC servo motors, achieving up-and-down flapping (stroke ±60°) and fin surface twisting (stroke ±45°) respectively. Fin surface structure: the pectoral fin skeleton is made of titanium alloy and covered with a flexible silicone membrane, which can mimic the wave-like movement of a ray fin.

[0071] The forward guide and sensor compartment 10 integrates a sensor array, including a multispectral lidar 11 that emits 905 nm pulsed laser light with a detection range of 50 meters, used for obstacle identification and underwater terrain scanning; a high-frequency obstacle avoidance sonar 12 with an operating frequency of 200 kHz, a beam angle of 15°, and a detection range of 10 meters, used for close-range obstacle avoidance; a high-definition underwater camera 13 with a 2-megapixel CMOS sensor and dual white LED and blue LED supplementary lights, used for optical observation, target identification, and image recording; and a three-dimensional electric field sensor 14 with triaxial orthogonal electrodes, used to detect weak bioelectric fields generated by organisms.

[0072] The central processing and sampling chamber 20 is equipped with continuous monitoring sensors, including: fiber optic dissolved oxygen sensor 21: based on the principle of fluorescence quenching, with a response time of <30 seconds; four-ring conductivity probe 22: AC impedance method, measurement range 0-100 mS / cm; combined pH / ORP electrode 23: a composite probe of glass and platinum electrodes; turbidity sensor 24: dual-channel detection of 90° scattered light and transmitted light; chlorophyll / cyanobacteria fluorometer 25: 470nm LED excitation, 685nm detection; and miniature CTD 26: integrating a ceramic piezoresistive pressure sensor (depth measurement), a platinum resistance temperature sensor (temperature measurement), and a conductivity sensor (salt measurement).

[0073] The central processing and sampling chamber 20 is also equipped with a retractable vertical profile sensor chain (27) in the abdomen. The chain is made of Kevlar tensile cable and has a built-in multi-core data transmission line. Six micro sensor nodes are fixed on the chain at equal intervals (e.g., 0.5 meters). Each node contains a temperature and dissolved oxygen micro sensor. The retraction mechanism is achieved by a winch driven by a micro DC motor. The chain can be completely stored in the special sealed chamber in the abdomen.

[0074] This portable multi-parameter water quality analyzer achieves continuous and adaptive coverage of water areas through mobile monitoring and intelligent path planning, effectively eliminating monitoring blind spots and improving spatial resolution and monitoring efficiency. It also introduces multi-node collaborative diagnosis and Byzantine fault tolerance mechanisms, enabling the identification and calibration of sensor faults or data anomalies without manual intervention, greatly enhancing system reliability and data quality. Finally, by deeply integrating in-situ sensor data and sampling analysis results and using spatial interpolation technology to construct a comprehensive water quality field, it not only provides a more comprehensive and accurate water quality assessment but also lays the data foundation for advanced applications such as pollution source tracing and situation prediction, realizing a leap in monitoring capabilities from "single-point discrete" to "full-field intelligence."

[0075] Figure 3 A flowchart illustrating the intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer provided in this application is shown below. Figure 3 As shown, this embodiment provides an intelligent calibration and remote transmission method, which includes:

[0076] The real-time position of the water quality analyzer is obtained by the autonomous navigation and positioning unit, the real-time position is corrected and a preset route is planned, and the water quality analyzer is driven to navigate accurately according to the preset route by the propulsion and maneuvering unit, while real-time obstacle avoidance is performed based on the real-time position.

[0077] The system uses a sensing and sensor unit to collect multiple water quality data and multiple water body data of the target water area and sends them to the computing and communication unit. The system uses a sampling and analysis unit to collect multiple water body samples of the target water area and performs preprocessing and data analysis on the multiple water body samples to obtain multiple first analysis results of the target water area. The system then sends the multiple first analysis results to the computing and communication unit.

[0078] The system uses computing and communication units to process multiple water quality data and multiple water body data to obtain processing results. It presets anomaly event thresholds and compares the processing results with the anomaly event thresholds to issue corresponding anomaly event alarms.

[0079] When an abnormal event alarm is triggered, multiple sensing and sensor units and multiple sampling and analysis units exchange multiple water quality data, multiple water body data, and multiple first analysis results respectively; the computing and communication unit compares and analyzes the multiple water quality data, multiple water body data, and multiple first analysis results to obtain the fault node, and performs autonomous calibration on the fault node;

[0080] The computing and communication unit is used to perform spatiotemporal alignment and data fusion on multiple water quality data, multiple water body data and multiple first analysis results to obtain a first comprehensive water quality field. The spatial monitoring gaps of the first comprehensive water quality field are filled by interpolation to obtain a second comprehensive water quality field. The second comprehensive water quality field is then processed to obtain a second analysis result.

[0081] This application provides an intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer. This method first plans the route and controls navigation and real-time obstacle avoidance through an autonomous navigation and positioning unit. During navigation, the sensing and sensor unit continuously collects water quality and water body data, while the computing and communication unit detects anomalies and triggers alarms based on preset thresholds. Simultaneously, the sampling and analysis unit preprocesses and analyzes water samples from the abnormal area to obtain preliminary results. When an alarm is triggered, the system achieves autonomous diagnosis and calibration of faulty nodes through multi-node data exchange and comparative analysis based on Byzantine fault-tolerant consensus. Finally, all valid data undergoes spatiotemporal alignment, multi-source fusion, and spatial interpolation to construct a high-resolution second comprehensive water quality field, generating the final analysis results. This achieves a complete closed-loop process from autonomous monitoring and intelligent diagnosis to full-field analysis.

[0082] This intelligent calibration and remote transmission method eliminates spatial blind spots through mobile monitoring and, through multi-node collaboration and Byzantine fault tolerance mechanisms, achieves highly reliable autonomous diagnosis and calibration of sensor faults and data anomalies without relying on manual intervention, significantly improving system robustness and data quality. Simultaneously, this method innovatively integrates real-time sensor data with laboratory-level sampling and analysis results, and uses spatial interpolation technology to generate a continuous, high-resolution comprehensive water quality field. This not only provides more accurate and comprehensive water quality assessments but also offers a reliable data foundation for advanced applications such as pollution source tracing and dynamic simulation, achieving a leap from discrete data acquisition to intelligent situational awareness across the entire field.

[0083] Specifically, the real-time position of the water quality analyzer is obtained using an autonomous navigation and positioning unit, the real-time position is corrected, and a preset route is planned. The propulsion and maneuvering unit then drives the water quality analyzer to navigate precisely according to the preset route, while simultaneously performing real-time obstacle avoidance based on the real-time position, including:

[0084] The autonomous navigation and positioning unit collects GPS signals, acoustic positioning signals, diving depth, acceleration, and heading angle. The extended Kalman filter is used to fuse the GPS signals, acoustic positioning signals, diving depth, acceleration, and heading angle to obtain fused positioning data.

[0085] The autonomous navigation and positioning unit plans the route based on the starting and ending points of the route, while the propulsion and maneuvering unit performs real-time obstacle avoidance based on the fused positioning data and the route.

[0086] This application provides an intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer. This method uses an autonomous navigation and positioning unit to simultaneously collect multi-source information such as GPS signals, acoustic positioning signals, diving depth, acceleration, and heading angle. It then utilizes an extended Kalman filter algorithm to fuse these heterogeneous data in real time, generating high-precision, robust fused positioning data. Based on this, the system automatically plans the optimal path according to the start and end points of a preset route. The propulsion and maneuvering unit, based on the fused positioning data and real-time environmental perception, dynamically performs obstacle avoidance maneuvers while precisely navigating along the planned route, ensuring navigation safety and mission continuity.

[0087] The portable multi-parameter water quality analyzer receives a GPS signal once per second during ascent and continuously receives ultra-short baseline (USBL) acoustic positioning signals underwater, acquiring diving depth, acceleration, and heading angle in real time.

[0088] The specific steps for fusing positioning data are: initialization, state prediction, observation update, and output. Initialization begins by defining the state vector, using the formula:

[0089]

[0090] in, Indicates the position in the NED coordinate system. This represents the velocity components in the NED coordinate system. This represents the roll angle (in radians around the x-axis), with positive values ​​pointing downwards to the right. This represents the pitch angle (in radians around the y-axis), with a positive value indicating the head is pointing upwards. It represents the yaw angle (in radians around the z-axis), with positive values ​​indicating clockwise.

[0091] Secondly, the formula for calculating state prediction is:

[0092]

[0093] in, Indicates the current time step. This represents the state prediction at time k based on time k-1. Represents a nonlinear state transition function. This represents the control inputs (including position in the NED coordinate system, velocity components in the NED coordinate system, roll angle, pitch angle, and yaw angle). express.

[0094] The formula for calculating the state transition Jacobian matrix is:

[0095]

[0096] in, This represents the state transition Jacobian matrix.

[0097] The formula for calculating covariance prediction is:

[0098]

[0099] in, This indicates the covariance prediction for the next time step. This represents the state estimation covariance matrix. This represents the process noise covariance matrix.

[0100] The specific observation equation for the observation update is as follows:

[0101]

[0102] in, Represents the observation function, , and This indicates the location coordinates for ultra-short baseline (USBL) acoustic positioning. This indicates the magnetic declination correction value.

[0103] The Jacobian matrix updated by observations is calculated using the following formula:

[0104] ;

[0105] ;

[0106] in, Represents the Jacobian matrix. This indicates the distance from the current location to the USBL array.

[0107] The formula for calculating the noise covariance of observation updates is:

[0108]

[0109] in, Represents the noise covariance. , , , , .

[0110] The Kalman gain is calculated using the following formula:

[0111]

[0112] in, This represents the Kalman gain matrix.

[0113] The state update is calculated using the following formula:

[0114]

[0115] in, This represents the innovation (residual) vector, the difference between the observed value and the predicted observed value. This represents the updated state vector.

[0116] Covariance update, the calculation formula is:

[0117]

[0118] in, This represents the updated covariance.

[0119] This intelligent calibration and remote transmission method significantly improves the accuracy and reliability of underwater positioning through multi-source information fusion, effectively overcoming the limitations of single signals (such as GPS failure underwater). The application of extended Kalman filtering enables optimal estimation of dynamic position, ensuring the accuracy of trajectory control. Simultaneously, combining path planning with real-time obstacle avoidance allows the system to not only navigate according to predetermined tasks but also intelligently handle unknown obstacles, greatly enhancing its autonomous operation capability and safety in complex and dynamic aquatic environments. This integrated solution provides a stable and reliable mobile platform foundation for subsequent water quality monitoring.

[0120] Specifically, the system utilizes sensing and sensor units to collect multiple water quality data and multiple water body data of the target water area, and utilizes computing and communication units to process the multiple water quality data and multiple water body data to obtain processing results, including:

[0121] The sensing and sensor unit collects multiple water quality data and corresponding multiple water body data in the target water area. The water quality data includes pH value, dissolved oxygen, conductivity and chlorophyll value, and the water body data includes temperature and turbidity.

[0122] The computing and communication unit performs sliding window feature expansion on multiple water quality data and corresponding multiple water body data to obtain water quality feature vectors;

[0123] The isolated forest algorithm is used to process water quality feature vectors to obtain global anomaly scores and local contribution vectors.

[0124] This application provides an intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer. This method first uses a sensing and sensor unit to simultaneously collect multi-dimensional water quality parameters such as pH, dissolved oxygen, conductivity, and chlorophyll content, as well as key water body physical parameters such as temperature and turbidity from the target water area. Subsequently, a computing and communication unit uses a sliding window technique to extract and expand features from continuous time-series data, constructing a water quality feature vector that reflects short-term change patterns. Finally, the system introduces an unsupervised isolated forest machine learning algorithm to analyze the constructed feature vector. This not only calculates a global anomaly score representing the overall degree of data anomaly but also outputs a local contribution vector revealing the contribution of each specific parameter to the anomaly, thereby achieving a deep assessment of water quality status and preliminary identification of abnormal events.

[0125] The collected data is packaged into a vector with timestamps, and the calculation formula is as follows:

[0126]

[0127] in, Represents the sampling vector. Represents a timestamp. Indicates pH value, Indicates dissolved oxygen. Indicates electrical conductivity. Indicates chlorophyll, Indicates temperature. Indicates turbidity.

[0128] The sliding window statistics are calculated using the following formula:

[0129] ;

[0130] ;

[0131] in, This represents the vector of mean values ​​for all parameters within the window. This represents the variance vector of each parameter within the window. Indicates the size of the sliding window. Let i represent the sampling vector at time i.

[0132] The sampled vector is standardized using the following formula:

[0133]

[0134] in, This represents the standardized sample vector.

[0135] Anomaly scores are calculated using a lightweight isolated forest model, using the following formula:

[0136]

[0137] in, Indicates the global anomaly score. This represents the expected value of the path length across all isolated trees. Indicates the subsample size. This represents the normalization factor.

[0138] ,when At this time, it may be abnormal. At this time, it may be normal, when At that time, there were no obvious abnormalities.

[0139] The local contribution vector is calculated using the following formula:

[0140]

[0141] in, Let represent the contribution of the i-th parameter to the anomaly score, satisfying ... , This represents the reduction in path length caused by splitting at the i-th parameter.

[0142] This intelligent calibration and remote transmission method first transforms time-series data into feature vectors rich in variation information through sliding window feature expansion, enhancing its ability to capture gradual, periodic, or complex anomaly patterns and improving the sensitivity and predictability of anomaly detection. Second, employing the unsupervised algorithm of isolated forests, it can automatically adapt to the data distribution of different water bodies without relying on a large amount of pre-labeled normal or anomalous data, demonstrating good adaptability. Finally, combining the algorithm's output global anomaly score with the local contribution vector not only determines whether an anomaly has occurred but also locates the main water quality parameters causing the anomaly, providing precise guidance for subsequent root cause analysis, fault diagnosis, and targeted sampling, achieving a leap from "determining whether an anomaly exists" to "analyzing what kind of anomaly."

[0143] Specifically, a preset abnormal event threshold is established, and the processing result is compared with the abnormal event threshold to issue a corresponding abnormal event alarm, including:

[0144] Preset thresholds for pH, dissolved oxygen, conductivity, chlorophyll, temperature, and turbidity. Compare these thresholds with the following values ​​to obtain multiple first comparison results: pH, dissolved oxygen, conductivity, chlorophyll, temperature, turbidity, and the corresponding thresholds for pH, dissolved oxygen, conductivity, chlorophyll, temperature, and turbidity.

[0145] A preset abnormal score threshold is set, and a second comparison result is obtained by comparing the global abnormal score with the abnormal score threshold. Based on multiple first comparison results, second comparison results and local contribution vectors, corresponding abnormal event alarms are issued.

[0146] This application provides an intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer. This method first presets independent thresholds for each key water quality parameter (including pH, dissolved oxygen, conductivity, chlorophyll content, temperature, and turbidity). Real-time sensor data is compared one by one with the corresponding thresholds to form a "first comparison result" reflecting whether a single parameter exceeds its limit. Simultaneously, a comprehensive "anomaly score threshold" is preset, and the global anomaly score calculated by the isolated forest algorithm is compared with this threshold to form a "second comparison result" reflecting whether the overall data pattern is abnormal. Finally, the system comprehensively considers the single parameter exceeding the limit result (first comparison result), the overall anomaly score judgment result (second comparison result), and the contribution of each parameter to the anomaly (local contribution vector), making collaborative decisions and issuing anomaly event alarms with different directions and confidence levels.

[0147] The preset threshold values ​​are: pH 5.5-9.2, dissolved oxygen 2, conductivity ±100% of background value, chlorophyll 25, temperature ±8, and turbidity 30.

[0148] A pH abnormality alarm will be issued when the pH value is <5.5 or >9.2, with a weight of 0.2; a dissolved oxygen abnormality alarm will be issued when the dissolved oxygen value is <2, with a weight of 0.25; a conductivity abnormality alarm will be issued when the conductivity exceeds ±100%, with a weight of 0.15; a chlorophyll value >25 will be issued when the chlorophyll value is >25, with a weight of 0.15; a temperature abnormality alarm will be issued when the temperature exceeds ±8℃, with a weight of 0.1; and a turbidity abnormality alarm will be issued when the turbidity value is >30, with a weight of 0.15.

[0149] Six typical pollution modes were defined: when pH < 5.5, dissolved oxygen < 2, conductivity > ±100%, and temperature is normal, it is considered industrial acidic wastewater with a weight of 1.2; when chlorophyll value > 25, dissolved oxygen < 2, and daytime pH > 9.2, it is considered eutrophication with a weight of 1.1; when dissolved oxygen < 2, turbidity > 30, and conductivity > ±100%, it is considered organic pollution with a weight of 1.3; when dissolved oxygen < 2 and temperature exceeds 8℃, it is considered thermal pollution with a weight of 1.0; when conductivity exceeds -100% and temperature exceeds ±8℃, it is considered freshwater intrusion with a weight of 1.0; when any single parameter changes drastically within a short period of time, it is considered chemical leakage with a weight of 1.5.

[0150] The comprehensive abnormality score is calculated using the following formula:

[0151]

[0152] in, Indicates the overall abnormal score. , and Indicates the weighting coefficient. This represents the largest difference among the six original data sets. This represents the product of the global anomaly score and the weight of the corresponding contamination mode. This indicates the similarity between the current abnormal event and the preset contamination pattern.

[0153] The preset anomaly score thresholds are 0.55, 0.70, and 0.85. When the overall anomaly score > 0.85, a Level 1 alarm (emergency pollution event) is triggered; when 0.85 > overall anomaly score > 0.70, a Level 2 alarm (significant anomaly event) is triggered; when 0.70 > overall anomaly score > 0.55, a Level 3 alarm (minor anomaly event) is triggered; and when 0.55 > overall anomaly score, no alarm is triggered.

[0154] This intelligent calibration and remote transmission method combines a single-parameter threshold method based on fixed rules with a holistic anomaly score method based on data-driven pattern recognition, forming a dual verification. This effectively avoids false alarms caused by occasional drift of a single sensor (threshold method misses) or missed alarms caused by interference from complex environmental factors (model method misjudgment), significantly improving the accuracy of alarms. Simultaneously, the introduction of a local contribution vector allows the alarm to not only indicate "an anomaly has occurred" but also further explain "which parameter(s) are the main contributing factors," achieving interpretability and operability of alarm information. This provides direct and powerful decision support for subsequent fault diagnosis, preliminary pollution type assessment, and emergency response strategy formulation.

[0155] Specifically, multiple water samples from the target water area are collected through a sampling and analysis unit, and preprocessing and data analysis are performed on these samples to obtain multiple first analysis results for the target water area, including:

[0156] The sampling and analysis unit collects water samples from the area in the target water body where an abnormal event alarm was generated.

[0157] The water sample is filtered, digested online, and enriched to obtain the treated water sample;

[0158] The first analytical results were obtained by colorimetric, electrochemical and optical detection of the treated water sample.

[0159] This application provides an intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer, which offers a mechanism for rapid on-site verification and in-depth analysis of abnormal water quality events. Specifically, when the system triggers an abnormal event alarm, the sampling and analysis unit immediately and automatically collects on-site water samples from the target water area corresponding to the alarm. Subsequently, the sample undergoes a standardized online pretreatment process within the device, including physical filtration, online chemical digestion, and target enrichment to remove interfering substances and concentrate the analyte components, obtaining a high-quality treated water sample. Finally, the system simultaneously or sequentially applies multiple analytical techniques, such as colorimetry, electrochemical methods, and optical detection, to perform qualitative and quantitative analysis of specific pollutants (such as heavy metals, nutrients, and organic pollutants), generating accurate, multi-index laboratory-grade "first analysis results," thereby achieving verification and in-depth analysis of abnormal data from in-situ sensors.

[0160] When the computing and communication unit triggers an abnormal event alarm, it immediately switches to "high-precision positioning mode", which increases the grid resolution of multi-source fusion positioning from 10 meters to 1 meter, and combines real-time water flow data to correct its own position, ensuring that it can accurately return to the abnormal coordinate point, with the positioning error controlled within ±0.5 meters.

[0161] Based on the local contribution vector of the abnormal parameters, a three-dimensional sampling array is planned: if the contribution of chlorophyll value and turbidity is high, vertical stratified dense sampling is adopted, with one sampling point every 0.5 meters in the water depth of 0-5 meters; if the contribution of pH value and conductivity is high, horizontal radial sampling is adopted, with 8 sampling points evenly distributed on concentric circles with radii of 1, 3 and 5 meters centered on the abnormal point; if the contribution of dissolved oxygen and temperature is high, key sampling is adopted above and below the thermocline.

[0162] After water samples were collected, an online three-stage gradient filtration system was used for filtration. Stage 1: Coarse filtration (anti-clogging): The sample passed through a stainless steel screen (100μm pore size) with automatic reverse cleaning, trapping large particles and algal clumps. A pressure differential sensor monitored the screen in real time; when the pressure difference ΔP > 5kPa, reverse water flow (0.5 seconds) and mechanical vibration were activated for self-cleaning. Stage 2: Fine filtration (particulate removal): A cellulose ester membrane belt (0.45μm pore size) with automatic winding function was used. The membrane belt moved slowly at a speed of 0.1mm / s to ensure fresh membrane surface for each filtration, avoiding clogging and cross-contamination. The effective filtration area A = 1.0, and the design flux J = 10. Stage 3: Ultrafiltration: For projects requiring the determination of dissolved metals or organic matter, a tangential flow ultrafiltration module (molecular rejection 10 kDa) was activated, operating at low pressure (<50 kPa) with a recovery rate >90%. The filtered sample immediately enters the dispensing valve, with one portion (approximately 10 mL) going directly into the "Instant Analysis Flow Path" and the other portion (approximately 40 mL) going into the "Deep Processing Flow Path".

[0163] Online digestion is performed for total phosphorus (TP) and total nitrogen (TN). The digestion reactor is a high-temperature, high-pressure resistant miniature titanium alloy reaction tube (internal volume...). It is wrapped with a heating wire and a platinum resistance temperature sensor (PT100).

[0164] Digestion procedure: Chemical addition: Using a precision syringe pump, add potassium persulfate solution sequentially to the reaction tube. (Final concentration approximately 0.015M). Sodium hydroxide solution (to adjust pH to alkaline, for TN) or sulfuric acid solution (to adjust pH to acidic, for TP). Add sample .

[0165] Sealing and Heating: The reaction tube is mechanically sealed, and the temperature is programmed under pressure sensor monitoring: First stage: Increase from room temperature to 105°C within 3 minutes, hold for 5 minutes to expel dissolved oxygen. Second stage: Increase to 120°C within 5 minutes and hold. minutes, digestion time It can be adjusted as needed, and the system's built-in algorithm dynamically optimizes based on historical digestion efficiency (assessed through spiked recovery rate).

[0166] Cooling and Transfer: After the procedure is completed, the Peltier semiconductor cooler cools the reaction tube to below 30°C within 2 minutes. After depressurization, the digestion solution is transferred to the detection cell.

[0167] For online enrichment of trace heavy metals, the process involves: First, solid-phase extraction microcolumn: using a 2 mm inner diameter, 30 mm length PEEK tube filled with 50 mg of Chelater Resin (such as Chelex-100 or iminodiacetic acid type resin). Second, column pretreatment: passing the microcolumn sequentially with high-purity water and pH buffer (such as 0.1 M ammonium acetate, pH 5.5) to activate the functional groups. Sample loading and adsorption: loading the filtered sample at a low flow rate... The sample is pumped through a microcolumn. The sample volume is dynamically determined based on the target detection limit.

[0168]

[0169] in, Indicates the sample volume. Indicates the allocation coefficient. Indicates resin capacity.

[0170] Washing: Wash away adsorbed alkali metal and alkaline earth metal ions with a small amount of weak acid or chelating agent solution (such as 0.01 MH NO3). Elution: Quantitatively elute the target heavy metal ions with a small amount of strong acid, and then introduce them into the subsequent electrochemical detection cell.

[0171] Colorimetric detection (for nutrients: nitrate NO3⁻, nitrite NO2⁻, phosphate PO4³⁻, etc.). Sequential injection analysis (SIA) or microfluidic chip technology is employed. Taking nitrate detection as an example (cadmium column reduction-naphthylethylenediamine azo colorimetric method):

[0172] Reduction: The sample is mixed with ammonium chloride buffer and flows through a miniaturized cadmium reduction column (filled with cadmium particles) to quantitatively reduce NO3⁻ to NO2⁻. Diazotization and Coupling: The effluent is mixed with sulfonamide reagent (in acidic medium) to generate a diazonium salt, which is then coupled with N-(1-naphthyl)-ethylenediamine dihydrochloride (NED) to generate a purple-red azo dye. Photometric Detection: The reaction solution flows into a Z-shaped long-pathway micro-detector cell. Illumination is performed by an LED light source with a center wavelength of 540 nm. The absorbance is detected by a photodiode and calculated using Lambert-Beer's law. The calculation formula is as follows:

[0173]

[0174] in, Indicates absorbance. Indicates the molar absorptivity. Indicates the concentration of the analyte. This indicates a Z-type long optical path.

[0175] The system uses multi-channel selector valves and precise timing control to sequentially detect NO3⁻, NO2⁻, PO4³⁻ and other items within 10-15 minutes.

[0176] Electrochemical detection (for heavy metals: Pb²⁺, Cd²⁺, Cu²⁺, etc.). Differential pulsed anodic stripping voltammetry (DPASV) is employed, offering high sensitivity (down to the nM level). Pre-electrolytic enrichment: Under stirring, heavy metal ions are reduced to metals at a constant potential, forming amalgam or bismuth amalgam. The diffusion field is homogenized after 10 seconds. Stripping scan: A positive differential pulse voltage is applied, causing the metal to oxidize and dissolve, producing a characteristic oxidation current peak. Quantitative analysis: Each heavy metal ion exhibits a current peak at a specific potential, with the peak current proportional to its concentration.

[0177] Optical detection (for chlorophyll a, CDOM, turbidity, etc.). Multispectral fluorescence detection: fluorescence intensity is detected at 685 nm using LED excitation with a center wavelength of 470 nm. A pre-established, turbidity-corrected calibration curve is used. Calculate the concentration.

[0178] Colored dissolved organic matter (CDOM): The relative content of CDOM was characterized by excitation with a UV LED at 350 nm and detection of fluorescence intensity at 450 nm.

[0179] The process for generating the first analytical result includes: Full quality control and a process blank: A high-purity water sample is inserted as a process blank for each analytical sequence to correct for reagent background and system contamination. Parallel sample analysis: At least one sample from each batch is analyzed in duplicate, with a relative deviation requirement of <10%. Spike recovery: Periodically (or in case of abnormal events), a portion of the samples are spiked with a standard substance of known concentration, and the recovery rate is calculated (required to be 80-120%) to verify the accuracy of the method.

[0180] Data structure of the first analysis results: After each analysis task is completed, the sampling and analysis unit generates a structured "first analysis results" data packet, including:

[0181] Sample Information: Sampling time, precise coordinates (x, y, z), water depth, water temperature, pH. Detection Results List: Value, unit, detection limit, and uncertainty for each test item (e.g., total phosphorus, lead ion concentration). Quality Identification: Data quality level (A / B / C / D), anomaly marker, analytical method used, instrument status code.

[0182] This intelligent calibration and remote transmission method achieves closed-loop verification on-site, from "anomaly detection" to "cause confirmation," significantly improving the reliability and decision-making value of monitoring results. Its automated and integrated online preprocessing and analysis process transforms raw water samples into stable, directly detectable samples without manual intervention, avoiding sample deterioration, time delays, and operational errors caused by traditional post-sampling laboratory analysis. Combining colorimetric, electrochemical, and optical detection principles, it can cover a wider range of pollutant types, providing more accurate and comprehensive chemical composition information than a single in-situ sensor. This serves as both an authoritative verification of sensor data and a direct revelation of the types and concentrations of potential pollutants causing anomalies, providing crucial and immediate factual evidence for the characterization, classification, and emergency response of pollution incidents.

[0183] Specifically, when an abnormal event alarm is triggered, multiple sensing and sensor units and multiple sampling and analysis units exchange multiple water quality data, multiple water body data, and multiple first analysis results, including:

[0184] When an abnormal event alarm is triggered, the autonomous navigation and positioning unit at the trigger point broadcasts a request signal to the target water area;

[0185] Upon receiving the request signal, multiple sensing and sensor units and multiple sampling and analysis units exchange multiple water quality data, multiple water body data, and multiple first analysis results in a unified data format.

[0186] This application provides an intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer. This intelligent calibration and remote transmission method defines a set of efficient distributed data collaborative exchange protocols. The core process is as follows: when a monitoring node triggers an alarm due to an abnormal event, the node's autonomous navigation and positioning unit, as the initiator, immediately broadcasts a signal containing its own location and request information to all other working nodes within the monitoring network coverage area. All nodes in the network that receive this broadcast request signal (including other sensing and sensor units, sampling and analysis units) respond to this request and, according to a predefined and unified data format, exchange their collected water quality data, water body data, and completed first analysis results, etc., point-to-point or multi-point, through communication units, thereby quickly aggregating a multi-node, multi-type data set around the abnormal event point.

[0187] This intelligent calibration and remote transmission method, triggered by broadcast requests and using a unified format for data exchange, enables the system to automatically organize a small-scale, highly correlated data synchronization within a very short time after an anomaly occurs, breaking down data silos between monitoring nodes. This provides an indispensable multi-source, heterogeneous data foundation for subsequent comparative analysis, fault diagnosis, and consensus computation, greatly enhancing the system's overall cognitive ability and response speed to local events. It is a key prerequisite for achieving distributed intelligent diagnosis and collaborative decision-making. This design also gives the system excellent scalability, allowing new nodes to seamlessly integrate with the collaborative protocol.

[0188] Specifically, the fault nodes are identified by comparing and analyzing multiple water quality data, multiple water body data, and multiple first analysis results using computing and communication units. Autonomous calibration of the fault nodes is then performed, including:

[0189] The computing and communication unit uses the Byzantine fault-tolerant consensus algorithm to calculate the water quality consensus result, water body consensus result, and first analysis consensus result of multiple water quality data, multiple water body data, and multiple first analysis results.

[0190] Preset water quality consensus threshold, water body consensus threshold and first analysis consensus threshold, and compare the water quality consensus result, water body consensus result and first analysis consensus result with the water quality consensus threshold, water body consensus threshold and first analysis consensus threshold respectively to obtain the fault node;

[0191] Perform soft and hard calibration on the faulty nodes.

[0192] This application provides an intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer. This intelligent calibration and remote transmission method adopts the Byzantine fault-tolerant consensus algorithm in the field of distributed systems to achieve reliable verification of multi-source monitoring data and accurate location of faulty nodes. The specific steps are as follows: First, the computing and communication unit takes the collected water quality data, water body data, and first analysis results from multiple nodes as independent input sets, runs the Byzantine fault-tolerant consensus algorithm, and calculates the water quality consensus result, water body consensus result, and first analysis consensus result that represent the data status of most reliable nodes. Then, the system presets consensus thresholds for various types of data and compares the raw data reported by each node with the corresponding type of consensus result. If the deviation exceeds the preset threshold, the node is determined to be a suspected faulty node. Finally, for the identified faulty node, the system executes an autonomous calibration process, including software parameter correction based on the consensus result (soft calibration) or triggering hardware maintenance commands such as physical cleaning and calibration (hard calibration), thereby restoring the node to its normal working state.

[0193] Byzantine Fault Tolerant Consensus Algorithm Implementation Process:

[0194] First, local data preparation is performed. It aggregates all multi-source data collected within the most recent consensus cycle, including multiple water quality data points, multiple water body data points, and multiple initial analysis results. This data is packaged into a structured local data packet. To ensure data authenticity and non-repudiation, the node generates a digital signature for this data packet using its private key. Subsequently, the node broadcasts or multicasts the signed local data packet to all other consensus-participating nodes in the network via the underwater acoustic mesh network. Simultaneously, each node also begins receiving data packets from other nodes.

[0195] For each water quality parameter to be agreed upon (e.g., dissolved oxygen concentration), a node extracts all relevant measurements (including its own) from the validated dataset. To mitigate the impact of up to f faulty nodes, the node first sorts all received values ​​for that parameter. Then, it removes the f largest and f smallest values. This step is crucial; it assumes that the most extreme erroneous values ​​will appear at the extremes of the sequence, and removing them filters out malicious or severely faulty data to a great extent. For the remaining (N-2f) values, instead of a simple arithmetic average, the node performs a weighted average based on quality scores. Data from nodes with good historical performance and current good status have higher quality scores and are given greater weight in the final average. Conceptually, the calculation formula is: multiply each remaining value by the combined weight of the sending node (determined by quality scores, etc.), sum them, and then divide by the sum of all weights. This process is repeated for each water quality parameter, until each node independently calculates the same set of consensus values ​​(among honest nodes). This set represents the network's collective, interference-resistant judgment of the water quality status at the current point in time. After the consensus value is calculated, each node will perform a final round of comparison to identify which nodes may be faulty or malicious.

[0196] Each node compares its own raw measurements (or raw values ​​received from other nodes) with the consensus value reached, calculating the relative deviation for each data point. For example, it calculates the absolute difference between the pH value reported by a node and the consensus value for that parameter, then divides it by the consensus value to obtain a percentage deviation. If the relative deviations of multiple key parameters of a node exceed a preset maximum tolerance threshold, and its deviation pattern significantly and consistently deviates from the network consensus, then the node is marked as a "suspected faulty node."

[0197] Based on the severity of the fault, the system will take tiered measures: For minor or first-time deviations, the system will send a command to the node to trigger its internal self-calibration procedure (such as releasing the standard solution to recalibrate the sensor). After calibration, the node must remeasure and report the data to verify whether it has returned to normal. For nodes with severe deviations or that refuse calibration, the network will temporarily isolate them. This means that in subsequent consensus rounds, other nodes will reduce or even ignore the data weight from this node, or directly not invite it to participate in consensus voting, to prevent it from continuing to pollute the consensus results. At the same time, the node will be marked and manual maintenance is recommended. If organized and coordinated malicious attacks are detected (such as multiple nodes colluding to provide inconsistent erroneous data), the algorithm will trigger a high-level security alert and upload relevant evidence logs to the control center.

[0198] This intelligent calibration and remote transmission method, by introducing Byzantine fault-tolerant consensus, enables the system to reach a correct and consistent consensus result based on data from the majority of normal nodes, even when some nodes (including sensors or analysis units) provide erroneous, abnormal, or even malicious data. This effectively resists "Byzantine faults." This not only achieves high-precision, automated diagnosis of faulty nodes or data anomalies, avoiding the impact of a single node failure on overall judgment, but also significantly improves the long-term operational stability and data output quality of the entire monitoring network in unattended environments. The hardware and software combined calibration strategy allows the system to intelligently take recovery measures based on the fault type, achieving a complete autonomous closed loop from "problem detection" to "problem resolution."

[0199] Specifically, the computing and communication unit is used to perform spatiotemporal alignment and data fusion of multiple water quality data, multiple water body data, and multiple first analysis results to obtain a first comprehensive water quality field, including:

[0200] The computing and communication unit establishes a unified time reference and performs cubic spline interpolation on multiple water quality data, multiple water body data, and asynchronous data from multiple first analysis results;

[0201] Convert multiple water quality data, multiple water body data, and location data from multiple first analysis results into the UTM coordinate system;

[0202] The first comprehensive water quality field is obtained by using confidence-weighted fusion of multiple water quality data, multiple water body data, and multiple first analysis results.

[0203] This application provides an intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer. Firstly, in the time dimension, the computing and communication units establish a unified time reference and use cubic spline interpolation to perform high-precision interpolation and alignment of time-series data from different nodes with incompletely synchronized acquisition times, ensuring that all data are comparable on the time axis. Secondly, in the spatial dimension, the system uniformly transforms the location information recorded in all data into the UTM (Universal Transverse Mercator) coordinate system, eliminating spatial misalignment caused by using different geographical coordinate systems. Finally, based on information such as the reliability of the data source and the accuracy of the sensor, a confidence weight is assigned to each data point, and a weighted fusion algorithm is used to integrate multi-source, heterogeneous but spatiotemporally aligned water quality data, water body data, and analysis results to generate a spatiotemporally continuous and information-consistent "first comprehensive water quality field."

[0204] The system unifies timestamps from different nodes, based on their respective local clocks, onto a single absolute time base. GPS time (UTC) or Network Time Protocol (NTP) synchronization time is used as the primary time base.

[0205] Local clock drift is modeled as a linear model, and the calculation formula is as follows:

[0206]

[0207] in, Indicates the primary time base. Indicates the local clock of each node. Indicates the clock frequency drift coefficient. Indicates the clock offset. This represents random error.

[0208] Cubic spline interpolation unifies the time axis, interpolating all asynchronous data onto a unified, equally spaced time axis. The flow of the cubic spline interpolation algorithm is as follows:

[0209] Construct a cubic spline function for the time series of each parameter. The spline conditions are as follows: , exist Second-order continuous differentiability, natural boundary conditions: .

[0210] in, Represents a cubic spline function. Indicates the analysis time window.

[0211] The spline function is in the form of: in the interval superior,

[0212]

[0213] in, , , and Represents the coefficient.

[0214] Interpolation calculation and quality assessment: For each uniform time point, calculate the interpolation:

[0215]

[0216] in, Indicates a unified point in time. This indicates interpolation.

[0217] The interpolation uncertainty estimate is calculated using the following formula:

[0218]

[0219] in, This represents the variance of the original measurement error. Represents the average value of the original timestamps This indicates the interpolation uncertainty.

[0220] The specific process of confidence-weighted data fusion is as follows:

[0221] Define the neighborhood and spatial radius: Time radius: .

[0222] in, Indicates spatial radius, Indicates the time radius. Indicates grid spacing. , .

[0223] For grid point g, search for all data points d that satisfy the following conditions. k :

[0224]

[0225] in, Represents a spatial location vector.

[0226] The formula for calculating data quality weights is:

[0227]

[0228] in, This represents the original quality score of data point k. This indicates the data quality weight.

[0229] The formula for calculating spatial distance weight is:

[0230]

[0231] in, Indicates spatial distance weights. Indicates spatial scale parameters.

[0232] The formula for calculating the time distance weight is:

[0233]

[0234] in, Indicates the time distance weight. Indicates the time scale parameter.

[0235] The formula for calculating data type weights is:

[0236]

[0237] in, Indicates the weight of the data type.

[0238] The formula for calculating the overall weight is:

[0239]

[0240] in, Indicates the overall weight. .

[0241] The formula for calculating the weighted mean in weighted fusion and uncertainty estimation is as follows:

[0242]

[0243] in, This represents the weighted average.

[0244] The formula for calculating the weighted variance is:

[0245]

[0246] in, This represents the weighted variance.

[0247] The specific steps for generating the first comprehensive water quality field are as follows:

[0248] The system iterates through all spatiotemporal grid points, repeating the fusion calculation in step two for each point, ultimately generating a four-dimensional data field covering the entire analysis area and time period. This first comprehensive water quality field contains the following information for all monitored parameters (pH, dissolved oxygen, conductivity, etc.) at each grid point: weighted average estimate, estimated variance (uncertainty), effective sample size, and supplementary quality score based on interpolation distance.

[0249] A final overall quality score is calculated for each grid point. This score is the product of three sub-scores: a score based on the number of valid samples (more samples, higher score), a score based on the estimated variance (smaller variance, higher score), and a score based on the interpolation distance (closer to the original data point, higher score). The final score is quantified as a value between 0 and 1 and divided into four levels: A (high confidence), B (medium confidence), C (low confidence), and D (speculation), providing a reliability basis for subsequent analysis.

[0250] The system calculates the overall "coverage" of the data field, i.e., the proportion of grid points with valid estimates out of the total number of grid points, and identifies the largest continuous area without data. These metrics reflect the completeness of the monitoring network. All metadata from the processing, including the list of nodes participating in the fusion, the weight coefficients of each node, time correction parameters, coordinate transformation parameters, algorithm version, etc., is fully recorded and appended to the data field to ensure the traceability and repeatability of the results. Thus, a complete, continuous, and uncertainty-quantified integrated water quality field with quality labels is constructed, providing a fused and purified core data foundation for the subsequent spatial interpolation and the generation of second-order analysis results.

[0251] This intelligent calibration and remote transmission method effectively solves the problem of data spatiotemporal mismatch caused by independent node operation, sampling frequency, and differences in positioning systems through high-precision spatiotemporal alignment (cubic spline interpolation and unified UTM coordinates), laying a solid foundation for subsequent analysis and modeling. The confidence-weighted fusion strategy is an intelligent information integration method that differentiates data from different sources and with varying reliability, giving higher weight to high-confidence data in the fusion result. This effectively suppresses the impact of noise and low-quality data while preserving data diversity. The resulting "First Integrated Water Quality Field" is a high-quality, highly consistent spatial dataset, providing reliable input for more advanced gap-filling and spatial analysis.

[0252] Specifically, the second comprehensive water quality field is obtained by using interpolation to fill the spatial monitoring gaps in the first comprehensive water quality field, including:

[0253] Multiple constraints are defined, and a second integrated water quality field is generated based on these constraints using the Kriging space interpolation method.

[0254] This application provides an intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer. Based on the generated first comprehensive water quality field (composed of discrete monitoring point data after spatiotemporal alignment and fusion), the system first defines and incorporates various constraints reflecting the physicochemical laws of the aquatic environment (such as water temperature gradient, water flow direction, pollutant diffusion model, etc.). Subsequently, based on these multidimensional constraints, the Kriging space interpolation method in geostatistics is used to make an optimal and unbiased estimate of the spatial "gap" between monitoring points, and finally generates a "second comprehensive water quality field" that covers the entire target water area, is spatially continuous, and has reasonable physical meaning.

[0255] First, a scanning analysis of the first comprehensive water quality field was conducted, identifying two types of areas: one is the effective data area, which consists of grid points supported by direct measurement data; the other is the monitoring gap area, which consists of grid points with zero effective samples or below the quality threshold. The gap area is usually formed due to uneven distribution of bionic fish, communication interruptions, or terrain obstruction.

[0256] Overall coverage is calculated as the proportion of valid data grid points to the total number of grid points. If the coverage is > 90%, only local fine interpolation is needed; if it is between 70% and 90%, medium-scale interpolation is required; if it is < 70%, large-scale interpolation is initiated and may trigger an alarm indicating insufficient monitoring network density. Spatial distribution of gaps is identified as either discrete small voids or large, contiguous areas. The system uses a three-dimensional connected component analysis algorithm to label each independent gap region and record its volume, surface complexity, and boundary contact area with the valid data region. Interpolation priority is ranked according to the principles of "edges first, then centers" and "smallest first, then largest." Small gaps adjacent to multiple valid data regions are prioritized because they have more interpolation constraints and more reliable results.

[0257] Specific process of spatial interpolation based on Kriging:

[0258] The experimental variability function is calculated by pairing effective data points together and calculating the square of the difference between their spatial distances and parameter values. All closely spaced pairs of points are grouped into the same distance bin, and the mean semivariogram within that bin, i.e., the experimental variability function value, is calculated using the following formula:

[0259]

[0260] in, This represents the experimental variation function value. This represents the number of point pairs with a spacing of h. This indicates the parameter value at position x.

[0261] The system fits the experimental variogram values ​​to a theoretical model, most commonly the spherical model. The model formula is:

[0262]

[0263] in, This represents the semivariance when the distance is zero. This represents the variance caused by spatially dependent structures. The maximum distance at which spatial autocorrelation exists. This represents the theoretical variation function.

[0264] The specific steps for generating the optimal unbiased estimate from the Kriging equations are as follows: Construct the Kriging equations:

[0265]

[0266] in, This represents the theoretical semivariance between data points i and j. This represents the theoretical semivariance between the interpolation point and the data point i. Let represent the Lagrange multipliers. Solving the above system of linear equations yields the optimal weights and multipliers.

[0267] Calculate the parameter estimates for the interpolation points. .

[0268] Various constraints and anisotropic correction for water flow direction: The system acquires the current water flow velocity and direction data. When calculating spatial distances, it does not simply use Euclidean distance, but instead uses anisotropic distance corrected for the water flow field. Along the water flow direction, material diffusion is faster and the correlation is stronger. Therefore, the system "stretches" the spatial scale in the downstream direction and "compresses" the spatial scale in the perpendicular direction. This is achieved by defining an anisotropic transformation matrix, which is integrated into the variogram calculation and neighborhood search, making the interpolation results more consistent with the physical laws of pollutant diffusion.

[0269] Boundary Constraints: Land / Shoreline Boundaries: For interpolation points near the shoreline, the system imposes "zero flux" or "fixed value" boundary conditions. For example, the gradient of certain conservative substance concentrations along the shoreline normal is forced to zero. Surface and Bottom Boundaries: At the surface, the system considers the effects of atmospheric exchange; at the bottom, it considers potential release or adsorption at the sediment-water interface. These are achieved by introducing virtual boundary data points or adjusting the variogram model near the boundary.

[0270] The specific steps for generating the second integrated water quality field are as follows:

[0271] For each spatiotemporal grid point, the data field includes: Optimal estimate: for the original valid points, it is a weighted fusion value; for the interpolation points, it is a Kriging estimate; Total uncertainty: it combines measurement error (for valid points) and Kriging variance (for interpolation points); Data source label: it clearly marks each value as "measured fusion", "Kriging interpolation", or "constrained interpolation"; Interpolation contribution: it records the weight proportion of the interpolation point estimate from the most recent measured point.

[0272] The system automatically generates an uncertainty field layer and a data source confidence layer to complement the second integrated water quality field. Uncertainty field: Overlaid on the concentration field as a semi-transparent cloud map or contour lines, visually displaying which areas have high certainty (light color / transparent) and which areas have high uncertainty (dark color / opaque). Confidence layer: Uses different colors or textures to distinguish between "high-confidence measured areas," "medium-confidence interpolation areas," and "low-confidence extrapolation areas," providing decision-makers with clear quality indicators.

[0273] Compared to traditional simple interpolation (such as inverse distance weighting), this intelligent calibration and remote transmission method not only considers the spatial correlation between data points but also incorporates prior domain knowledge into the interpolation process through various custom constraints (such as integrating hydrodynamic or ecological models). This allows the generated second comprehensive water quality field to not only smoothly fill monitoring gaps but also reflect the actual migration and diffusion trends of pollutants in water bodies, the effects of temperature stratification, and other real physicochemical processes. This significantly improves the scientific rigor and accuracy of the spatial interpolation results, providing a higher-quality data foundation for comprehensive visual assessment of water quality, pollution source tracing, and trend prediction.

Claims

1. A portable multi-parameter water quality analyzer, characterized in that, The portable multi-parameter water quality analyzer includes: The main structure is a streamlined shell structure resembling a tuna; An autonomous navigation and positioning unit is located at the head of the main structure. The autonomous navigation and positioning unit is used to obtain the real-time position of the water quality analyzer, correct the real-time position, and plan a preset route. The propulsion and maneuvering unit is located at the tail and two fins of the main structure and is connected to the autonomous navigation and positioning unit. The propulsion and maneuvering unit is used to drive the water quality analyzer to navigate precisely according to the preset route, and at the same time to avoid obstacles in real time based on the real-time position. A sensing and sensor unit is located at the head and two gills of the main structure and is connected to the computing and communication unit. The sensing and sensor unit is used to collect multiple water quality data and multiple water body data of the target water area and send the water quality data and the water body data to the computing and communication unit. The sampling and analysis unit is located in the abdomen of the main structure and connected to the sensing and sensor unit. The sampling and analysis unit is used to collect multiple water samples from the target water area and perform preprocessing and data analysis on the multiple water samples to obtain multiple first analysis results of the target water area, and send the multiple first analysis results to the computing and communication unit. The computing and communication unit is located inside the abdominal cavity of the main structure and is connected to the sensing and sensor unit and the sampling and analysis unit. The computing and communication unit is used to preset anomaly thresholds, process multiple water quality data and multiple water body data to obtain processing results, and compare the processing results with the anomaly thresholds to issue corresponding anomaly alarms. The computing and communication unit is also used to compare and analyze multiple water quality data, multiple water body data and multiple first analysis results to obtain fault nodes, and to perform autonomous calibration on the fault nodes. The computing and communication unit is also used to obtain a first comprehensive water quality field based on multiple water quality data, multiple water body data and multiple first analysis results, use interpolation to fill the spatial monitoring gaps of the first comprehensive water quality field to obtain a second comprehensive water quality field, and process the second comprehensive water quality field to obtain the second analysis result.

2. A method for intelligent calibration and remote transmission of a portable multi-parameter water quality analyzer, characterized in that, The calibration and remote transmission method is applied to the water quality analyzer of claim 1, and the calibration and remote transmission method includes: The autonomous navigation and positioning unit acquires the real-time position of the water quality analyzer, corrects the real-time position and plans a preset route, and drives the water quality analyzer to navigate precisely according to the preset route through the propulsion and maneuvering unit, while simultaneously performing real-time obstacle avoidance based on the real-time position. The sensing and sensor unit collects multiple water quality data and multiple water body data of the target water area and sends the multiple water quality data and multiple water body data to the computing and communication unit. The sampling and analysis unit collects multiple water body samples of the target water area and performs preprocessing and data analysis on the multiple water body samples to obtain multiple first analysis results of the target water area. The multiple first analysis results are then sent to the computing and communication unit. The computing and communication unit processes multiple water quality data and multiple water body data to obtain the processing result, presets the abnormal event threshold, compares the processing result with the abnormal event threshold, and issues a corresponding abnormal event alarm. When the abnormal event alarm is triggered, the multiple sensing and sensor units and the multiple sampling and analysis units exchange multiple water quality data, multiple water body data, and multiple first analysis results respectively; the computing and communication unit compares and analyzes the multiple water quality data, multiple water body data, and multiple first analysis results to obtain the fault node, and performs the autonomous calibration on the fault node; The computing and communication unit is used to perform spatiotemporal alignment and data fusion on multiple water quality data, multiple water body data and multiple first analysis results to obtain the first comprehensive water quality field. The interpolation method is used to fill the spatial monitoring gaps of the first comprehensive water quality field to obtain the second comprehensive water quality field. The second comprehensive water quality field is processed to obtain the second analysis result.

3. The intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer according to claim 2, characterized in that, The process of acquiring the real-time position of the water quality analyzer using the autonomous navigation and positioning unit, correcting the real-time position and planning a preset route, and driving the water quality analyzer to navigate precisely according to the preset route via the propulsion and maneuvering unit, while simultaneously performing real-time obstacle avoidance based on the real-time position, includes: The autonomous navigation and positioning unit collects GPS signals, acoustic positioning signals, diving depth, acceleration, and heading angle. An extended Kalman filter is used to fuse the GPS signals, acoustic positioning signals, diving depth, acceleration, and heading angle to obtain fused positioning data. The autonomous navigation and positioning unit plans the route based on the starting and ending points of the route, and the propulsion and maneuvering unit performs real-time obstacle avoidance based on the fused positioning data and the route.

4. The intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer according to claim 2, characterized in that, The process of collecting multiple water quality data and multiple water body data of the target water area using the sensing and sensor unit, and processing the multiple water quality data and multiple water body data using the computing and communication unit to obtain the processing result includes: The sensing and sensor unit collects multiple water quality data and corresponding multiple water body data in the target water area, wherein the water quality data includes pH value, dissolved oxygen, conductivity and chlorophyll value, and the water body data includes temperature and turbidity; The computing and communication unit performs sliding window feature expansion on the plurality of water quality data and the corresponding plurality of water body data to obtain a water quality feature vector; The water quality feature vector is processed using the isolated forest algorithm to obtain the global anomaly score and the local contribution vector.

5. The intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer according to claim 4, characterized in that, The step of setting a preset abnormal event threshold and comparing the processing result with the abnormal event threshold to issue a corresponding abnormal event alarm includes: A preset threshold values ​​for pH, dissolved oxygen, conductivity, chlorophyll, temperature, and turbidity are used to compare the pH, dissolved oxygen, conductivity, chlorophyll, temperature, and turbidity with the respective threshold values ​​to obtain multiple first comparison results. A preset anomaly score threshold is set, and a second comparison result is obtained by comparing the global anomaly score with the anomaly score threshold. Based on multiple first comparison results, the second comparison result, and the local contribution vector, a corresponding anomaly event alarm is issued.

6. The intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer according to claim 2, characterized in that, The process of collecting multiple water samples from the target water area through the sampling and analysis unit, and preprocessing and analyzing the multiple water samples to obtain multiple first analysis results of the target water area includes: The sampling and analysis unit is used to collect water samples from the area in the target water body where the abnormal event alarm was generated; The water sample was filtered, digested online, and enriched to obtain the treated water sample; The first analytical result was obtained by performing colorimetric, electrochemical, and optical detection on the treated water sample.

7. The intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer according to claim 2, characterized in that, When the abnormal event alarm is triggered, the plurality of sensing and sensor units and the plurality of sampling and analysis units exchange a plurality of water quality data, a plurality of water body data, and a plurality of the first analysis results, including: When the abnormal event alarm is triggered, the autonomous navigation and positioning unit at the trigger point broadcasts a request signal to the target water area; Upon receiving the request signal, the plurality of sensing and sensor units and the plurality of sampling and analysis units exchange multiple water quality data, multiple water body data, and multiple first analysis results in a unified data format.

8. The intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer according to claim 2, characterized in that, The step of using the computing and communication unit to compare and analyze multiple water quality data, multiple water body data, and multiple first analysis results to obtain the fault node, and performing the autonomous calibration on the fault node, includes: The computing and communication unit uses the Byzantine fault-tolerant consensus algorithm to calculate multiple water quality data, multiple water body data, and multiple first analysis results for water quality consensus, water body consensus, and first analysis consensus. The water quality consensus threshold, water body consensus threshold, and first analysis consensus threshold are preset, and the fault node is obtained by comparing the water quality consensus result, the water body consensus result, and the first analysis consensus result with the water quality consensus threshold, the water body consensus threshold, and the first analysis consensus threshold, respectively. The faulty node is subjected to both soft and hard calibration.

9. The intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer according to claim 2, characterized in that, The step of using the computing and communication unit to perform spatiotemporal alignment and data fusion on multiple water quality data, multiple water body data, and multiple first analysis results to obtain the first comprehensive water quality field includes: The computing and communication unit establishes a unified time reference and performs cubic spline interpolation on the multiple water quality data, multiple water body data, and multiple asynchronous data in the first analysis results; Convert the multiple water quality data, the multiple water body data, and the location data in the multiple first analysis results into the UTM coordinate system; The first comprehensive water quality field is obtained by fusing multiple water quality data, multiple water body data, and multiple first analysis results using confidence weighting.

10. The intelligent calibration and remote transmission method for a portable multi-parameter water quality analyzer according to claim 2, characterized in that, The process of obtaining the second comprehensive water quality field by filling the spatial monitoring gaps of the first comprehensive water quality field using the interpolation method includes: Multiple constraints are defined, and the second integrated water quality field is generated based on these constraints using the Kriging space interpolation method.

Images