A smart water quality detection system and method

Abstract

This invention discloses an intelligent water quality detection system and method, belonging to the field of water quality detection technology. It solves the problems of mixed peaks in industrial wastewater, where 90% of the pollutants are unknown and lack toxicological understanding, the cocktail effect is difficult to predict, massive amounts of data are uninterpretable, and there are no countermeasures after detection, leading to delayed remediation. The system includes a field sensing module, an edge computing module, and a central decision-making module. The field sensing module connects to the field communication port of the edge computing module through its field communication interface, and the edge computing module connects to the network access port of the central decision-making module through its uplink communication interface. This invention collects, fuses, and analyzes complex pollution information in water bodies in real time, dynamically assesses the risk of mixed toxicity, and intelligently traces the source, ultimately generating accurate optimized control decisions and visualized simulation schemes.

Description

Technical Field

[0001] This invention relates to the field of water quality testing technology, and in particular to an intelligent water quality testing system and method. Background Technology

[0002] Modern water quality testing has moved from targeted monitoring of known pollutants to a new stage of non-targeted screening using tools such as high-resolution mass spectrometry, aiming to comprehensively reveal thousands of unknown chemical substances in water bodies, especially in the combined runoff areas of industrial and domestic sewage.

[0003] In the downstream section where the industrial park and the urban sewage treatment plant meet, when high-resolution mass spectrometry is used for non-targeted water quality screening, the water body is actually a complex mixture of thousands of industrial chemicals, pharmaceuticals and their unknown transformation products, which causes the detection instrument to generate a data tsunami containing tens of thousands of characteristic peaks.

[0004] The vast majority of detected chemicals are unknown because they lack identification information in existing databases. Secondly, even if resources are spent to identify some of their structures, key data such as their toxicity and environmental behavior are completely blank. More importantly, the possible synergistic or antagonistic cocktail effects between these known and unknown pollutants make it impossible to predict mixed toxicity from data on a single substance.

[0005] Ultimately, the massive amounts of information generated by current water quality testing technologies cannot be effectively interpreted and transformed into clear risk assessment conclusions. They cannot accurately pinpoint pollution sources or support effective governance decisions, resulting in a management dilemma where technological investment has findings but no solutions, thus delaying the opportunity for risk intervention.

[0006] Therefore, an intelligent water quality detection system and method are proposed to solve or alleviate the above problems. Summary of the Invention

[0007] The purpose of this invention is to address the shortcomings of existing technologies by proposing an intelligent water quality detection system and method.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: An intelligent water quality detection system includes a field sensing module, an edge computing module, and a central decision-making module. The field sensing module is connected to the field communication port of the edge computing module through its field communication interface, and the edge computing module is connected to the network access port of the central decision-making module through its uplink communication interface.

[0009] Preferably, the field sensing module includes a nano-sensor node circuit, which includes a microcontroller, a sensor probe array, a signal conditioning circuit, an analog-to-digital converter, and a communication interface chip. The sensor probe array includes a silicon nanowire field-effect transistor, an upconversion nanoparticle optical sensor, a microbial fuel cell sensor, and a multi-parameter integrated probe for temperature, salinity, turbidity, and dissolved oxygen. The signal conditioning circuit includes an instrumentation amplifier, a first transimpedance amplifier, a low-power operational amplifier, and a multiplexer. The source and drain pins of the silicon nanowire field-effect transistor are respectively connected to the positive and negative input pins of the corresponding instrumentation amplifier. The photoelectric signal output pin of the upconversion nanoparticle optical sensor is connected to the input pin of the first transimpedance amplifier, and the electrode output pin of the microbial fuel cell sensor is connected to the low-power operational amplifier. The input pins of the operational amplifier and the analog output pins of the temperature, salinity, turbidity, and dissolved oxygen multi-parameter integrated probe are connected to the input channel pins of the multiplexer. The output pins of the instrumentation amplifier, the first transimpedance amplifier circuit, the micropower operational amplifier circuit, and the multiplexer are respectively connected to the analog input channel pins of the analog-to-digital converter. The serial clock pin and serial data pin of the analog-to-digital converter are respectively connected to the integrated circuit bus clock pin and integrated circuit bus data pin of the microcontroller. The transmit pin of the microcontroller's universal asynchronous transceiver is connected to the driver input pin of the communication interface chip. The receive pin of the microcontroller's universal asynchronous transceiver is connected to the receiver output pin of the communication interface chip. The differential data output positive pin and differential data output negative pin of the communication interface chip are connected to the standard serial bus interface.

[0010] Preferably, the nanosensor node circuit further includes a power management circuit, which includes a lithium battery, a charging management chip, a buck switching regulator chip, and a low-dropout linear regulator chip. The positive power input pin of the charging management chip can be used for power connection, the battery charging pin of the charging management chip is connected to the positive terminal of the lithium battery, the positive terminal of the lithium battery is connected to the voltage input pin of the buck switching regulator chip, the voltage output pin of the buck switching regulator chip is connected to the voltage input pin of the low-dropout linear regulator chip, and the voltage output pin of the low-dropout linear regulator chip provides operating voltage for the microcontroller, analog-to-digital converter, and communication interface chip. The nanosensor node circuit is connected to the corresponding downlink serial communication port of the edge gateway circuit in the edge computing module through its standard serial bus interface.

[0011] Preferably, the field sensing module further includes a surface acoustic wave (SAW) microfluidic sorting array circuit. The SAW microfluidic sorting array circuit includes a main controller, a radio frequency (RF) signal generator, a power amplifier, an interdigital transducer array, an optical detection unit, and a communication control interface chip. The optical detection unit includes an ultraviolet-visible fiber optic spectrometer and a photomultiplier tube module. The serial peripheral interface master output / slave input pin, serial peripheral interface master input / slave output pin, and serial peripheral interface clock pin of the main controller are respectively connected to the serial data input pin, serial data output pin, and serial clock input pin of the RF signal generator. The RF signal output pin of the RF signal generator is connected to the power amplifier via a coaxial transmission line. The RF signal input pin and the RF signal output pin of the power amplifier are connected to the signal input pads of the corresponding interdigital transducers in the interdigital transducer array through an impedance matching network composed of inductors and capacitors. The transmit and receive pins of the universal asynchronous transceiver of the UV-Vis fiber optic spectrometer are connected to the driver input pin and receiver output pin of the communication control interface chip, respectively. The high-voltage power enable control pin of the photomultiplier tube module is connected to the digital output pin of the main controller, and its analog current output pin is connected to another built-in analog-to-digital converter input channel pin of the main controller through a second transimpedance amplifier. The differential data line of the communication control interface chip is connected to the downlink communication interface of the edge computing module through a first industrial Ethernet cable.

[0012] Preferably, the surface acoustic wave microfluidic sorting array circuit further includes a fluid control unit, which includes a motor drive chip, a microfluidic peristaltic pump, a solenoid valve, a flow sensor, and a pressure sensor. Multiple pulse width modulation signal output pins of the main controller are respectively connected to the enable control pins of multiple motor drive chips. The power output pins of each motor drive chip are respectively connected to the DC motor control pins of the corresponding microfluidic peristaltic pump. Multiple general-purpose input / output pins of the main controller drive the coil pins of multiple solenoid valves through a Darlington transistor array chip. The serial clock pin and serial data pin of the integrated circuit bus of the flow sensor are respectively connected to the spare integrated circuit bus clock pin and spare integrated circuit bus data pin of the main controller. The analog voltage output pin of the pressure sensor is connected to the analog input channel pin of the built-in analog-to-digital converter of the main controller.

[0013] Preferably, the on-site sensing module further includes an in-situ photochemical reaction chamber circuit. The in-situ photochemical reaction chamber circuit includes a programmable logic controller (PLC) module, a light source control unit, a sample reaction and temperature control unit, an online detection unit, and an industrial communication interface module. The light source control unit includes a tunable LED array, LED constant current driving chips, and a laser diode driving power supply. The sample reaction and temperature control unit includes a quartz reaction cell, a magnetic stirrer, a temperature sensor, and a pH electrode. The online detection unit includes a miniature mass spectrometer, an electron paramagnetic resonance spectrometer, and a time-correlated single-photon counter. Multiple analog output channel pins of the PLC module are respectively connected to the pulse width modulation (PWM) dimming pins of each LED constant current driving chip. The current output pin of each LED constant current driving chip is connected to the positive terminal of the tunable LED array. The laser diode driving power supply... The analog enable control pin of the power supply is connected to the digital output channel pin of the programmable logic controller (PLC) module. The motor speed control pin and power control pin of the magnetic stirrer are connected to the analog output channel pin of the PLC module. The resistance signal output pin of the temperature sensor and the analog voltage signal output pin of the pH electrode are respectively connected to the analog input channel pin of the PLC module. The signal output terminals of the micro mass spectrometer, electron paramagnetic resonance spectrometer, and time-correlated single-photon counter are respectively connected to the expansion port of the PLC module through a high-speed data acquisition card. The industrial fieldbus interface of the PLC module is converted to the standard industrial Ethernet protocol through an industrial communication interface module and connected to an industrial Ethernet switch. The in-situ photochemical reaction chamber circuit is connected to the downlink communication interface of the edge computing module through an industrial Ethernet switch.

[0014] Preferably, the edge computing module includes an edge gateway circuit, which comprises an embedded industrial control board core module, a field-programmable gate array (FPGA) accelerator card, a communication interface array, and a data preprocessing and security module. The data preprocessing and security module includes an encryption chip and a real-time clock chip. The embedded industrial control board core module is connected to the FPGA accelerator card. Multiple universal asynchronous transceiver (API) transmit and receive pins of the embedded industrial control board core module are respectively connected to the driver input pins and receiver output pins of the communication interface array. The differential output pins of the communication interface array are connected to the field sensor via its standard serial bus interface and controller area network (CLAN) bus interface. The nanosensor node circuit in the knowledge module is connected, and the Ethernet controller medium-independent interface of the embedded industrial control board core module is connected to the surface acoustic wave microfluidic sorting array circuit and the in-situ photochemical reaction chamber circuit. The secure digital input / output interface of the embedded industrial control board core module is connected to the server cluster circuit in the central decision-making layer module for data communication through the fifth-generation mobile communication technology module. The serial clock pin and serial data pin of the integrated circuit bus of the encryption chip are respectively connected to the integrated circuit bus clock pin and data pin of the embedded industrial control board core module. The interrupt output pin of the real-time clock chip is connected to the external interrupt input pin of the embedded industrial control board core module.

[0015] Preferably, the central decision-making module includes a server cluster circuit, which includes multiple parallel computing servers, a centralized data storage server, and a network switching device. The parallel computing servers include a central processing unit (CPU) chip, a graphics processing unit (GPU) accelerator card, a double data rate (DFR) memory module, and a solid-state drive (SSD). The centralized data storage server includes an independent redundancy array (RDA) controller, a mechanical hard disk array (HDD) array, and a high-speed SSD cache. The network switching device includes a core layer switch and multiple access layer switches. The CPU chips in each parallel computing server are interconnected. Each GPU accelerator card is connected to a parallel computing server. The DFR memory module is connected to the CPU chip. The SSD is connected to the CPU chip via a high-speed non-volatile memory interface. The DRA controller connects the HDD array and the high-speed SSD cache to the CPU chip. The CPU chip is connected to the core layer switch. The access layer switches are connected to the stacked ports of the core layer switch to form an internal network. The uplink communication interface of the edge gateway circuit is connected to the corresponding port of the access layer switch. The graphics workstation in the digital twin and visualization circuit is connected to the core layer switch.

[0016] Preferably, the central decision-making module further includes a digital twin and visualization circuit, which includes a graphics workstation, a splicing processor, and multiple display units. The display port output interfaces of the multiple graphics cards of the graphics workstation are connected to the multiple display port input interfaces of the splicing processor through display port cables. The multiple high-definition multimedia interface output ports of the splicing processor are respectively connected to the high-definition multimedia interface input ports of the multiple display units through high-definition multimedia interface cables. The network interface of the graphics workstation is connected to the network switching equipment in the central decision-making module through fiber optic patch cords.

[0017] This invention provides an intelligent water quality detection method, executed by the intelligent water quality detection system described above, comprising the following steps: The S1 field perception module collects multi-source heterogeneous data, preprocesses and aligns the data spatiotemporally to generate a standardized input dataset; The S2 edge computing module analyzes mass spectrometry data and generates an information list containing known and unknown risk substances through feature extraction and structure inference. The S3 center decision module models mixed toxicity effects based on biotoxicity test data and a chemical substance inventory, quantifying the toxicity contribution of various pollutants. The S4 central decision-making module uses a chemical fingerprint database and a Bayesian inference model to conduct source tracing analysis of pollutants, calculate and dynamically track the contribution of each potential pollution source; The S5 central decision-making module integrates chemical, ecological, and source tracing risks, calculates dynamic environmental risk indices, and sets tiered early warning thresholds. The S6 central decision-making module generates control and management decision-making schemes through multi-objective optimization and scenario simulation, and drives digital twins and visualization circuits to perform simulations and displays.

[0018] The present invention has the following beneficial effects: This invention collects, integrates, and analyzes information on complex pollutants in water bodies in real time, dynamically assesses their mixed toxicity risks, and intelligently traces their sources, ultimately generating precise optimized control decisions and visualized simulation schemes. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the structure of the present invention; Figure 2This is a flowchart of the present invention.

[0021] In the figure: 1. On-site sensing module; 101. Nano-sensor node circuit; 102. Surface acoustic wave microfluidic sorting array circuit; 103. In-situ photochemical reaction chamber circuit; 2. Edge computing module; 3. Central decision-making module; 301. Server cluster circuit; 302. Digital twin and visualization circuit. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0023] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0024] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0025] In the description of this invention, it should be understood that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use, or the orientation or positional relationship commonly understood by those skilled in the art. They are only used to facilitate the description of this invention and to simplify the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0026] Furthermore, the terms "first," "second," and "third" are used only to distinguish descriptions and should not be interpreted as indicating or implying relative importance.

[0027] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0028] A smart water quality detection system, such as Figure 1 As shown, it includes a field sensing module 1, an edge computing module 2, and a central decision-making module 3. The field sensing module 1 is connected to the field communication port of the edge computing module 2 through its field communication interface, and the edge computing module 2 is connected to the network access port of the central decision-making module 3 through its uplink communication interface.

[0029] The field sensing module 1 includes a nanosensor node circuit 101, which includes a microcontroller STM32F407VGT6, a sensor probe array, a signal conditioning circuit, an analog-to-digital converter ADS1115, and a communication interface chip MAX485. The sensor probe array includes a silicon nanowire field-effect transistor, an upconversion nanoparticle optical sensor, a microbial fuel cell sensor, and a multi-parameter integrated probe for temperature, salinity, turbidity, and dissolved oxygen. The signal conditioning circuit includes an instrumentation amplifier AD8421, a first transimpedance amplifier OPA657, a micropower operational amplifier AD8630, and a multiplexer CD4051B. The source and drain pins of the silicon nanowire field-effect transistor are connected to the positive and negative input pins of the corresponding instrumentation amplifiers, respectively. The photoelectric signal output pin of the upconversion nanoparticle optical sensor is connected to the input of the first transimpedance amplifier. The electrode output pins of the microbial fuel cell sensor are connected to the input pins of the micropower operational amplifier. The analog output pins of the temperature, salinity, turbidity, and dissolved oxygen multi-parameter integrated probe are connected to the input channel pins of the multiplexer. The output pins of the instrumentation amplifier, the first transimpedance amplifier circuit, the micropower operational amplifier circuit, and the multiplexer are respectively connected to the analog input channel pins of the analog-to-digital converter. The serial clock pin and serial data pin of the analog-to-digital converter are respectively connected to the integrated circuit bus clock pin and integrated circuit bus data pin of the microcontroller. The transmit pin of the microcontroller's universal asynchronous transceiver is connected to the driver input pin of the communication interface chip. The receive pin of the microcontroller's universal asynchronous transceiver is connected to the receiver output pin of the communication interface chip. The differential data output positive pin and differential data output negative pin of the communication interface chip are connected to the standard serial bus interface.

[0030] The nanosensor node circuit 101 also includes a power management circuit, which includes a lithium battery, a charging management chip BQ24610, a buck switching regulator chip LTM4644, and a low dropout linear regulator chip ADP7118. The positive power input pin of the charging management chip can be used for power connection. The battery charging pin of the charging management chip is connected to the positive terminal of the lithium battery. The positive terminal of the lithium battery is connected to the voltage input pin of the buck switching regulator chip. The voltage output pin of the buck switching regulator chip is connected to the voltage input pin of the low dropout linear regulator chip. The voltage output pin of the low dropout linear regulator chip provides operating voltage for the microcontroller, analog-to-digital converter, and communication interface chip. The nanosensor node circuit 101 is connected to the corresponding downlink serial communication port of the edge gateway circuit in the edge computing module 2 through its standard serial bus interface.

[0031] The field sensing module 1 also includes a surface acoustic wave (SAW) microfluidic sorting array circuit 102. The SAW microfluidic sorting array circuit 102 includes a main controller STM32H743IIT6, an RF signal generator ADF4351, a power amplifier, an interdigital transducer array, an optical detection unit, and a communication control interface chip LAN8720A. The optical detection unit includes an ultraviolet-visible fiber optic spectrometer and a photomultiplier tube module H10722-01. The main controller's serial peripheral interface master output / slave input pin, serial peripheral interface master input / slave output pin, and serial peripheral interface clock pin are respectively connected to the RF signal generator's serial data input pin, serial data output pin, and serial clock input pin. The RF signal generator outputs an RF signal. The pins are connected to the RF signal input pins of the power amplifier via coaxial transmission lines. The RF signal output pins of the power amplifier are connected to the signal input pads of the corresponding interdigital transducers in the interdigital transducer array via an impedance matching network composed of inductors and capacitors. The transmit and receive pins of the universal asynchronous transceiver of the UV-Vis fiber optic spectrometer are connected to the driver input pin and receiver output pin of the communication control interface chip, respectively. The high-voltage power enable control pin of the photomultiplier tube module is connected to the digital output pin of the main controller. Its analog current output pin is connected to another built-in analog-to-digital converter input channel pin of the main controller via a second transimpedance amplifier. The differential data lines of the communication control interface chip are connected to the downlink communication interface of the edge computing module 2 via a first industrial Ethernet cable.

[0032] The surface acoustic wave microfluidic sorting array circuit 102 also includes a fluid control unit, which includes a motor driver chip DRV8833, a microfluidic peristaltic pump, a solenoid valve, a flow sensor SLF3S-1300F, and a pressure sensor ASDXRRX100PD2A5. Multiple pulse width modulation signal output pins of the main controller are respectively connected to the enable control pins of multiple motor driver chips. The power output pins of each motor driver chip are respectively connected to the DC motor control pins of the corresponding microfluidic peristaltic pump. Multiple general-purpose input / output pins of the main controller drive the coil pins of multiple solenoid valves through a Darlington transistor array chip. The serial clock pin and serial data pin of the integrated circuit bus of the flow sensor are respectively connected to the spare integrated circuit bus clock pin and spare integrated circuit bus data pin of the main controller. The analog voltage output pin of the pressure sensor is connected to the analog input channel pin of the built-in analog-to-digital converter of the main controller.

[0033] The on-site sensing module 1 also includes an in-situ photochemical reaction chamber circuit 103. This circuit includes a programmable logic controller (PLC) module (SIMATIC S7-1200), a light source control unit, a sample reaction and temperature control unit, an online detection unit, and an industrial communication interface module. The light source control unit includes a tunable LED array, an LT3763 LED constant current driver chip, and an LDC205C laser diode driver. The sample reaction and temperature control unit includes a quartz reaction cell, a magnetic stirrer, a PT100 temperature sensor, and a pH electrode. The online detection unit includes a miniature mass spectrometer, an electron paramagnetic resonance spectrometer, and a time-correlated single-photon counter. Multiple analog output channel pins of the PLC module are connected to the pulse width modulation (PWM) dimming pins of each LED constant current driver chip. The current output pin of each LED constant current driver chip is connected to the positive terminal of the tunable LED array. The analog enable control pin of the laser diode driver is connected to... The digital output channel pins of the programmable logic controller (PLC) module, the motor speed control pin and power control pin of the magnetic stirrer are connected to the analog output channel pins of the PLC module, the resistance signal output pin of the temperature sensor and the analog voltage signal output pin of the pH electrode are connected to the analog input channel pins of the PLC module, respectively, the signal output terminals of the micro mass spectrometer, electron paramagnetic resonance spectrometer and time-correlated single-photon counter are connected to the expansion port of the PLC module through a high-speed data acquisition card, the industrial fieldbus interface of the PLC module is converted to the standard industrial Ethernet protocol through the industrial communication interface module and connected to an industrial Ethernet switch, and the in-situ photochemical reaction chamber circuit 103 is connected to the downlink communication interface of the edge computing module 2 through the industrial Ethernet switch.

[0034] Edge computing module 2 includes an edge gateway circuit, which comprises an embedded industrial control board core module (NVIDIA Jetson AGX Orin), a field-programmable gate array (FPGA) accelerator card (Xilinx Alveo U50), a communication interface array, and a data preprocessing and security module. The data preprocessing and security module includes an encryption chip (ATECC608A) and a real-time clock chip (DS3231). The embedded industrial control board core module is connected to the FPGA accelerator card. Multiple universal asynchronous transceiver (API) transmit and receive pins of the embedded industrial control board core module are connected to the driver input pins and receiver output pins of the communication interface array, respectively. The differential output pins of the communication interface array are connected to the nanosensor node circuit 101 in the field sensing module 1 via its standard serial bus interface and controller area network (CLAN) bus interface, respectively. The Ethernet controller media-independent interface of the embedded industrial control board core module is connected to the surface acoustic wave microfluidic sorting array circuit 102 and the in-situ photochemical reaction chamber circuit 103. The secure digital input / output interface of the embedded industrial control board core module is connected to the fifth-generation mobile communication technology module (Quectel). The RM500Q-GL is connected to the server cluster circuit 301 in the central decision module 3 for data communication. The serial clock pin and serial data pin of the encryption chip's integrated circuit bus are connected to the integrated circuit bus clock pin and data pin of the core module of the embedded industrial control board, respectively. The interrupt output pin of the real-time clock chip is connected to the external interrupt input pin of the core module of the embedded industrial control board.

[0035] The central decision module 3 includes a server cluster circuit 301, which comprises multiple parallel computing servers, a centralized data storage server, and network switching equipment. The parallel computing servers include an Intel Xeon Platinum 8468 CPU chip, an NVIDIA A100 80GB PCIe graphics card, a Double Data Rate (DDR) memory module (M393AAG40M3-CWE), and a PM9A3 solid-state drive (SSD). The centralized data storage server includes a Broadcom MegaRAID 9560-16i independent disk redundant array controller, an ST20000NM007D hard disk array, and an Intel Optane SSD P5800X high-speed SSD cache. The network switching equipment includes a Cisco Nexus 9336C-FX2 core layer switch and multiple Cisco Catalyst access layer switches. The 9200L-48T-4X system interconnects the central processing unit (CPU) chips within each parallel computing server. Each graphics processing unit (GPU) accelerator card is connected to the parallel computing server. Double data rate (DFR) memory modules are connected to the CPU chips. Solid-state drives (SSDs) are connected to the CPU chips via a high-speed non-volatile memory interface. An independent redundant disk array (DBA) controller connects the mechanical hard disk array and high-speed SSD cache to the CPU chips. The CPU chips are connected to the core layer switch. Access layer switches are connected to the stacked ports of the core layer switch to form an internal network. The uplink communication interface of the edge gateway circuit is connected to the corresponding port of the access layer switch. The graphics workstation in the digital twin and visualization circuit 302 is connected to the core layer switch.

[0036] The central decision-making module 3 also includes a digital twin and visualization circuit 302, which includes an NVIDIA RTX A6000 graphics workstation, a Barco Folsom 9600 splicing processor, and multiple PlanarCarbonLight CL55 display units. The display port output interfaces of the multiple graphics cards of the graphics workstation are connected to the multiple display port input interfaces of the splicing processor through display port cables. The multiple high-definition multimedia interface output ports of the splicing processor are connected to the high-definition multimedia interface input ports of the multiple display units through high-definition multimedia interface cables. The network interface of the graphics workstation is connected to the network switching equipment in the central decision-making module 3 through fiber optic patch cords.

[0037] This invention provides an intelligent water quality detection method, executed by the intelligent water quality detection system described above, such as... Figure 2 As shown, it includes the following steps: S1 field perception module 1 collects multi-source heterogeneous data, preprocesses and aligns the data spatiotemporally to generate a standardized input dataset; S101, the nano-sensor network node circuit collects real-time water quality parameters, the surface acoustic wave microfluidic sorting array circuit 102 processes the sample feature data, and the in-situ photochemical reaction chamber circuit 103 transmits the reaction monitoring data to the edge computing module 2 respectively. S102, Edge computing module 2 timestamps the received multi-source data and compensates for the time delay caused by the transmission distance according to the preset hydrodynamic model; S103. For non-uniformly sampled data sequences, an adaptive sliding window filtering algorithm based on polynomial fitting is used for interpolation processing. S104. Construct a network graph of connections between sensors, use multivariate statistical distance method to identify abnormal data points, and combine graph neural network to aggregate neighborhood information to collaboratively correct outliers, forming a high-quality spatiotemporal synchronization dataset. The S2 edge computing module 2 analyzes the mass spectrometry data and generates an information list containing known and unknown risk substances through feature extraction and structure inference. S201, edge computing module 2 or central decision module 3 receives raw mass spectrometry data stream from field sensing module 1; S202. In the chromatographic dimension, continuous wavelet transform function is applied to identify chromatographic peaks, and in the mass spectrometry dimension, peaks are merged and combined by calculating isotope distribution similarity. S203. Using a deconvolution algorithm based on multivariate curve resolution, overlapping chromatographic mass spectrometry peaks are decomposed into pure single compound peaks, and the precise mass number, retention time and signal intensity characteristics of each compound are extracted. S204. Using machine learning models, based on mass spectrometry fragment information, nitrogen rules and unsaturation calculations, the structure category or molecular fragment of unknown mass spectrometry peaks is inferred. S205. Construct a multi-criteria evaluation matrix that includes detection frequency, signal strength, predicted environmental behavior parameters and correlation with toxicity. Use the approximation ideal solution ranking method to conduct risk assessment and priority ranking of unknown features and generate a list of high-risk unknowns. S3 Center Decision Module 3 models mixed toxicity effects based on biotoxicity test data and chemical substance inventory, and quantifies the toxicity contribution of various pollutants. S301, the central decision module 3 receives multi-endpoint effect data from the biotoxicity testing unit, performs four-parameter logistic curve fitting on each endpoint data, and calculates the toxicity unit of a single pollutant. S302. Based on the concentration summation model, calculate the predicted baseline toxicity effect of the mixture using the toxicity units of each pollutant; S303. Compare the ratio of the actual observed toxic effects to the predicted baseline toxic effects. When the ratio exceeds a set threshold, determine that there is a synergistic or antagonistic effect and initiate an interaction effect analysis. S304. Apply the Shapley value calculation method based on cooperative game theory, treat all detected known and unknown pollutant features as members of the alliance, and fairly allocate the observed comprehensive toxicity effect value to each feature, thereby quantifying the independent and interactive contributions of each chemical feature to the total toxicity. S4 Central Decision Module 3, based on a chemical fingerprint database and a Bayesian inference model, performs source tracing analysis of pollutants and calculates and dynamically tracks the contribution of each potential pollution source. S401, Central Decision Module 3 calls the pre-built pollution source chemical fingerprint database, which contains identification information such as the concentration ratio of characteristic pollutants and stable isotope ratios of each potential emission source; S402. Establish a linear mixture model to describe the chemical fingerprint of the downstream monitoring section as a linear superposition of the fingerprints of each upstream pollution source multiplied by their contribution ratio. S403. Under the Bayesian inference framework, prior information on the non-negative contribution of pollution sources is introduced, and the Markov chain Monte Carlo sampling algorithm is used to solve the posterior probability distribution and confidence interval of the contribution ratio of each pollution source. S404. Construct a state-space model with the contribution of pollution sources as the state variable, apply the Kalman filter algorithm to recursively estimate and update the contribution in real time, and use the cumulative sum control chart method to monitor sudden changes in the contribution, so as to realize the dynamic tracking of pollution sources. S5 Central Decision Module 3 integrates chemical, ecological, and source tracing risks, calculates dynamic environmental risk indices, and sets tiered early warning thresholds; S501, Central Decision Module 3 calculates the chemical risk sub-index, which is the root mean square of the sum of squares of the ratios of the concentrations of each pollutant to their corresponding standard limits, and introduces a risk amplification coefficient that reflects the proportion of unknown substances. S502. Calculate the ecological risk sub-index, which is the maximum value of the ratio of observed toxicity units to a preset safety threshold among multiple biotoxicity test endpoints, and introduces a correction factor that reflects the degree of synergistic effect. S503. Calculate the source risk sub-index, which is the sum of the weighted product of the contribution of each pollution source, its historical environmental performance score, and the current controllability factor. S504. The information entropy method is used to analyze the volatility of historical data of each sub-index to determine the objective weights, and then combined with the preset subjective weights to obtain the final comprehensive weight coefficient. S505. Using the weighted geometric mean method, the dynamic comprehensive environmental risk index is calculated by combining the above three risk sub-indices and their comprehensive weights. S506. Based on the peak value method, perform extreme value statistical analysis on the historical risk index sequence, calculate the risk warning threshold corresponding to different return periods by fitting the generalized Pareto distribution function, and classify the risk level accordingly. The S6 central decision module 3 generates control decision schemes through multi-objective optimization and scenario simulation, and drives the digital twin and visualization circuit 302 to perform deduction and display. S601, Central Decision Module 3 constructs a constrained multi-objective optimization model with the emission reduction ratio of each pollution source as the decision variable and the objectives of minimizing the predicted environmental risk index and minimizing the total economic cost; S602. The non-dominated sorting genetic algorithm based on reference points is used to solve the optimization model. By simulating the selection, crossover and mutation operations in the biological evolution process, a set of Pareto optimal solutions is generated iteratively, which are multiple control scenario schemes that do not dominate each other. S603. For each scheme in the Pareto optimal solution set, evaluate its risk reduction efficiency, cost-benefit ratio and implementation feasibility, and apply the multi-criteria compromise solution ranking method to comprehensively calculate the group utility value and individual regret value of each scheme, thereby obtaining the ranking of recommended implementation schemes. S604. Generate a hash digest of the final decision-making scheme and the complete analytical evidence chain data, store it in the blockchain node for evidence storage, and generate an automatically executable smart contract based on the decision-making logic. S605 drives the digital twin and visualization circuit 302, loads the hydrodynamic and water quality coupling model in the server cluster circuit 301, performs scenario simulation and visualization deduction of the recommended implementation scheme, and dynamically presents the results through multiple display units.

[0038] The nanosensor node circuit 101 deployed in the downstream provincial control section and upstream key nodes of this system operates continuously, with its microcontroller acting as the local brain, coordinating the working rhythm of the entire sensor array.

[0039] The silicon nanowire field-effect transistors in the sensor probe array, with their surface-modified specific probes, can sense in real time the minute changes in channel conductivity caused by specific heavy metal ions or organic pollutants in the water. The upconversion nanoparticle optical sensor emits visible light signals under near-infrared light excitation, effectively avoiding background fluorescence interference and achieving highly sensitive detection of trace organic pollutants. The microbial fuel cell sensor directly reflects the bioavailability of pollutants through the current changes generated by electroactive microorganisms metabolizing pollutants. The temperature, salinity, turbidity, and dissolved oxygen multi-parameter integrated probe simultaneously acquires the basic physicochemical parameters of the water.

[0040] The raw, weak analog signals generated by these sensors are immediately fed into the signal conditioning circuit. The instrumentation amplifier amplifies the weak differential signal of the silicon nanowire device with high fidelity. The transimpedance amplifier converts the photocurrent of the optical sensor into a voltage signal. The micropower operational amplifier processes the microampere-level current of the microbial fuel cell. The multiplexer switches the multiple output channels of the multi-parameter probe in an orderly manner.

[0041] After conditioning, the various analog signals are sent to the analog-to-digital converter, which converts the information of the analog world into binary code of the digital world with high precision. The microcontroller reads these digitized water quality parameters through the bus, performs preliminary processing and packaging, and finally converts the data into differential signals through its serial port control communication interface chip. These signals are then sent to the edge computing module 2 via a long-distance communication link with strong anti-interference capabilities.

[0042] Meanwhile, once an acute increase in toxicity occurs underwater, the system immediately triggers the emergency monitoring mode. The surface acoustic wave microfluidic sorting array circuit 102 is then activated, and its main controller issues a command. The motor drive chip in the fluid control unit drives the microfluidic peristaltic pump to accurately introduce the abnormal water sample into the chip channel. The flow sensor and pressure sensor monitor the flow rate and pressure in real time to ensure process stability.

[0043] Next, the main controller controls the radio frequency signal generator to generate a radio frequency signal of a specific frequency via the bus. After being amplified by the power amplifier, the signal drives the interdigital transducer to generate a strong surface acoustic wave standing wave field in the microchannel. Under the action of acoustic radiation force, particles and molecular aggregates of different sizes, densities and compressibility coefficients in the water are enriched at pressure nodes or antinodes, thereby achieving physical pre-sorting and concentration of pollutants by hundreds of times, greatly enriching trace pollutants and simplifying the matrix for subsequent analysis.

[0044] The ultraviolet-visible fiber optic spectrometer and photomultiplier tube module in the optical detection unit monitor the changes in the spectral characteristics of the samples in real time during the sorting process, and the data is converted into Ethernet frames through the communication control interface chip.

[0045] The concentrated high-risk sample is automatically transported to the in-situ photochemical reaction chamber circuit 103. The programmable logic controller module acts as the general commander, controlling the light source control unit. Its analog output module precisely adjusts the duty cycle of the LED constant current drive chip, thereby controlling the tunable LED array to emit light pulses of different wavelengths from ultraviolet to visible light to simulate the photolysis process of natural water bodies. The laser diode drive power supply emits lasers of specific wavelengths to trigger photochemical reactions when needed. On the other hand, it controls the sample reaction and temperature control unit. The magnetic stirrer ensures that the sample is uniformly mixed in the quartz reaction cell, and the temperature sensor and pH electrode monitor the reaction conditions in real time.

[0046] The atmospheric pressure photoionization source of the miniature mass spectrometer, the resonant cavity of the electron paramagnetic resonance spectrometer, and the time-correlated single-photon counter of the online detection unit simultaneously capture the degradation products, free radical intermediates, and transient spectral signals of pollutants during the reaction process. These high-speed dynamic data are acquired by the programmable logic controller module through the high-speed data acquisition card and converted into industrial Ethernet protocol through the industrial communication interface module. Thus, the multi-source, heterogeneous, and massive raw data streams generated by all circuits in the field sensing layer are converged to the edge gateway circuit of the edge computing module 2 through a reliable link.

[0047] With its powerful computing capabilities, the core module of the embedded industrial control board of the edge gateway circuit, in collaboration with the field-programmable gate array (FPGA) accelerator card, begins to execute the data preprocessing and spatiotemporal alignment steps in the method.

[0048] The embedded industrial control board core module receives sensor data and receives data from the sorting array and reaction chamber via the Ethernet physical layer chip. It adds high-precision timestamps to all data, calls the preset hydrodynamic model, calculates the theoretical delay time of the pollution plume from different upstream points to the downstream section based on real-time hydrological data, and performs time shift compensation on the data to achieve alignment of all data on a unified time axis.

[0049] Next, to address the issue of different sampling frequencies of various sensors, an adaptive interpolation algorithm based on filtering is adopted, and with hardware acceleration from a field-programmable gate array (FPGA) accelerator card, a continuous data sequence with consistent time resolution is generated.

[0050] Then, the system constructs a sensor correlation graph and uses multivariate Mahalanobis distance to perform rapid calculations on the graphics processor of the core module of the embedded industrial control board to identify outlier data points caused by equipment failure or abnormal interference. Furthermore, it uses a graph neural network model to refer to the readings of adjacent and functionally similar sensors to intelligently correct and fill out the outliers.

[0051] During data preprocessing, the encryption chip generates digital signatures for key data to ensure its authenticity and integrity. After cleaning, alignment, and normalization, the high-quality spatiotemporal synchronized dataset is efficiently transmitted to the central decision module 3 through the communication module integrated in the core module of the embedded industrial control board or a high-speed fiber optic network.

[0052] In the central decision module 3, multiple parallel computing servers begin to operate. The high-performance computing cluster composed of the central processing unit chip and the graphics processing unit accelerator card inside undertakes the task of in-depth analysis of massive amounts of data.

[0053] High-speed network switching equipment ensures low-latency data exchange between servers and storage systems.

[0054] The mass spectrometry data analysis and risk ranking steps were then initiated.

[0055] The server cluster receives pre-processed raw mass spectrometry data streams from the edge gateway.

[0056] The algorithm first performs continuous wavelet transform on the data in the chromatographic dimension, and uses Mexican cap wavelets to identify the start, apex and end of the chromatographic peaks.

[0057] In terms of mass-to-charge ratio, the algorithm calculates the isotopic distribution pattern of each mass spectrum peak and matches it with the theoretical distribution. It then merges different adduct peaks and isotopic peaks belonging to the same compound to form compound characteristics.

[0058] For severely overlapping chromatographic peaks, a multivariate curve resolution algorithm is used for mathematical deconvolution. With the acceleration of a graphics processing unit (GPU), the pure concentration profile and mass spectrum are iteratively solved.

[0059] For mass spectral peaks that still cannot be matched with commercial mass spectral libraries after deconvolution, the algorithm enters the structure inference stage: based on the mass defect and neutral loss rules of mass spectral fragment ions, combined with nitrogen rules and unsaturation calculations, it uses a graph neural network model trained on a graphics processor accelerator card to predict the chemical category or core molecular fragment to which the unknown substance may belong.

[0060] The next crucial step is risk ranking: the algorithm constructs an evaluation vector for each unknown feature peak, encompassing five dimensions: detection frequency, signal strength, predicted environmental persistence, predicted bioaccumulation, and temporal correlation with biotoxic effects. It then uses a ranking method that approximates the ideal solution to calculate the comprehensive risk score for each unknown. The revolutionary effect of this step is that it transforms and prioritizes thousands of unnamed mass spectrometry peaks from the vast ocean of data into a limited list of high-risk unknowns, allowing management attention to be focused from a boundless ocean of chemical unknowns to a limited number of risky target islands.

[0061] Almost simultaneously, the mixed toxicity effect modeling and contribution decomposition steps were carried out, with the central decision module 3 calling up multi-endpoint effect data from the biotoxicity testing unit.

[0062] The algorithm first performs four-parameter curve fitting on the dose and effect data at each endpoint to calculate the half-effect concentration and toxicity unit of each identified pollutant.

[0063] Next, based on the concentration summation model, the theoretical mixed toxicity baseline for all known pollutants was calculated.

[0064] When the observed combined toxicity significantly exceeds the theoretical baseline, the algorithm determines that a significant cocktail effect exists.

[0065] At this point, the system initiates a value-based contribution decomposition model.

[0066] The model treats the observed total toxicity effect as the total payoff of a cooperative game, with each identified pollutant and the sorted key high-risk unknown features considered as players in the game.

[0067] Through complex combinatorial mathematical calculations, the model is able to fairly quantify each participant's marginal contribution to the total toxicity.

[0068] The steps of pollution source contribution analysis and dynamic tracking are beginning to point the finger at the responsible parties.

[0069] The system calls the pollution source chemical fingerprint database pre-built in the centralized data storage server. This database contains characteristic mass spectrometry maps, specific pollutant proportions, and stable isotope fingerprints formed by historical and real-time monitoring data of emission outlets of upstream industrial parks and key enterprises.

[0070] The algorithm establishes a Bayesian normal mixture model, which describes the chemical fingerprint detected at the downstream section as a linear superposition of the fingerprints of each potential pollution source upstream according to their contribution ratio, plus the measurement error.

[0071] Within the Bayesian framework, physical constraints such as the non-negativity of contribution ratios are introduced as prior information. The Markov chain Monte Carlo algorithm is used to perform large-scale sampling on a server cluster to solve for the posterior probability distribution of the contribution of each pollution source and to give its confidence interval.

[0072] Meanwhile, in order to cope with the dynamic changes in emissions, the algorithm constructs a state-space model and applies a Kalman filter to recursively estimate and update the contribution of pollution sources in real time.

[0073] Subsequently, the dynamic environmental risk index calculation and early warning process was initiated. Instead of simply listing chemical concentrations, a multi-dimensional risk assessment system was constructed, including chemical risk sub-indices, ecological risk sub-indices, and source tracing risk sub-indices. The server cluster used the information entropy method to automatically analyze the volatility of historical data for each sub-indice, dynamically determine its objective weight, and combine it with the subjective weight determined by expert experience to obtain a comprehensive weight.

[0074] Finally, a dynamic comprehensive environmental risk index between 0 and 1 is calculated using the weighted geometric mean formula. Based on long-term historical risk index data, the system uses the peak value method in extreme value statistics theory to fit a generalized Pareto distribution, thereby scientifically calculating the risk warning thresholds corresponding to different return periods.

[0075] When the real-time calculated risk index exceeds the once-in-a-year threshold, the system automatically triggers a yellow alert; A red alert is triggered when the threshold of a once-in-a-decade event is exceeded.

[0076] Scientific data containing massive amounts of information about unknown substances and complex toxic interactions has been compressed and translated into clear and easy-to-understand risk signals that are directly linked to environmental emergency response levels.

[0077] Finally, with the powerful support of the digital twin and visualization circuit 302, the multi-objective optimization decision-making and scenario simulation steps propel the system towards the endpoint of intelligent decision-making. When a risk warning is triggered, the digital twin engine is activated. Multiple graphics cards in the graphics workstation provide powerful graphics and computing capabilities, driving a high-precision hydrodynamic and water quality coupling model. This model incorporates the river's three-dimensional topography, real-time hydrology, and the analyzed locations, emission characteristics, and contribution ratios of key pollution sources, constructing a real-time virtual mapping of the river in digital space.

[0078] Based on this, the algorithm constructs a multi-objective optimization problem, with the decision variables being the emission reduction ratio of each key pollution source. The first optimization objective is to minimize the predicted comprehensive environmental risk index of the downstream section, and the second objective is to minimize the socio-economic costs brought about by measures such as production restrictions and shutdowns.

[0079] Using a reference-point-based non-dominated sorting genetic algorithm, with the aid of a graphics processor, the system quickly simulates hundreds of different combinations of control scenarios and calculates the expected risk reduction effect and economic cost under each scenario, forming a series of Pareto optimal solutions that achieve the best trade-off between risk and cost.

[0080] Decision-makers can intuitively see the diffusion changes of pollution clouds in the virtual river, the decline curve of the risk index, and the corresponding economic impact bar charts under different control schemes on multiple display units driven by the splicing processor.

[0081] The system further applies a multi-criteria decision-making method to rank the solutions in the Pareto optimal solution set and recommend the 2-3 control solutions with the best overall performance. The specific instructions contained in the final selected solution will be automatically generated and stored through blockchain technology to form an immutable electronic evidence chain. It can even be encoded into a smart contract, which will automatically trigger the sending of execution instructions when the conditions are met. Thus, the entire system completes the closed loop from detecting anomalies to issuing precise control instructions.

[0082] In summary, through the coordinated sensing, filtering, and activation actions of various precision electrical components in each circuit of the field sensing layer, the active capture and feature enhancement of information on complex pollutant mixtures were achieved. Through the fusion and normalization processing of edge gateway circuits, real-time standardization of multi-source heterogeneous data was realized.

[0083] Ultimately, the central decision-making module 3, through the execution of algorithmic steps such as in-depth analysis, quantitative attribution, risk translation, and optimization deduction, successfully transformed the once daunting data tsunami and unknown dilemmas generated by high-resolution mass spectrometry technology into specific control schemes and enforcement bases with clear ecological risk indications, probabilistic pollution source responsibility identification, quantitative graded early warning, and cost-benefit optimization analysis.

[0084] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An intelligent water quality detection system, characterized in that, It includes a field sensing module (1), an edge computing module (2) and a central decision-making module (3). The field sensing module (1) is connected to the field communication port of the edge computing module (2) through its field communication interface, and the edge computing module (2) is connected to the network access port of the central decision-making module (3) through its uplink communication interface.

2. The intelligent water quality detection system according to claim 1, characterized in that, The field sensing module (1) includes a nanosensor node circuit (101), which includes a microcontroller, a sensor probe array, a signal conditioning circuit, an analog-to-digital converter, and a communication interface chip. The sensor probe array includes a silicon nanowire field-effect transistor, an upconversion nanoparticle optical sensor, a microbial fuel cell sensor, and a multi-parameter integrated probe for temperature, salinity, turbidity, and dissolved oxygen. The signal conditioning circuit includes an instrumentation amplifier, a first transimpedance amplifier, a low-power operational amplifier, and a multiplexer. The source and drain pins of the silicon nanowire field-effect transistor are respectively connected to the positive and negative input pins of the corresponding instrumentation amplifier. The photoelectric signal output pin of the upconversion nanoparticle optical sensor is connected to the input pin of the first transimpedance amplifier. The electrode output pin of the microbial fuel cell sensor is connected to... The analog output pin of the temperature, salinity, turbidity, and dissolved oxygen multi-parameter integrated probe is connected to the input pin of the multiplexer. The output pins of the instrumentation amplifier, the first transimpedance amplifier circuit, the micropower operational amplifier circuit, and the multiplexer are respectively connected to the analog input channel pins of the analog-to-digital converter. The serial clock pin and serial data pin of the analog-to-digital converter are respectively connected to the integrated circuit bus clock pin and integrated circuit bus data pin of the microcontroller. The transmit pin of the microcontroller's universal asynchronous transceiver is connected to the driver input pin of the communication interface chip. The receive pin of the microcontroller's universal asynchronous transceiver is connected to the receiver output pin of the communication interface chip. The differential data output positive pin and differential data output negative pin of the communication interface chip are connected to the standard serial bus interface.

3. The intelligent water quality detection system according to claim 2, characterized in that, The nanosensor node circuit (101) also includes a power management circuit, which includes a lithium battery, a charging management chip, a buck switching regulator chip, and a low dropout linear regulator chip. The positive power input pin of the charging management chip can be used for power connection. The battery charging pin of the charging management chip is connected to the positive terminal of the lithium battery. The positive terminal of the lithium battery is connected to the voltage input pin of the buck switching regulator chip. The voltage output pin of the buck switching regulator chip is connected to the voltage input pin of the low dropout linear regulator chip. The voltage output pin of the low dropout linear regulator chip provides operating voltage for the microcontroller, analog-to-digital converter, and communication interface chip. The nanosensor node circuit (101) is connected to the corresponding downlink serial communication port of the edge gateway circuit in the edge computing module (2) through its standard serial bus interface.

4. The intelligent water quality detection system according to claim 1, characterized in that, The field sensing module (1) further includes a surface acoustic wave microfluidic sorting array circuit (102). The surface acoustic wave microfluidic sorting array circuit (102) includes a main controller, a radio frequency signal generator, a power amplifier, an interdigital transducer array, an optical detection unit, and a communication control interface chip. The optical detection unit includes an ultraviolet-visible fiber optic spectrometer and a photomultiplier tube module. The serial peripheral interface master device output slave device input pin, serial peripheral interface master device input slave device output pin, and serial peripheral interface clock pin of the main controller are respectively connected to the serial data input pin, serial data output pin, and serial clock input pin of the radio frequency signal generator. The radio frequency signal output pin of the radio frequency signal generator is connected to the power amplifier through a coaxial transmission line. The RF signal input pin of the power amplifier and the RF signal output pin of the power amplifier are connected to the signal input pad of the corresponding interdigital transducer in the interdigital transducer array through an impedance matching network composed of inductors and capacitors. The transmit pin and receive pin of the universal asynchronous transceiver of the ultraviolet-visible fiber optic spectrometer are respectively connected to the driver input pin and receiver output pin of the communication control interface chip. The high voltage power enable control pin of the photomultiplier tube module is connected to the digital output pin of the main controller. Its analog current output pin is connected to another built-in analog-to-digital conversion analog input channel pin of the main controller through a second transimpedance amplifier. The differential data line of the communication control interface chip is connected to the downlink communication interface of the edge computing module (2) through a first industrial Ethernet cable.

5. The intelligent water quality detection system according to claim 4, characterized in that, The surface acoustic wave microfluidic sorting array circuit (102) further includes a fluid control unit, which includes a motor drive chip, a microfluidic peristaltic pump, a solenoid valve, a flow sensor, and a pressure sensor. The multiple pulse width modulation signal output pins of the main controller are respectively connected to the enable control pins of multiple motor drive chips. The power output pins of each motor drive chip are respectively connected to the DC motor control pins of the corresponding microfluidic peristaltic pump. The multiple general-purpose input / output pins of the main controller drive the coil pins of multiple solenoid valves through a Darlington transistor array chip. The serial clock pin and serial data pin of the integrated circuit bus of the flow sensor are respectively connected to the spare integrated circuit bus clock pin and spare integrated circuit bus data pin of the main controller. The analog voltage output pin of the pressure sensor is connected to the analog input channel pin of the built-in analog-to-digital converter of the main controller.

6. The intelligent water quality detection system according to claim 1, characterized in that, The on-site sensing module (1) further includes an in-situ photochemical reaction chamber circuit (103). The in-situ photochemical reaction chamber circuit (103) includes a programmable logic controller module, a light source control unit, a sample reaction and temperature control unit, an online detection unit, and an industrial communication interface module. The light source control unit includes a tunable light-emitting diode array, a light-emitting diode constant current driving chip, and a laser diode driving power supply. The sample reaction and temperature control unit includes a quartz reaction cell, a magnetic stirrer, a temperature sensor, and a pH electrode. The online detection unit includes a miniature mass spectrometer, an electron paramagnetic resonance spectrometer, and a time-correlated single-photon counter. The multiple analog output channel pins of the programmable logic controller module are respectively connected to the pulse width modulation dimming pins of each light-emitting diode constant current driving chip. The current output pin of each light-emitting diode constant current driving chip is connected to the positive electrode of the tunable light-emitting diode array. The laser diode... The analog enable control pin of the drive power supply is connected to the digital output channel pin of the programmable logic controller module. The motor speed control pin and power control pin of the magnetic stirrer are connected to the analog output channel pin of the programmable logic controller module. The resistance signal output pin of the temperature sensor and the analog voltage signal output pin of the pH electrode are respectively connected to the analog input channel pin of the programmable logic controller module. The signal output terminals of the micro mass spectrometer, electron paramagnetic resonance spectrometer and time-correlated single-photon counter are respectively connected to the expansion port of the programmable logic controller module through a high-speed data acquisition card. The industrial fieldbus interface of the programmable logic controller module is converted to the standard industrial Ethernet protocol through the industrial communication interface module and connected to an industrial Ethernet switch. The in-situ photochemical reaction chamber circuit (103) is connected to the downlink communication interface of the edge computing module (2) through the industrial Ethernet switch.

7. The intelligent water quality detection system according to claim 1, characterized in that, The edge computing module (2) includes an edge gateway circuit, which includes an embedded industrial control board core module, a field-programmable gate array (FPGA) accelerator card, a communication interface array, and a data preprocessing and security module. The data preprocessing and security module includes an encryption chip and a real-time clock chip. The embedded industrial control board core module is connected to the FPGA accelerator card. The multiple universal asynchronous transceiver transmit and receive pins of the embedded industrial control board core module are respectively connected to the driver input pins and receiver output pins of the communication interface array. The differential output pins of the communication interface array are connected to the nanosensors in the field sensing module (1) through their standard serial bus interface and controller area network (CLAN) bus interface, respectively. The device node circuit (101) is connected, and the Ethernet controller medium-independent interface of the embedded industrial control board core module is connected to the surface acoustic wave microfluidic sorting array circuit (102) and the in-situ photochemical reaction chamber circuit (103). The secure digital input / output interface of the embedded industrial control board core module is connected to the server cluster circuit (301) in the central decision layer module for data communication through the fifth generation mobile communication technology module. The serial clock pin and serial data pin of the integrated circuit bus of the encryption chip are respectively connected to the integrated circuit bus clock pin and data pin of the embedded industrial control board core module. The interrupt output pin of the real-time clock chip is connected to the external interrupt input pin of the embedded industrial control board core module.

8. The intelligent water quality detection system according to claim 1, characterized in that, The central decision module (3) includes a server cluster circuit (301), which includes multiple parallel computing servers, a centralized data storage server, and a network switching device. The parallel computing server includes a central processing unit (CPU) chip, a graphics processing unit (GPU) accelerator card, a double data rate (DFR) memory module, and a solid-state drive (SSD) storage device. The centralized data storage server includes an independent disk redundancy array (RDA) controller, a mechanical hard disk array (HDD) array, and a high-speed SSD cache. The network switching device includes a core layer switch and multiple access layer switches. The CPU chips in each parallel computing server are interconnected. Each GPU accelerator card is connected to the parallel computing server. The DFR memory module is connected to the CPU chip. The SSD storage device is connected to the CPU chip via a non-volatile memory high-speed interface. The DRA controller connects the HDD array and the high-speed SSD cache to the CPU chip. The CPU chip is connected to the core layer switch. The access layer switches are connected to the stacked ports of the core layer switches to form an internal network. The uplink communication interface of the edge gateway circuit is connected to the corresponding port of the access layer switch. The graphics workstation in the digital twin and visualization circuit (302) is connected to the core layer switch.

9. The intelligent water quality detection system according to claim 8, characterized in that, The central decision module (3) also includes a digital twin and visualization circuit (302), which includes a graphics workstation, a splicing processor and multiple display units. The display port output interfaces of the multiple graphics cards of the graphics workstation are connected to the multiple display port input interfaces of the splicing processor through display port cables. The multiple high-definition multimedia interface output ports of the splicing processor are connected to the high-definition multimedia interface input ports of the multiple display units through high-definition multimedia interface cables. The network interface of the graphics workstation is connected to the network switching device in the central decision module (3) through fiber optic patch cords.

10. A smart water quality detection method, executed by the smart water quality detection system as described in any one of claims 1-9, characterized in that, Includes the following steps: The S1 field perception module (1) collects multi-source heterogeneous data, preprocesses and aligns the data spatiotemporally to generate a standardized input dataset; The S2 edge computing module (2) analyzes the mass spectrometry data and generates an information list containing known and unknown risk substances through feature extraction and structure inference. The S3 central decision module (3) models mixed toxicity effects based on biotoxicity test data and chemical substance inventory, and quantifies the toxicity contribution of various pollutants; The S4 central decision module (3) uses a chemical fingerprint database and a Bayesian inference model to conduct source analysis of pollutants, calculate and dynamically track the contribution of each potential pollution source; The S5 central decision-making module (3) integrates chemical, ecological and source tracing risks, calculates dynamic environmental risk index, and sets graded early warning thresholds; The S6 central decision module (3) generates control decision schemes through multi-objective optimization and scenario simulation, and drives the digital twin and visualization circuit (302) to perform simulation and display.

Images