A depth and quantity fixed intelligent water quality sampling control method and system

Abstract

The application discloses a kind of depth setting quantitative intelligent water quality sampling control method and system, belong to water quality analysis and monitoring technical field, it includes by obtaining environmental perception data, the depth deviation trend of preset sampling head is predicted to generate active stable control instruction;Dynamic response data of sampling process is obtained, the single pulse effective displacement of sampling pump group is calibrated online, and real-time displacement coefficient is obtained, and flow integration is carried out based on real-time displacement coefficient until reaching preset sampling volume;Water quality parameters of next sampling point are obtained and adaptive cleaning scheme is generated by querying cleaning knowledge base;System state evaluation result is obtained by diagnosis based on dynamic response data and environmental perception data;Based on the change characteristics of the real-time monitoring data of online water quality sensor and the system state evaluation result, adaptive sampling strategy is generated and executed.Combination of prediction compensation, online calibration and adaptive decision, a collaborative control mechanism is used to achieve high precision, high efficiency and intelligent sampling in complex water environment.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of water quality analysis and monitoring technology, in particular to a depth and quantity fixed intelligent water quality sampling control method and system. BACKGROUND

[0002] Water sampling is a basic link in water environment monitoring and scientific research, and its purpose is to obtain water samples that can represent specific time and space characteristics in natural water bodies such as rivers, lakes and oceans for subsequent laboratory chemical or physical property analysis. The water sampling control system is usually an automatic device integrating sensors, controllers and actuators, which controls the sampling head to dive to a specified depth through a pre-set program, and extracts a specified volume of water sample, and the accuracy and representativeness of the sampling directly determine the effectiveness of the final analysis results.

[0003] In the related art, the Chinese invention patent application with the publication number CN116338128A discloses a flowing water body water quality monitoring and early warning system and method, which comprises: a plurality of monitoring units distributed along the flow direction of the flowing water body and a remote monitoring unit in communication connection with the monitoring units; the monitoring unit comprises a sensor module, a water quality sampling unit and a control transmission unit; the water quality sampling unit comprises a water pump and a sampling device, the water pump extracts water samples from the measured flowing water body and outputs them to the sampling device, the sensor module is arranged in the sampling device and detects the water samples, the output end of the sensor module is connected to the control transmission unit, the control transmission unit is in communication connection with the remote monitoring unit, and the control transmission unit controls the water pump to work.

[0004] However, the above-mentioned related technology still has limitations when facing complex and dynamically changing water environments. First, in rivers, lakes and reservoirs with large flow rate, turbulent flow or turbulent waves, the sampling head is easily impacted by fluid resistance to produce displacement deviation, and it is difficult to offset the dynamic deviation of the physical depth by relying only on the pre-set diving instruction of the depth, resulting in a lack of representativeness of the sampling site. Secondly, the existing quantitative control method is mostly based on fixed pump speed, running time or factory-calibrated displacement coefficient, without considering the real-time influence of pipeline aging, fluid viscosity fluctuation, pipeline resistance change and bubble interference on single-pulse effective displacement during sampling, resulting in a cumulative error that cannot be ignored in the collected volume. In addition, the related technology usually executes a fixed cleaning and sampling process, lacks adaptive quality control logic for the pollution gradient difference between the front and rear sites, and cannot dynamically adjust the sampling strategy according to its own health status such as filter screen blockage risk and online monitored water quality mutation characteristics. This passive execution rather than active perception decision-making mode results in insufficient robustness in harsh working conditions, and makes it difficult to accurately capture instantaneous water quality abnormal events, limiting the scientific value and research depth of the monitoring data. SUMMARY

[0005] To solve the above problems, the application provides a depth-constant and quantitative intelligent water quality sampling control method and system, which adopts a cooperative control mechanism integrating prediction compensation, online calibration and adaptive decision-making, and can realize high-precision, high-efficiency and intelligent sampling in complex water environment.

[0006] The above object can be achieved by the following scheme:

[0007] A depth-constant and quantitative intelligent water quality sampling control method comprises the following steps: acquiring environmental perception data comprising pressure data, temperature-salinity data and attitude data; predicting a depth deviation trend of a preset sampling head based on the environmental perception data, and generating an active stabilization control instruction for driving a preset lifting mechanism to perform advance depth compensation; after the sampling head is stabilized at a target depth, acquiring dynamic response data of a sampling process, wherein the dynamic response data comprises operating parameters of a sampling pump group and pressure fluctuation data of a pipeline; based on the dynamic response data, calibrating a single-pulse effective displacement of the sampling pump group online to obtain a real-time displacement coefficient, and performing flow integration based on the real-time displacement coefficient until a preset sampling volume is reached; after sampling is completed, acquiring water quality parameters of a next sampling point and querying a cleaning knowledge base storing historical cleaning records to generate an adaptive cleaning scheme; based on the dynamic response data and the environmental perception data, performing diagnosis to obtain a system state evaluation result comprising an abnormal risk probability; based on a change feature of real-time monitoring data of an online water quality sensor and the system state evaluation result, generating and executing an adaptive sampling strategy.

[0008] Optionally, the steps of predicting a depth deviation trend of a preset sampling head and generating an active stabilization control instruction for driving a preset lifting mechanism to perform advance depth compensation comprise the following steps: calculating a current theoretical depth of the sampling head based on the pressure data and the temperature-salinity data; calculating a current inclination angle of the sampling head and a depth compensation amount based on the attitude data; acquiring a current traction force parameter of a cable traction mechanism, and combining the current inclination angle to predict a future depth deviation amount and an attitude change trend caused by water flow; and generating an active stabilization control instruction comprising a feedforward compensation component according to the future depth deviation amount and the attitude change trend.

[0009] Optionally, the step of acquiring a current traction force parameter of a cable traction mechanism comprises the following step: acquiring a traction force parameter representing a current cable tension through a current sensor or a torque sensor arranged in a motor driver of the cable traction mechanism.

[0010] Optionally, the online calibration of the real-time displacement coefficient of the single-pulse effective displacement of the sampling pump group based on the dynamic response data comprises: controlling the sampling pump group to run at a set of test pulses, and synchronously collecting pressure fluctuation waveforms and pump group motor current curves as calibration input data; performing feature analysis on the calibration input data to extract pressure build-up rate and fluctuation damping features; and calculating the real-time displacement coefficient according to the pressure build-up rate and fluctuation damping features.

[0011] Optionally, the generation of the adaptive cleaning scheme comprises: obtaining real-time online water quality parameters of the next sampling point; querying historical cleaning records with similar pollution characteristics in the cleaning knowledge base according to the real-time online water quality parameters; predicting specific residual risks possibly generated in the pipeline after sampling based on the historical cleaning records; determining the cleaning stage, cleaning medium and intensity of the cleaning action according to the specific residual risks, and forming the adaptive cleaning scheme.

[0012] Optionally, the cleaning knowledge base comprises: recording sampling pipeline residual detection data under different water quality parameter combinations in historical sampling tasks; recording different cleaning schemes adopted for different water quality parameter combinations and their final cleaning effect verification data; establishing a mapping relationship between the water quality parameter combinations, the cleaning schemes and the cleaning effect verification data, and continuously updating the cleaning knowledge base.

[0013] Optionally, the abnormal risk probability comprises: extracting pump group current spectrum features and pressure fluctuation feature frequencies from the dynamic response data; extracting flow velocity disturbance features of the current water area from the environmental perception data; performing fusion analysis on the pump group current spectrum features, the pressure fluctuation feature frequencies and the flow velocity disturbance features to evaluate filter screen clogging risk probability and pipeline leakage risk probability; and taking the filter screen clogging risk probability and the pipeline leakage risk probability as the abnormal risk probability in the system state evaluation result.

[0014] Optionally, the generation and execution of the adaptive sampling strategy based on the change characteristics of the real-time monitoring data of the online water quality sensor and the system state evaluation result comprises: calculating a change gradient of the real-time monitoring data over time or over depth; determining that a suspected water quality event occurs when the change gradient exceeds a preset event triggering threshold; generating an emergency sampling scheme according to the spatial position or time attribute of the suspected water quality event in combination with the current system state evaluation result; and replacing the original preset sampling plan with the emergency sampling scheme and executing the same.

[0015] Optionally, the generating the emergency sampling scheme comprises: generating an emergency sampling scheme containing multi-layer encrypted sampling when the filter clogging risk probability in the system state evaluation result is lower than a preset safety threshold; generating an optimization scheme reducing single sampling volume and increasing preventive flushing instructions in the emergency sampling scheme when the filter clogging risk probability is higher than the safety threshold.

[0016] Based on the same inventive concept, the application further provides a depth- and quantity-determining intelligent water quality sampling control system, which comprises:

[0017] a multi-dimensional environment perception module for acquiring environment perception data containing pressure data, temperature-salinity data and attitude data;

[0018] an active depth stabilization module for predicting a depth deviation trend of a preset sampling head based on the environment perception data and generating an active stabilization control instruction for driving a preset lifting mechanism to perform advanced depth compensation;

[0019] a sampling process perception module for acquiring dynamic response data of a sampling process after the sampling head is stabilized at a target depth, wherein the dynamic response data contains operating parameters of a sampling pump group and pressure fluctuation data of a pipeline;

[0020] an intelligent quantity control module for online calibrating a single-pulse effective displacement of the sampling pump group based on the dynamic response data to obtain a real-time displacement coefficient and performing flow integration based on the real-time displacement coefficient until a preset sampling volume is reached;

[0021] an adaptive cleaning scheme module for acquiring water quality parameters of a next sampling point after sampling is completed and querying a cleaning knowledge base storing historical cleaning records to generate an adaptive cleaning scheme;

[0022] a system health diagnosis module for diagnosing based on the dynamic response data and the environment perception data to obtain a system state evaluation result containing an abnormal risk probability;

[0023] an intelligent strategy scheduling module for generating and executing an adaptive sampling strategy based on a change feature of real-time monitoring data of an online water quality sensor and the system state evaluation result.

[0024] Compared with the prior art, the application has the following advantages:

[0025] 1. Through the deep coupling of a series of means such as predictive compensation, online calibration, adaptive cleaning and intelligent decision-making, high-precision and high-reliability water quality sampling in complex dynamic water areas is realized. The advanced deep compensation mechanism can actively resist the interference of water flow and waves, ensuring that the sampling head is accurately and stably positioned on the target isobath, thereby fundamentally ensuring the spatial representativeness of the sampling data. At the same time, the quantitative method of online self-calibration can correct the measurement error caused by changes in the equipment working condition in real time, ensuring the long-term accuracy of the sampling volume and providing high-quality raw samples for subsequent precise chemical analysis.

[0026] 2. The automation efficiency of water quality sampling and the reliability of equipment operation are improved. The adaptive cleaning strategy based on water quality difference prediction can intelligently match the cleaning intensity according to the actual pollution risk, significantly shortening the preparation time while ensuring extremely low cross-contamination rate compared to traditional fixed processes. In addition, it also has online composite diagnosis and risk prediction capabilities, which can identify potential faults such as filter screen blockage in advance and perform preventive maintenance, effectively reducing the failure rate and the need for manual intervention in field operations, and enhancing the stability of long-term autonomous operation of the equipment.

[0027] 3. The sampling methodology has made a leap from passive execution to intelligent exploration, improving the scientific value of monitoring data. This method gives the system the ability to perceive environmental changes and autonomously respond scientifically. When an abnormal gradient or mutation of water body parameters is identified, it can automatically trigger encrypted or high-frequency sampling, thereby accurately capturing transient pollution events or fine water layering structures that are easily missed by traditional timed sampling modes. This event-driven adaptive sampling strategy enables the sampling equipment to evolve from a simple task execution tool to an intelligent detection platform that can actively discover scientific phenomena.

[0028] Other features and advantages of the present application will be set forth in the following description, and in part will be apparent from the description, or can be learned by practice of the present application. The objects and other advantages of the present application will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings. BRIEF DESCRIPTION OF DRAWINGS

[0029] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0030] Figure 1 is a flowchart of a depth-setting and quantitative intelligent water quality sampling control method according to an embodiment of the present application.

[0031] Figure 2 is a posture compensation characteristic surface schematic diagram of a depth calculation model of a depth-quantitative intelligent water quality sampling control method and system according to an embodiment of the present application.

[0032] Figure 3 is a sampling pump group current signal spectrum feature comparison diagram schematic diagram of a depth-quantitative intelligent water quality sampling control method and system according to an embodiment of the present application.

[0033] Figure 4 is a structural schematic diagram of a depth-quantitative intelligent water quality sampling control system according to an embodiment of the present application. DETAILED DESCRIPTION

[0034] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0035] With reference to Figure 1 , one embodiment of the present application proposes a depth-quantitative intelligent water quality sampling control method, which adopts a cooperative control mechanism integrating prediction compensation, online calibration and adaptive decision-making, and can realize high-precision, high-efficiency and intelligent sampling in complex water environment.

[0036] The method of the embodiment specifically includes:

[0037] obtain environment perception data containing pressure data, temperature-salinity data and posture data;

[0038] based on the environment perception data, predict the depth offset trend of a preset sampling head, and generate active stabilization control instructions for driving a preset lifting mechanism to perform advance depth compensation;

[0039] after the sampling head is stabilized at a target depth, obtain dynamic response data of the sampling process, wherein the dynamic response data contains operating parameters of a sampling pump group and pressure fluctuation data of a pipeline;

[0040] based on the dynamic response data, calibrate the single-pulse effective displacement of the sampling pump group online to obtain a real-time displacement coefficient, and perform flow integration based on the real-time displacement coefficient until a preset sampling volume is reached;

[0041] after the sampling is completed, obtain water quality parameters of a next sampling point and query a cleaning knowledge base storing historical cleaning records, and generate an adaptive cleaning scheme;

[0042] diagnose based on the dynamic response data and the environment perception data to obtain a system state evaluation result containing an abnormal risk probability;

[0043] generate and execute an adaptive sampling strategy based on the change characteristics of the real-time monitoring data of the online water quality sensor and the system state evaluation result.

[0044] Specifically, the external environmental disturbance is predicted by a prediction model, and a leading compensation action is applied to actively stabilize the spatial position of the sampling head. On this basis of stability, the system dynamic response during the sampling process is taken as a self-detection signal to perform real-time online calibration of the performance of the core executive component, thereby realizing high-precision quantitative control matching the current real state of the system. An intelligent decision-making mechanism based on a knowledge base is introduced, the pollution risk is predicted according to the characteristics of the sample to be measured, and the most efficient adaptive cleaning scheme is dynamically generated. The internal operation characteristics and the external environment data are fused for diagnosis to evaluate the potential abnormal risk. The perception ability of the water body environment change and the evaluation result of the health state of the system are combined to autonomously generate and evolve the sampling strategy, so that the whole sampling behavior is changed from static execution to dynamic intelligent exploration. The leading deep compensation improves the depth accuracy and stability in the dynamic water area, ensuring the spatial representativeness of the sampling data. Secondly, the online self-calibration quantitative method solves the metering error caused by equipment aging or working condition change, ensuring the long-term high precision and high repeatability of the sampling volume. The adaptive cleaning scheme ensures the sample purity and eliminates cross contamination, while shortening the sampling preparation time and greatly improving the continuous operation efficiency. In addition, the compound diagnosis and risk prediction ability endows the system with high autonomy and reliability, reducing the sampling interruption caused by unexpected failures. Finally, the event-driven adaptive sampling strategy enables the device to intelligently capture instantaneous or local key environmental changes, improving the scientific value of the obtained data.

[0045] Optionally, the depth offset trend of the preset sampling head is predicted, and an active stabilization control instruction for driving the preset lifting mechanism to perform leading depth compensation is generated.

[0046] calculate the current theoretical depth of the sampling head based on the pressure data and the temperature-salinity data;

[0047] calculate the current inclination angle and depth compensation amount of the sampling head based on the attitude data;

[0048] obtain the current traction force parameter of the cable traction mechanism, and combine the current inclination angle to predict the future depth offset amount and attitude change trend caused by the water flow;

[0049] generate an active stabilization control instruction containing a feedforward compensation component according to the future depth offset amount and attitude change trend.

[0050] Specifically, the pressure data of the sampling head, the water temperature and salinity data, and the attitude data are obtained as initial environmental perception data. The system calculates the real-time density of the current water area by consulting the preset water state equation using the temperature and salinity data, and calculates the current inclination angle of the sampling head relative to the vertical direction based on the attitude data. The accurate vertical depth of the sampling head at the current time is calculated by the following formula :

[0051] ;

[0052] wherein P is the hydrostatic pressure obtained by the pressure sensor, and g is the local gravity acceleration. The accurate vertical depth is compared with the preset target depth to obtain the static error of the current depth, that is, the depth compensation amount. The current traction force parameter reflecting the water flow drag force is obtained by monitoring the driving current or torque of the cable traction motor in the lifting mechanism. The system inputs the current traction force parameter and the current inclination angle into the preset fluid dynamics prediction model. The model can be simplified as a second-order dynamic system in engineering, which regards the sampling head as an underwater pendulum. The model takes the current traction force parameter as the main indicator of external disturbance force, and calculates the depth deviation and attitude change trend caused by the water flow impact in the future preset period of 0.5 seconds to 2.0 seconds. The active stabilization control instruction containing the feedforward compensation component is generated. The instruction is composed of two parts:

[0053] ;

[0054] The first three terms constitute a standard PID feedback controller for correcting the existing depth compensation amount ΔH. are proportional, integral, and differential coefficients, respectively. The fourth term represents the feedforward compensation component, which is a control amount positively related to the depth deviation and the attitude change trend but in the opposite direction. For example, if the model predicts that the sampling head will be pressed down by 10 cm by the water flow in the next second, the feedforward compensation component will generate an instruction to drive the lifting mechanism to pre-lift the cable length by about 10 cm. Finally, the complete active stabilization control instruction is sent to the lifting mechanism and the cable traction mechanism to perform the composite compensation operation containing preventive actions, thereby realizing the advanced stabilization control of the sampling head position. As Figure 2 shown is a schematic diagram of the attitude compensation characteristic surface of the depth calculation model, which shows how to combine the hydrostatic pressure P to real-time correct and calculate the true vertical depth H when the sampling head is tilted by the water flow impact.

[0055] For example, in nearshore watershed monitoring conditions, the real-time pressure sensor reading P is 102500 Pa at a water depth of approximately 10 m, and the temperature T is 25°C. C, Salinity S is 30 PSU, attitude angle θ is 15 Generated by the impact of water flow. The system is based on a simplified formula of the International Equation of State for Seawater. The calculated real-time density ρ≈1019.8 kg / m³. Therefore... If the preset target depth is 10.0m, then the static error ΔH = 0.09m. At this time, the traction force parameter is 50N. Inputting this into the second-order pendulum dynamic model, derived based on Newton's laws of motion and considering the fluid resistance coefficient Cd = 1.05, the predicted offset after 1.0s is 0.12m, with a trend of increasing by 5. The default PID parameters are set to Kp=0.8, Ki=0.1, and Kd=0.05. The feedforward function f is defined as linear compensation kff×ΔHpred, where kff is set to a default value of 0.9 to prevent overshoot. Ignoring the instantaneous values ​​of integrals and derivatives... The lifting mechanism was instructed to immediately retract the cable by 0.18m, with a lead of 0.108m effectively offsetting the predicted downward pressure, ensuring that the sampling head had been actively reset before the water flow disturbance arrived.

[0056] Optionally, obtaining the current traction force parameters of the cable traction mechanism includes:

[0057] By using a current sensor or torque sensor installed in the motor driver of the cable traction mechanism, traction parameters characterizing the current cable tension are obtained.

[0058] Specifically, the status of the servo system of the cable traction mechanism is passively monitored. In the cable traction mechanism, the motor used for cable winding and unwinding is controlled by a motor driver, and traction force parameters are acquired through sensors installed inside the motor driver. In a preferred embodiment, a high-precision current sensor, such as a Hall effect sensor, integrated on the motor driver circuit board is used to measure the operating current of the drive motor in real time. The physical basis for this measurement is that when the water flow exerts tension on the cable and sampling head, the motor must output a larger torque to counteract this tension in order to maintain positional stability or uniform motion. For DC or AC servo motors, the output torque is proportional to the input current within a certain range. Therefore, changes in motor current can accurately reflect changes in cable tension. After acquiring the raw motor current, a data processing step is performed to obtain the final traction force parameters. This processing is shown in the following formula:

[0059] ;

[0060] in, I0 is the reference current of the motor when it runs without external load, i.e. only to overcome its own and the internal friction of the gearbox, etc. The value is calibrated and stored when the equipment is shipped or regularly maintained. I is the current value measured by the sensor in real time. k is the comprehensive conversion coefficient, which will convert the current difference value into the dimension of force in engineering. Its value depends on the torque constant of the motor, the reduction ratio of the gearbox, and the radius of the winch drum. It is also determined by the equipment calibration process. The calculated traction force parameter, with the unit of Newton, directly quantifies the tension applied on the cable by external factors such as water flow, providing real-time and accurate external disturbance force input for the subsequent dynamic prediction model.

[0061] Exemplarily, the traction mechanism is driven by a permanent magnet synchronous servo motor, and the Hall sensor is integrated on the DC bus side of the driver. The real-time measured motor operating current is 4.2A. It is known that the no-load current measured by the motor during factory calibration is 0.8A. The calculation of the conversion coefficient k is determined according to the torque constant of the motor , the reduction ratio n = 10 of the speed reducer, and the radius r = 0.05m of the drum, and the derivation process is: . Substituting the formula gives . The calculation result 340N represents the total external force applied on the entire underwater sampling system by the current. This traction force parameter, as a real-time external disturbance force input, directly participates in the prediction model, enabling the prediction model to perceive the change of water flow speed in real time according to the sudden change of tension, and accurately calculate the displacement offset that will occur, solving the problem of response lag in conventional depth control due to ignoring the cable tension.

[0062] Optionally, the real-time displacement coefficient is obtained by online calibration of the single-pulse effective displacement of the sampling pump group based on the dynamic response data, which comprises:

[0063] The sampling pump group is controlled to run with a set of test pulses, and the pressure fluctuation waveform and pump group motor current curve are synchronously collected as calibration input data;

[0064] The calibration input data is analyzed for features, and the pressure build-up rate and fluctuation damping characteristics are extracted;

[0065] According to the pressure build-up rate and fluctuation damping characteristics, the real-time displacement coefficient is calculated.

[0066] Specifically, after completing the pipeline air exhaust and switching to the precision sampling mode, the central controller instructs the sampling pump group to execute a preset test pulse sequence, for example, a short pulse sequence containing 3 to 5 different frequencies, the frequency range can be between 5 Hz to 20 Hz. During this test pulse operation, two key data streams are synchronously collected. One is the pressure fluctuation waveform fed back from the pressure sensor at the end of the pipeline, and the other is the motor current curve fed back from the pump motor driver. These two synchronized time series data together constitute the calibration input data of this calibration. Real-time feature analysis is performed on the calibration input data. The pressure build-up rate is extracted from the pressure fluctuation waveform, which reflects the ability of the pump to overcome system resistance to build up pressure, and is usually obtained by calculating the first derivative of the pressure curve rising segment or the average slope in a specified time window. Secondly, the fluctuation damping feature is extracted, which characterizes the absorption and attenuation effect of the pipeline system on the pressure fluctuation caused by the pump pulse, and is closely related to the real-time working conditions such as pipe wall elasticity and fluid viscosity, and is usually quantified by extracting the main frequency amplitude decay rate through Fourier transform or wavelet analysis of the pressure fluctuation waveform. The extracted pressure build-up rate and fluctuation damping feature are substituted into the preset pump-pipeline system model to calculate the real-time displacement coefficient of this sampling. The model is an empirical or semi-empirical formula that relates the theoretical displacement of the pump, the pipeline damping, and the fluid characteristics, as shown in the following formula:

[0067] ;

[0068] ;

[0069] wherein, is the real-time displacement coefficient used for flow integration, is the theoretical single-pulse displacement coefficient calibrated at the pump factory. is the dynamic calibration factor, the value of which is calculated by the function g. g is a mapping function or a calibration table stored in the controller, which maps the measured pressure build-up rate dP / dt and the fluctuation damping feature to the corresponding calibration factor. This function g is calibrated through a large number of experiments during the development stage of the device. After calibration, the calibration input data obtained this time, especially the pressure fluctuation feature frequency and the current spectrum feature, will be stored and used as basic feature data for subsequent composite diagnosis, to judge the system health status by comparing with the real-time data during sampling.

[0070] Exemplarily, during the sampling task start-up stage, the test pulse with execution frequencies of 5 Hz, 10 Hz, and 15 Hz is executed, and the theoretical displacement of the peristaltic pump is 0.10 mL / pulse. The pressure sensor waveform is recorded synchronously, and the pressure build-up rate is extracted by differential calculation to obtain the pressure build-up rate under the 10 Hz pulse Wave fluctuation damping characteristics are calculated by wavelet analysis of pressure wave peak attenuation ratio The above parameters reflect the decrease of elastic modulus and the increase of liquid viscosity due to continuous use of the pump pipe. The mapping function g adopts a linear weighting model , and the reference value is set as , The sensitivity coefficients a = 0.0002 and b = 0.3. The calibration factor is calculated by substituting the formula . The calibration process explains the phenomenon that the actual displacement is higher than the theoretical value due to the decrease of liquid viscosity, which makes the pressure build up faster. Then, the flow integral is performed with 0.1045 as the step reference, and when the pulse number reaches 957, 957 x 0.1045 ≈ 100 mL, the sampling is stopped to achieve accurate quantification, and the calibration error is reduced from the traditional 5% to within 0.5%.

[0071] Optionally, the generating the adaptive cleaning scheme comprises:

[0072] Obtaining the real-time online water quality parameters of the next sampling point;

[0073] In the cleaning knowledge base, querying historical cleaning records with similar pollution characteristics as the real-time online water quality parameters;

[0074] Based on the historical cleaning records, predicting the specific residual risk that may be generated in the pipeline after this sampling;

[0075] According to the specific residual risk, determining the cleaning stage, cleaning medium and intensity of cleaning action to form an adaptive cleaning scheme.

[0076] ​Specifically, after finishing the last sampling, before entering the next sampling point, the integrated online water quality sensor is activated first to obtain the real-time online water quality parameters of the next sampling point, mainly including conductivity and turbidity. At the same time, the water quality parameters recorded at the end of the last sampling, i.e. conductivity and turbidity, are retrieved from the memory. The current water quality parameter changes are taken as a query vector, and a search is performed in the preset cleaning knowledge base. The cleaning knowledge base is a database or a precision table in engineering, which stores a large number of historical cleaning records. Each historical cleaning record contains historical water quality parameter changes, the cleaning scheme used at the time, and the residual concentration verified afterwards. Through a similarity matching algorithm, a set of historical cleaning records closest to the current query vector is found. Based on the retrieved historical cleaning records, the risk of specific residue that may be generated in the pipeline after this sampling is predicted through a preset risk assessment model. This risk value is a comprehensive score, which quantifies the possibility and severity of cross-contamination. For example, if the historical records show that when switching from low-salt high-organic water samples to high-salt low-organic water samples, even after standard flushing, there is still a high adsorption of organic matter on the pipe wall, then when encountering a similar situation, a higher risk of specific residue will be generated. This risk of specific residue is used as the core input to dynamically generate an adaptive cleaning scheme containing specific execution parameters. This scheme is not a fixed three-step approach, but a set of instructions dynamically adjusted according to the risk value. The process can be represented by the following formula group to represent its logic:

[0077] ;

[0078] ;

[0079] ;

[0080] where f is a decision function that selects the combination of cleaning strategies according to the size of . For example, when is low, the scheme may only include a standard volume of pre-flushing; when is high, the scheme may include air emptying, multiple pre-flushing and high-intensity pulse flushing, etc. and represent the key pre-flushing times and single flushing volume in the scheme, respectively. and are the minimum basic cleaning parameters, and and are the weight coefficients obtained from the cleaning knowledge base, which convert the dimensionless risk value into specific cleaning parameter increments. The adaptive cleaning scheme generated finally contains a series of precise instructions such as cleaning stage, flushing times, flushing volume, flushing flow rate and rhythm, etc., and is sent to the execution mechanism for execution.

[0081] Exemplarily, the system completes a sampling of conductivity Cprev= 0.5 ms / cm (fresh water) and turbidity Tprev= 5 NTU, and detects a next point Cnext= 0.8 ms / cm (brackish water) and turbidity Tnext= 80 NTU (high turbidity). In the cleaning knowledge base, a similar water quality jump vector is searched, and the matched historical residual experimental data show that when the water quality changes from low salt and low turbidity to high salt and high turbidity, the pipe wall is prone to accumulate salt scale to capture suspended solids. The risk assessment model calculates a specific residual risk value of 0.85 in the value range of 0 1. Set the default weight coefficients Wn= 5.0 and Wv= 150.0 mL provided by the knowledge base, and the base values Nbase= 1 and Vbase= 100 mL. Substitute into the formula to obtain , and take the integer part as 5 times. After the decision function f identifies that the specific residual risk value is greater than 0.8, a pneumatic step is automatically inserted at the beginning of the cleaning sequence. The finally generated is: first perform high-pressure air blowing for 10 s, and then use the target point water sample to perform 5 times of pulse pre-flushing, each time 227.5 mL. This scheme effectively blocks the influence of high-concentration residues on subsequent analysis by increasing the flushing frequency and volume, and ensures the chemical purity of the sample.

[0082] Optionally, the cleaning knowledge base comprises:

[0083] record the pipe residual detection data after sampling under different combinations of water quality parameters in the historical sampling tasks;

[0084] record different cleaning schemes adopted for different combinations of water quality parameters and their final cleaning effect verification data;

[0085] establish a mapping relationship between the combinations of water quality parameters, the cleaning schemes and the cleaning effect verification data, and continuously update the cleaning knowledge base.

[0086] ​Specifically, first, a standard solution with a specific combination of water quality parameters is collected, such as a high turbidity or high salinity solution, and then a cleaning solution with clear parameters is executed. Next, a high-purity water sample is collected and analyzed using laboratory high-precision analytical instruments, such as a total organic carbon analyzer or an inductively coupled plasma mass spectrometer, to obtain specific post-sampling pipeline residual detection data. The water quality parameter combination of each experiment, the cleaning solution parameters used, and the final cleaning effect verification data are structured as a complete data tuple for structured storage. A large number of such data tuple sets form the basis of the cleaning knowledge base. After the device is put into actual use, the cleaning knowledge base supports continuous updating. After completing an adaptive cleaning, a quick self-verification is performed through the on-board sensor, and if the verification result deviates from the knowledge base prediction, the complete on-site data of this time is recorded. These new data can be used to iteratively optimize the mapping relationship offline or online, so that the cleaning decision of the system becomes more accurate as the use time grows.

[0087] For example, at the beginning of the device operation, the cleaning knowledge base contains a set of baseline data tuples D=(ΔP, S, E), where ΔP is the conductivity increase , the execution scheme S is 2 times of pre-washing, 150 mL each time, and the expected residual E is less than 0.1%. In an actual operation, the system enters the high salinity industrial wastewater area from the fresh water area, and the conductivity jump value is . The system calculates the scheme according to the knowledge base linear interpolation and executes it. After sampling is completed, the on-board conductivity sensor verifies the pipeline residual, and the residual rate is measured to be 0.15%, which exceeds the expected target. A new data tuple Dnew=([1500,...],[N=2,V=150],0.15%) is automatically created. In the idle stage, the offline optimization algorithm uses Dnew to re-fit the weight coefficient Wv in the formula G, adjusting it from the original 150 to 185, to reflect the need for more thorough cleaning under this specific salinity change. This continuous updating mechanism makes the cleaning scheme more targeted, ensuring that subsequent cleaning volumes generated in similar high-salinity conditions are increased to , thereby achieving reproducible accuracy.

[0088] Optionally, the abnormal risk probability includes:

[0089] Extracting pump group current spectrum features and pressure fluctuation feature frequencies from the dynamic response data;

[0090] Extracting flow rate disturbance features of the current water area from the environmental perception data;

[0091] Fusing and analyzing the pump group current spectrum features, pressure fluctuation feature frequencies, and flow rate disturbance features to evaluate the filter clogging risk probability and the pipeline leakage risk probability;

[0092] The filter clogging risk probability and the pipeline leakage risk probability are taken as the abnormal risk probability in the system state evaluation result.

[0093] Specifically, throughout the sampling execution process, internal state features are continuously extracted from dynamic response data. Specifically, the real-time current signal of the pump group motor is subjected to fast Fourier transform to generate pump group current spectrum features, and attention is focused on whether new abnormal harmonic peaks that do not exist under normal working conditions appear in the 0.1 to 5 Hz low frequency band. Such peaks are usually related to periodic load fluctuations of the motor caused by filter or pipeline partial clogging. At the same time, pressure fluctuation data is analyzed to extract its main frequency pressure fluctuation feature frequency, which is compared with the baseline frequency recorded during initial online calibration. A large frequency shift or amplitude attenuation often indicates a change in fluid passage impedance. External disturbance features are extracted from environmental perception data in parallel. By analyzing the time series data of the cable tension parameter, the mean and variance of the cable tension parameter in a short time window are calculated to quantify the flow disturbance feature of the current water area. A high mean and high variance flow disturbance feature means that the sampling head is in a turbulent and unstable water flow, which increases the external risk of filter clogging by suspended solids. The extracted internal and external features are fused and analyzed to evaluate the abnormal risk probability. The pump group current spectrum features, the shift of the pressure fluctuation feature frequency, and the flow disturbance feature together constitute a multi-dimensional state vector. The state vector is input into a pre-set risk assessment function, as shown in the following formula:

[0094] ;

[0095] wherein, is the input multi-dimensional state vector, is a nonlinear risk assessment function, which can be realized as an offline trained neural network classifier or a set of fuzzy logic reasoning rules in engineering. The output of the function is a probability vector containing two components, i.e., the filter clogging risk probability and the pipeline leakage risk probability. For example, when abnormal low-frequency peaks appear in the pump group current spectrum features, the pressure fluctuation feature frequency deviates significantly from the baseline value, and the flow disturbance feature shows that the water body is agitated, the function will output a higher filter clogging risk probability. Conversely, if the pressure cannot be established suddenly, and the motor current decreases significantly, the function will determine a higher pipeline leakage risk probability. These two probability values together constitute the core part of the system state evaluation result, i.e., the abnormal risk probability. For example, Figure 3The figure shows the comparison of the frequency spectrum characteristics of the pump group motor current after fast Fourier transform. The figure clearly shows that under abnormal conditions such as filter screen or pipeline partial blockage, the characteristic harmonic peak value of about 2.5Hz appears in the low frequency band. The characteristic and the pressure fluctuation frequency offset together constitute the key basis for judging the abnormal risk probability in system state evaluation.

[0096] Exemplarily, in the continuous sampling execution process, the motor current signal is extracted for fast Fourier transform, and it is found that an abnormal harmonic peak value of 0.4A appears at 2.5Hz in the frequency component, while the normal baseline should be less than 0.05A. This phenomenon indicates that the power consumption of the pump group to overcome the resistance is periodically fluctuating. At the same time, the pressure fluctuation characteristic frequency is offset from 10Hz at the time of calibration to 12.5Hz, indicating that the equivalent capacitance of the pipeline is reduced. The environmental perception module monitors that the average value of the traction force parameter is 80N, and the variance reaches , indicating that the current is a high flow rate turbulent flow condition. A multi-dimensional state vector is constructed and input into the fuzzy logic reasoning rule, and the blockage membership function parameters are set. The high current harmonic combined with the pressure frequency offset and the high turbulence background calculates the filter screen blockage risk probability as 0.78, while the pipeline leakage probability is determined as 0.05 because the current is not abnormal. The output probability vector clearly quantifies that the system is currently in a high-risk state of filter screen semi-blockage, providing a quantitative basis for the subsequent reduction of sampling frequency or the triggering of protective flushing instructions.

[0097] Optionally, the change characteristics of the real-time monitoring data of the online water quality sensor and the system state evaluation result generate and execute an adaptive sampling strategy, which includes:

[0098] Calculate the change gradient of the real-time monitoring data over time or over depth;

[0099] When the change gradient exceeds the preset event triggering threshold, it is determined that a suspected water quality event occurs;

[0100] According to the spatial position or time attribute of the suspected water quality event, combined with the current system state evaluation result, an emergency sampling scheme is generated;

[0101] Replace the original preset sampling plan with the emergency sampling scheme and execute it.

[0102] Specifically, the real-time monitoring data of the online water quality sensor is continuously analyzed to calculate its change characteristics. For example, for a certain water quality parameter concentration, the system measures at continuous time points or depth points, and calculates its change gradient over time or change gradient over depth by the following formula:

[0103] ;

[0104] ;

[0105] where dC is the concentration change, dt and dh are the corresponding time interval and depth interval, respectively. When the calculated change gradient, for example, , exceeds a pre-set event trigger threshold, such as a conductivity change of more than 50 microsiemens per centimeter per meter depth in the vertical direction, a suspected water quality event is determined to have occurred at the current location. Once a suspected water quality event is determined to have occurred, the central controller jointly decides the spatial location or temporal attribute of the event with the system state assessment results, especially the quantified abnormal risk probability therein, to dynamically generate an optimal emergency sampling scheme . The decision logic can be represented by the following function:

[0106] ;

[0107] In this function, represents the type of suspected water quality event, such as whether it is a sharp stratification in space or a transient pollution in time. is the abnormal risk probability vector obtained from the previous composite diagnosis step, which contains the filter clogging risk probability and the pipe leakage risk probability. The function g represents the decision rule set of the system. For example, if the filter clogging risk probability in the system state assessment results is below a safety threshold and the event type is a spatial stratification, the system will generate a high-density profile sampling instruction, i.e., increase 3 to 5 sampling points in the range of 1 meter above and below the event occurrence depth, as the emergency sampling scheme. Conversely, if the filter clogging risk probability is high, to avoid equipment clogging due to dense sampling of high-concentration suspended solids, the system will generate an optimized scheme that increases the sampling frequency while appropriately reducing the single sampling volume and adding a preventive flushing instruction. This emergency sampling scheme, which takes into account the urgency of the event and the health status of the system itself, is sent to the sampling task scheduler in the form of a series of specific execution instructions, to be inserted or replaced into the original pre-set sampling plan in real time and executed immediately, thereby achieving accurate capture and reliable sampling of key environmental changes.

[0108] For example, during the vertical descent process, the conductivity C is monitored in real time, and in the depth interval of 5.0m to 5.2m, the concentration increases from 400uS / cm to 420uS / cm. The vertical gradient is calculated using the formula . Since the value exceeds the preset event trigger threshold 50 (uS / cm) / m, the system determines that a water quality stratification event occurs. At this time, the decision function J calls the current system state evaluation result, and the filter clogging risk probability is known to be 0.78, which has exceeded the safety threshold 0.7. According to the preset logic of the decision rule set g: when water stratification is detected and the equipment clogging risk is high, the regular high-intensity intensive sampling cannot be performed, otherwise it will lead to the complete failure of the filter. The generated The strategy is: reduce the originally planned 5 sampling points in the stratification area to 2 key feature points, and reduce the single sampling volume from 250 mL to 100 mL, and stipulate that after each sampling is completed, perform an additional pulse-based reverse self-cleaning immediately. This method ensures the acquisition of key environmental data while reducing the load and increasing the maintenance frequency, thereby maximizing the service life of the system under abnormal working conditions.

[0109] Optionally, the generating an emergency sampling scheme comprises:

[0110] When the filter clogging risk probability in the system state evaluation result is lower than the preset safety threshold, an emergency sampling scheme containing multi-layer intensive sampling is generated;

[0111] When the filter clogging risk probability is higher than the safety threshold, an optimization scheme of reducing the single sampling volume and increasing the preventive flushing instruction in the emergency sampling scheme is generated.

[0112] Specifically, after determining a suspected water quality event, the central controller immediately executes a conditional decision logic. First, the controller obtains the latest system state assessment result from the preceding composite diagnosis module, and extracts the key filter clogging risk probability therefrom. The central controller compares this filter clogging risk probability with a preset safety threshold. This safety threshold is an empirical value in engineering, for example, set to 0.7, representing that the system judges that the possibility of filter clogging in subsequent sampling under the current working condition exceeds 70%. Based on the comparison result, two different strategies will be executed to generate a path. The first path is that if the filter clogging risk probability is lower than the safety threshold, it indicates that the system is in good health and has the ability to perform high-intensity sampling tasks. At this time, the controller will generate an emergency sampling scheme containing multi-layer encryption sampling aiming at maximizing data acquisition. For example, taking the core depth of the event as the benchmark, within the range of 0.5 meters above and below it, an encryption sampling instruction sequence containing 5 new sampling points is generated at an interval of 0.2 meters, and the standard single sampling volume, for example, 250 milliliters, is maintained. The second path is that if the filter clogging risk probability is equal to or higher than the safety threshold, it indicates that the system is facing a higher risk of operation. At this time, the controller will generate an optimization scheme prioritizing system safety, which aims to obtain the core information of the event while avoiding equipment failure. The optimization scheme specifically includes two adjustments, one is to modify the sampling instruction, reducing the single sampling volume from the standard value to a smaller value, for example, 100 milliliters, to reduce the total amount of suspended solids passing through the filter in a single pass; the second is to insert a preventive flushing instruction after each sampling action in the sampling instruction sequence, that is, to trigger a pulse reverse air pressure flushing to actively clean the filter. Finally, regardless of which path the scheme is generated, it will be the final emergency sampling scheme and will be issued to the task execution module to replace or insert the original sampling plan.

[0113] For example, a suspected water quality event of conductivity mutation occurs at the current depth of 6.0 m, and an emergency scheme needs to be generated. The central controller extracts the filter clogging risk probability output by the composite diagnosis module. The global preset safety threshold is set to the default value of 0.70. At this time, if the measured filter clogging risk probability is 0.82, the second defensive path is triggered. The execution logic is as follows: first, the single sampling volume is reduced to 100 milliliters, and the preventive flushing instruction is inserted after each sampling action in the sampling instruction sequence. The proportional coefficient 0.4 is reduced based on the decay function of the risk value, and the calculation is ; second, in the task sequence, originally continuous 3 sampling action between the forced insertion of reverse air pressure flushing instruction, flushing pressure set to 0.2MPa, last 5s. Through this example derivation, in the face of high risk of blockage, actively reduce the total amount of solid particles through the filter screen of a single sample, and maintain the flow path unobstructed by using high-frequency self-cleaning. This decision-making method based on probability threshold enables the device to capture sudden pollution events such as sewage discharge while avoiding shutdown failure due to overload in extreme harsh conditions, reflecting the intelligent compromise of the sampling strategy.

[0114] Reference Figure 4 Based on the same inventive concept, the application also provides a depth and quantity intelligent water quality sampling control system, which comprises:

[0115] A multi-dimensional environmental perception module is configured to obtain environmental perception data including pressure data, temperature-salinity data and attitude data;

[0116] An active depth stabilization module is configured to predict the depth deviation trend of a preset sampling head based on the environmental perception data, and generate an active stabilization control instruction for driving a preset lifting mechanism to perform advanced depth compensation;

[0117] A sampling process perception module is configured to obtain dynamic response data of the sampling process after the sampling head is stabilized at the target depth, wherein the dynamic response data includes the operating parameters of the sampling pump group and the pressure fluctuation data of the pipeline;

[0118] An intelligent quantitative control module is configured to calibrate the single pulse effective displacement of the sampling pump group based on the dynamic response data to obtain a real-time displacement coefficient, and perform flow integration based on the real-time displacement coefficient until the preset sampling volume is reached;

[0119] An adaptive cleaning scheme module is configured to obtain water quality parameters of the next sampling point after completing sampling and query a cleaning knowledge base storing historical cleaning records to generate an adaptive cleaning scheme;

[0120] A system health diagnosis module is configured to diagnose based on the dynamic response data and the environmental perception data to obtain a system state evaluation result including an abnormal risk probability;

[0121] An intelligent strategy scheduling module is configured to generate and execute an adaptive sampling strategy based on the change characteristics of the real-time monitoring data of the online water quality sensor and the system state evaluation result.

[0122] It should be noted that the electrical connection between the above-mentioned units does not necessarily represent the direct connection of the circuit, and the indirect connection mode can also be applied to the embodiments of the present application as long as the purpose of the present application is achieved. The above-mentioned is only an exemplary embodiment of the present application, and cannot limit the scope of the present application.

[0123] That is, any equivalents of the above described subject matter, as well as other modifications and variations that can occur to those skilled in the art are intended to be covered. Other embodiments of the application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the application being indicated by the following claims.

Claims

1. A method for intelligent water quality sampling and control at a fixed depth and quantity, characterized in that, include: Acquire environmental perception data that includes pressure, temperature and salinity, and attitude data; The current theoretical depth of the sampling head is calculated based on the pressure data and the temperature and salinity data. The current tilt angle and depth compensation of the sampling head are calculated based on the attitude data. The current traction force parameters of the cable traction mechanism are obtained. Combined with the current tilt angle, the future depth offset and attitude change trend caused by the water flow are predicted. Based on the future depth offset and attitude change trend, an active stabilization control command containing feedforward compensation components is generated. After the sampling head stabilizes at the target depth, dynamic response data of the sampling process is acquired, including the operating parameters of the sampling pump group and the pressure fluctuation data of the pipeline. Based on the dynamic response data, the effective single-pulse displacement of the sampling pump group is calibrated online to obtain the real-time displacement coefficient, and the flow rate is integrated based on the real-time displacement coefficient until the preset sampling volume is reached. The process of obtaining the real-time displacement coefficient includes: controlling the sampling pump group to run with a set of test pulses, and simultaneously collecting pressure fluctuation waveforms and pump group motor current curves as calibration input data; performing feature analysis on the calibration input data to extract pressure build-up rate and fluctuation damping characteristics; and calculating the real-time displacement coefficient based on the pressure build-up rate and fluctuation damping characteristics. After sampling is completed, the water quality parameters of the next sampling point are obtained and the cleaning knowledge base storing historical cleaning records is queried to generate an adaptive cleaning plan. Diagnosis is performed based on the dynamic response data and the environmental perception data to obtain a system state assessment result containing abnormal risk probabilities. The abnormal risk probabilities include: extracting pump current spectrum characteristics and pressure fluctuation characteristic frequencies from the dynamic response data, extracting current velocity disturbance characteristics from the current water area from the environmental perception data, fusing and analyzing the pump current spectrum characteristics, pressure fluctuation characteristic frequencies, and flow velocity disturbance characteristics, assessing the filter clogging risk probability and pipeline leakage risk probability, and using the filter clogging risk probability and pipeline leakage risk probability as the abnormal risk probabilities in the system state assessment result. Calculate the gradient of real-time monitoring data over time or depth. When the gradient exceeds a preset event trigger threshold, a suspected water quality event is determined. Based on the spatial location or temporal attributes of the suspected water quality event and the current system status assessment results, an emergency sampling plan is generated. The emergency sampling plan replaces the original preset sampling plan and is executed.

2. The intelligent water quality sampling and control method with fixed depth and quantity according to claim 1, characterized in that, The acquisition of the current traction force parameters of the cable traction mechanism includes: By using a current sensor or torque sensor installed in the motor driver of the cable traction mechanism, traction parameters characterizing the current cable tension are obtained.

3. The intelligent water quality sampling and control method with fixed depth and quantity according to claim 1, characterized in that, The adaptive cleaning scheme includes: Obtain the real-time online water quality parameters of the next sampling point; In the cleaning knowledge base, query historical cleaning records that have similar pollution characteristics to the real-time online water quality parameters; Based on historical cleaning records, we predict the specific residual risks that may occur in the pipeline after this sampling. Based on specific residual risks, the cleaning stages, cleaning media, and intensity of cleaning actions are determined to form an adaptive cleaning solution.

4. The intelligent water quality sampling and control method with fixed depth and quantity according to claim 3, characterized in that, The cleaning knowledge base includes: Record the residual detection data in the pipeline after sampling under different combinations of water quality parameters in historical sampling tasks; Record the different cleaning schemes used for different combinations of water quality parameters and the final cleaning effect verification data; Establish a mapping relationship between water quality parameter combinations, cleaning schemes, and cleaning effect verification data, and continuously update the cleaning knowledge base.

5. The intelligent water quality sampling and control method with fixed depth and quantity according to claim 1, characterized in that, The emergency sampling scheme includes: When the probability of filter blockage risk in the system status assessment result is lower than a preset safety threshold, an emergency sampling scheme containing multi-layer encrypted sampling is generated. When the probability of filter clogging exceeds the safety threshold, an optimized scheme is generated that reduces the single sampling volume and increases preventative flushing instructions in the emergency sampling plan.

6. A fixed-depth, quantitative intelligent water quality sampling and control system, applied to the fixed-depth, quantitative intelligent water quality sampling and control method as described in any one of claims 1-5, characterized in that, The system includes: The multi-dimensional environmental perception module is used to acquire environmental perception data including pressure data, temperature and salinity data, and attitude data. An active depth stabilization module is used to predict the depth offset trend of a preset sampling head based on the environmental perception data, and generate active stabilization control commands to drive a preset lifting mechanism to perform advanced depth compensation. The sampling process sensing module is used to acquire dynamic response data of the sampling process after the sampling head stabilizes at the target depth. The dynamic response data includes the operating parameters of the sampling pump group and the pressure fluctuation data of the pipeline. The intelligent quantitative control module is used to calibrate the effective displacement of the single pulse of the sampling pump group online to obtain the real-time displacement coefficient based on the dynamic response data, and to perform flow integration based on the real-time displacement coefficient until the preset sampling volume is reached. The adaptive cleaning solution module is used to obtain the water quality parameters of the next sampling point after sampling is completed and to query the cleaning knowledge base that stores historical cleaning records to generate an adaptive cleaning solution. The system health diagnosis module is used to perform diagnosis based on the dynamic response data and the environmental perception data to obtain a system status assessment result that includes the probability of abnormal risks. The intelligent strategy scheduling module is used to generate and execute an adaptive sampling strategy based on the changing characteristics of real-time monitoring data from online water quality sensors and the system state assessment results.

Images