A method for intelligent monitoring and retesting of wastewater with standard detection limit awareness

Abstract

The application discloses a standard detection limit sensing intelligent wastewater monitoring and retesting method, and belongs to the technical field of environmental monitoring and intelligent wastewater discharge supervision. The method comprises the following steps: S1, collecting multi-source water quality index data; S2, converting each water quality index data and corresponding background data into unified water quality index level observation records; S3, constructing a detection limit sensing feature block; S4, constructing a monitoring sample level risk representation; S5, identifying a fuzzy risk state; S6, generating a closed-loop retesting task; S7, selecting a retesting platform; S8, executing retesting, and forming a closed-loop monitoring of a risk identification-hardware retesting-result feedback-state updating. Through the fuzzy risk state identification, the method outputs the membership degree of the monitoring sample to the boundary transition state and the suspected high-risk state, and potential risk trends are found before formal over-standard. In combination with a risk score, an uncertainty score and a state mutation amplitude, a retesting task is automatically triggered, passive response is changed into active early warning, and the timeliness of risk discovery is improved.

Description

Technical Field

[0001] This invention relates to the field of environmental monitoring and intelligent wastewater discharge supervision technology, specifically a method for intelligent monitoring and retesting of wastewater based on standard detection limit sensing. Background Technology

[0002] Wastewater discharge regulation is a core aspect of environmental protection, directly impacting aquatic ecological quality and public health. Currently, my country has established a regulatory system based on the "Integrated Wastewater Discharge Standard," the "Pollutant Discharge Standard for Urban Wastewater Treatment Plants," and various industry and local standards. Regulatory methods typically rely on manual sampling and testing, enterprise self-monitoring and recording, and fixed online monitoring equipment installed at key wastewater discharge units. Traditional regulatory methods often directly compare the detected pollutant concentration values ​​with the corresponding emission standard limits, resulting in a binary judgment of "compliant" or "non-compliant." While this method is legally necessary in formal environmental enforcement and compliance determination, it presents several technical problems in practice that are difficult to address through simple threshold judgments and fragmented monitoring. These problems are as follows: (1) The separation of fixed online monitoring and manual sampling Some key discharge outlets are equipped with fixed online monitoring terminals capable of collecting water quality indicators such as pH, water temperature, conductivity, turbidity, COD, ammonia nitrogen, total phosphorus, and flow rate. However, fixed online equipment typically only covers fixed locations, making it difficult to flexibly verify subsurface discharges in pipe networks, shoreline outlets, ditches, emergency bypasses, water surface diffusion zones, and hard-to-reach areas. While manual sampling is highly accurate, it is infrequent and slow to respond, and cannot automatically supplement sampling when the model detects a risk transition or increased uncertainty.

[0003] (2) The inspection by drones or robots lacks a data closed loop. While existing drones, unmanned surface vessels, ground robots, or pipeline robots can perform inspections, photography, location tracking, or sampling, most of these operations involve pre-set route inspections or sampling under manual command. Their sampling tasks are typically not triggered by a combination of water quality risk status, detection limit uncertainty, and standard applicability, nor do they automatically update the risk status by writing back the retest results to the same water quality index-level observation and recording system. Therefore, mobile hardware is often merely a data collection tool and does not form an active closed loop with the risk identification model.

[0004] (3) Ordinary threshold alarms cannot express the risk of over-intervention. Emission status doesn't always jump instantly from compliance to exceeding limits. In many cases, pollution load, process fluctuations, combined sewer overflows, sewer erosion, or abnormal emissions can place monitoring samples in a transitional state between stable compliance and significant risk. Traditional threshold alarms can only indicate whether standard limits have been exceeded, making it difficult to represent whether a monitoring sample simultaneously exhibits both stable compliance and risk characteristics, and also failing to guide which water quality indicator, location, and time window should be retested.

[0005] (4) The method detection limit led to the misuse of left censoring observations. Laboratory tests and some online analysis results often contain records below the method detection limit. These records are not equivalent to missing values, nor are they exact measurements; rather, they indicate that the true value is below a certain detection limit. Directly replacing these with 0, half-detection limit, or detection limit can easily alter the distance relationships between monitored samples, thus affecting the status identification results. Directly deleting them would result in the loss of monitoring information. Existing monitoring platforms typically only include the detection limit as a report note, without incorporating the intensity of the detection limit, censoring status, and the reliability of water quality indicators into the model calculations.

[0006] (5) Water quality indicators with no applicable limits were deleted or incorrectly determined. Under different industry standards, local standards, discharge destinations, and standard table numbers, not all monitored water quality indicators have applicable limits. For example, some water quality indicators cannot form a formal judgment of exceeding the standard under the current regulatory scenario, but may still reflect changes in the state of wastewater. Existing systems often directly delete such water quality indicators or forcibly include them in threshold judgments. The former causes information loss, and the latter leads to invalid or erroneous regulatory interpretations.

[0007] (6) The reliability of sensors and mobile platforms was not included in the model calculation. Online sensors are affected by calibration status, drift, biofouling, pipe blockage, self-cleaning effectiveness, and communication quality; sampling by drones, unmanned surface vessels, or robots is affected by positioning errors, attitude stability, sampling depth, weather, water flow, and endurance. Existing methods typically treat these data as ordinary numerical inputs, lacking a mechanism to incorporate device operating status and sampling platform confidence into the observation reliability.

[0008] Therefore, existing technologies have shortcomings and deficiencies, and need further improvement and development. Specifically, there is a need for an intelligent wastewater discharge monitoring technology solution that can simultaneously address multi-source hardware acquisition, detection censoring, standard applicability, equipment reliability, potential risk status identification, and proactive retesting and scheduling. Summary of the Invention

[0009] To address the problems existing in the prior art, the present invention provides a method for intelligent monitoring and retesting of wastewater with standard detection limit sensing.

[0010] To achieve the above objective, the method includes the following steps: S1. Obtain online water quality index data through fixed online monitoring terminals; monitor the equipment status and sampling location in real time through drones, inspection robots and unmanned surface vessels, and obtain background data for each online water quality index data in conjunction with monitoring reports.

[0011] S2. Convert each water quality index data and corresponding background data obtained in S1 into a unified water quality index-level observation record.

[0012] S3, based on the unified observation record of S2, expands each water quality indicator into a detection limit sensing feature block that includes the standard relative level, deviation from the standard, non-judgment background, censoring status, detection limit intensity, and observation reliability.

[0013] S4. Based on the standard detection limit sensing feature blocks of each water quality indicator generated in S3, the standard detection limit sensing feature blocks of all water quality indicators of the same monitoring sample are connected end to end in sequence to form a long vector representing the entire monitoring sample, which is combined to form a monitoring sample-level risk characterization.

[0014] S5. Based on the monitoring sample-level risk characterization generated in S4, fuzzy risk state identification is performed by calculating the weighted distance between the monitoring sample and each preset risk state and the membership vector matrix of the monitoring sample to each risk state, and the risk results are output.

[0015] S6. Calculate the risk status change magnitude, monitoring sample risk score, and uncertainty score of the same emission outlet or the same monitoring point between adjacent sampling times based on the risk status membership degree. When the risk status change magnitude, monitoring sample risk score, or uncertainty score of the same emission outlet or the same monitoring point between adjacent sampling times exceeds the corresponding trigger threshold, generate a closed-loop retest task.

[0016] S7. After generating the retest task, calculate the platform adaptation score for all candidate retest devices, and select the platform with the highest score to perform the retest.

[0017] S8. Perform retesting through the platform selected in S7. After the retesting results are returned, update the water quality index level observation records, observation reliability, status membership degree and output risk results of the corresponding monitoring samples, forming a closed-loop monitoring system of risk identification, hardware retesting, result return and status update.

[0018] Preferably, the fixed online monitoring terminal integrates a water quality sensor and an online analyzer.

[0019] The water quality data include data on pH, water temperature, conductivity, dissolved oxygen, turbidity, oxidation-reduction potential, COD, TOC, ammonia nitrogen, total nitrogen, total phosphorus, and suspended solids.

[0020] The background data includes method detection limits, censoring flags, whether it was actually observed, whether it has applicable standard limits, standard limit type, sensor or detection method status, hardware platform status, sampling location, and sampling time. Among them, whether it has applicable standard limits and standard limit type are determined based on the local standards and industry standards corresponding to the discharge destination to determine whether each water quality indicator has applicable limits, and the limit type is recorded. The limit type is upper limit, lower limit, and interval.

[0021] Preferably, the specific process of S2 is as follows: convert each water quality indicator collected in S1 into a unified water quality indicator-level observation record. The formula is: in, i For monitoring sample index; j An index for monitoring water quality indicators; This indicates the reported value or sensor measurement value; Indicates the detection limit of the method; The censoring flag is indicated by 1 = below the detection limit and 0 = not censored. This indicates whether the water quality indicator was actually observed; 1 = yes, 0 = no. This indicates whether the water quality indicator has an applicable standard limit under the current regulatory scenario, 1 = applicable standard limit, 0 = no applicable standard limit; Indicates the standard limit type; Indicates the status of the sensor or detection method; Indicates the hardware platform status; Indicates the sampling location; Indicates the sampling time; For reported values ​​or sensor measurements And its corresponding background data.

[0022] Preferably, the specific process of S3 is as follows: S3.1 Construction of Water Quality Index Substitute Values For each water quality index-level observation record, input values ​​for continuous calculation are constructed. Censorship is indicated by water quality indices below the detection limit not being discarded; instead, alternative water quality indices are used. For continuous calculations, the formula is: in, =0 indicates that the corresponding water quality indicator was observed without censoring. =1 indicates that the corresponding water quality indicator was observed as a left-censored observation below the method detection limit; η The substitution factor is the detection limit.

[0023] S3.2, Construction of Standard Relative Level and Deviation Characteristics set up Indicates the first The first monitoring sample The input values ​​of each water quality indicator after being processed to the detection limit; Indicators of water quality The standard upper limit, Indicators of water quality The lower limit of the standard Indicators of water quality The standard range To prevent division by zero of positive numbers.

[0024] For upper limit water quality indicators, the following must be met: The standard relative level characteristics and standard deviation characteristics are constructed as follows: in, Indicates the position of the observation relative to the upper limit of the standard; when When, it indicates that the water quality indicator has not exceeded the upper limit; when When this occurs, it indicates that the water quality indicator exceeds the upper limit; This indicates the relative degree of deviation from the standard upper limit.

[0025] For lower limit water quality indicators, the following must be met: The standard relative level characteristics and standard deviation characteristics are constructed as follows: in, Indicates the position of the observed value relative to the lower limit of the standard; when When, it indicates that the water quality index is not lower than the lower limit; when When the water quality index is below the lower limit, it indicates that the water quality index is below the lower limit. This indicates the relative degree of deviation from the lower limit of the standard.

[0026] For interval-type water quality indicators, the following must be met: The standard relative level characteristics and standard deviation characteristics are constructed as follows: in, This indicates the relative position of the observed value within the standard interval; when the observed value is below the lower limit of the interval, Set to 0; when the observed value is higher than the upper limit of the interval, Take 1; This indicates the degree to which the observed value deviates from the standard range. A deviation value will be generated if the value is below the lower limit or above the upper limit, and the deviation value will be 0 when the value is within the range.

[0027] According to water quality indicators The standard limit type determines the relative level characteristics of the final standard used. and standard deviation characteristics : Water quality indicators of different standard limit types have all been converted into a unified form. and ; This indicates the relative level of observed water quality indicators under the constraints of corresponding standards. This indicates the degree of deviation of the observed water quality index value from the corresponding standard constraint.

[0028] S3.3 Construction of Non-determined Background Features For water quality indicators without applicable standard limits, construct non-judgment background features. The formula is: in, Indicators of water quality j Historical substitutes The median or a surrogate value from similar emission outlets the median; It indicates a robust scale.

[0029] S3.4 Construction of censoring state and detection limit intensity characteristics The formulas for constructing censoring state features and detection limit intensity features are as follows: in, Indicate whether the observation is a censored observation; It indicates the intensity of the detection limit relative to the normal fluctuation scale of the corresponding water quality index.

[0030] S3.5 Calculation of Observation Reliability Reliability The calculation formula is: in, Indicates the standard applicability factor; Indicates the censored state factor; Indicates the detection limit intensity factor; This indicates the reliability factor of fixed online equipment or laboratory testing methods; This represents the platform confidence factor for drones, inspection robots, or unmanned surface vessels.

[0031] The formula for determining the value of is: In the formula, The retention coefficient for water quality indicators without standard values ​​ranges from 0 to 1.

[0032] The formula for determining the value of is: In the formula, The coefficients for censored observations are retained, and their values ​​range from 0 to 1.

[0033] The formula for determining the value of is: In the formula, To detect the intensity attenuation parameter.

[0034] The calculation formula is: in, This is the minimum reliability reservation value, and its value is greater than [value missing]. and less than ; to Assign weights to each device status deduction item, and satisfy the following: This is a deduction item for calibration validity, indicating whether the equipment or testing instrument has exceeded its calibration validity period. The calculation method is as follows: in, This indicates the time interval since the last calibration. Indicators of water quality The allowable calibration validity period for the corresponding equipment or testing method. To prevent division by zero of constants.

[0035] This is a drift deduction item, representing the degree of deviation of the equipment verification value or quality control value from the standard value. The calculation method is as follows: in, This indicates the equipment verification value, quality control sample test value, or standard solution test value; This indicates the corresponding standard or reference value; This indicates the allowable drift threshold.

[0036] This is a deduction item for self-cleaning status, indicating the self-cleaning or backwashing status of fixed online equipment. The calculation method is as follows: For laboratory testing methods that do not have self-cleaning or backwashing units, Pick .

[0037] This is a deduction item for packet loss during communication, representing the communication stability during data upload. The calculation method is as follows: in, This indicates the packet loss rate during data collection or uploading. This indicates the maximum allowable packet loss rate; for offline lab report entries with no communication upload anomalies, Pick .

[0038] This is a deduction item for the status of the flow tank or sampling pipeline, indicating whether the water sample flow path is stable during fixed online monitoring. The calculation method is as follows: For laboratory testing methods that do not involve flow cells or sampling tubing, Pick .

[0039] This is a deduction item for maintenance status, indicating whether the equipment has any unresolved faults or overdue maintenance. The calculation method is as follows: Platform confidence factor The calculation formula is: in, to The weight of the deduction item for the platform status is determined, and the following conditions are met: The calculation method for the positioning error deduction item is as follows: in, This indicates that the retesting platform is performing monitoring of samples. Positioning error when corresponding to a task; This indicates the allowable positioning error threshold.

[0040] This is a deduction for attitude or hovering stability, calculated as follows: in, This indicates hovering fluctuations of drones, attitude fluctuations during robot sampling, or instantaneous attitude fluctuations during sampling of unmanned surface vessels. This indicates the permissible stability threshold.

[0041] The deduction for route or path offset is calculated as follows: in, This indicates the maximum deviation of the drone's flight path, the unmanned surface vessel's path, or the robot's travel path from the preset route. This indicates the allowed path offset threshold.

[0042] The sampling depth deviation deduction item is calculated as follows: in, Indicates the actual sampling depth; Indicates the target sampling depth set for the task; This indicates the allowed sampling depth deviation threshold.

[0043] Deductions are made for weather or water flow conditions, calculated as follows: in, Indicates the wind speed during task execution; Indicates the water flow velocity in the sampled area; Indicates rainfall intensity; , , These represent the permissible wind speed threshold, water flow velocity threshold, and rainfall intensity threshold, respectively.

[0044] This is a deduction item for sample storage status, calculated as follows: When the observation data is obtained not from mobile retesting platforms such as drones, inspection robots, or unmanned surface vessels, but from fixed online monitoring terminals or laboratory test reports, the platform confidence factor... Pick When the mobile retesting platform participates in sampling, retesting, or inspection, the platform confidence factor is calculated according to the formula. .

[0045] S3.6 Finally, the detection limit sensing feature block for water quality index level standards is formed: Based on the standard detection limit sensing features constructed from S3.1-S3.5, the first... The first monitoring sample Each water quality indicator is used to construct a water quality indicator-level standard detection limit sensing feature block, and the formula is as follows: in, Indicates the first The first monitoring sample A standard detection limit sensing feature block for each water quality indicator; This indicates whether the water quality indicator has an applicable standard limit under the current regulatory scenario. This indicates that applicable standard limits apply. This indicates that there are no applicable standard limits.

[0046] when When enabled and While retaining the relative level and deviation information of the standard, and disabling non-judgment background features, the detection limit sensing feature block for water quality index level standards is represented as follows: when When the standard relative level and standard deviation information are turned off, and the non-judgment background features are enabled, the water quality index level standard detection limit sensing feature block is represented as follows: The constructed standard detection limit sensing feature block can simultaneously express standard constraints, non-judgment background, censoring state, detection limit intensity, and observation reliability, avoiding the conflation of upper limit type, lower limit type, interval type, and water quality indicators without applicable limits.

[0047] Preferably, S4 includes the following process: Based on the standard detection limit sensing feature blocks of various water quality indicators generated by S3, the preset set of monitored water quality indicators is set as follows: in, J This indicates the total number of water quality indicators preset by the system. Indicates the first One water quality indicator is monitored; for the first Each monitoring sample, according to The standard detection limit sensing feature blocks corresponding to each water quality indicator are arranged in a fixed order to form a risk characterization at the monitoring sample level: in, Indicates the first Risk characterization at the monitoring sample level for each monitoring sample; Indicates the first The first monitoring sample A standard detection limit sensing feature block for each water quality indicator.

[0048] For monitoring samples The water quality indicators actually observed in China are... And construct the corresponding according to S3. .

[0049] For monitoring samples Water quality indicators that were not actually observed in China led to... And set placeholder feature blocks : .

[0050] Preferably, the specific process of S5 is as follows: in, For the first The monitoring sample and the first The square of the weighted distance between the risk states; For risk status index; The total number of water quality indicators to be monitored; Indicates risk status Water quality indicators State-specific weights; Representing state Water quality indicators The central feature; Do not participate The calculation.

[0051] Let the total number of risk states be The fuzzy coefficient is ,and To avoid division by zero when the distance is 0, a distance correction constant is introduced. ,in The corrected distance is: The system calculates monitoring samples based on the corrected distance. Risk status Distance affinity: in, Indicates monitoring samples Risk status The degree of closeness; For the same monitoring sample In all Normalize the distance affinity at each risk state to obtain the membership degree: in, Indicates the first The monitoring sample belongs to the first The degree of risk status. Since the distance affinity of the same monitoring sample across all risk statuses has been normalized, it satisfies: If monitoring samples Distance to a certain risk state center A value of 0 indicates that the monitoring sample completely overlaps with the center of the risk state. The membership degree of the monitoring sample to the risk state is set to 1, and the membership degree to other risk states is set to 0. If there are multiple risk states with a distance of 0 at the same time, the membership degree is evenly distributed among these risk states.

[0052] The membership matrix is ​​formed by the following formula: in, Indicates the total number of monitored samples; K Represents the number of risk states, and the state membership matrix. The degree to which each monitored sample was assigned to each risk state was recorded.

[0053] After obtaining the state membership degree, the monitoring samples are determined. The dominant risk state is represented by the formula: in, Indicates monitoring samples The dominant risk status; the dominant risk status refers to the monitoring sample The state with the highest membership degree among all risk states.

[0054] Calculate the impact of each monitored water quality index on the monitored samples The contribution of risk identification results; for the first The first monitoring sample Each water quality indicator, and its contribution to water quality. The calculation formula is: in, Indicators of water quality For monitoring samples The contribution of risk identification results; Indicates the dominant risk status Water quality indicators State-specific weights; This is the standard deviation item; This is a non-judgment background anomaly. This indicates non-judgmental background anomalies for water quality indicators for which no applicable standard limits exist. This is a deleted status item; To detect the limit strength item; For unreliable observation terms, Indicates unreliable contribution from observations; to Calculate weights for contributions, satisfying the following conditions: when At that time, the corresponding water quality index contribution If the value is 0, it will not be included in the ranking of major contributing water quality indicators; according to Monitoring samples from largest to smallest The observed water quality indicators were ranked, and one or more water quality indicators with the highest contribution were selected as the main contributing water quality indicators.

[0055] The output risk identification results include: the state membership matrix of the monitoring samples, the dominant risk state, the main contributing water quality indicators, the contribution degree of the main contributing water quality indicators, and risk interpretation labels; among them, the risk interpretation labels are generated from the dominant risk state, the main contributing water quality indicators, and the main sources of contribution, respectively generating state category labels, main explanation objects, and cause labels, as follows: 1. Determine the status category label based on the dominant risk status: When the dominant risk states are stable compliance, boundary transition, suspected increased risk, high uncertainty, censorship sensitive, and no applicable limit abnormality, respectively, corresponding stable compliance, boundary transition, suspected increased risk, high uncertainty, censorship sensitive, and no applicable limit abnormality labels are generated.

[0056] 2. The water quality indicators with the highest contribution were identified as the main objects of interpretation.

[0057] 3. Determine the cause label based on the highest percentage contribution item in the water quality indicator contribution formula: If the contribution mainly comes from the standard deviation, the reason label is "standard deviation contribution"; If the contribution mainly comes from non-judgment background anomalies, the reason label is "background anomaly contribution"; If the contribution mainly comes from the censored state item or the detection limit intensity item, the reason label is "detection limit uncertain contribution". If the contribution mainly comes from unreliable observations, the reason is labeled as insufficient observation reliability.

[0058] Preferably, S6 includes the following process: The risk score of the monitoring sample is calculated based on the membership degree of the risk status, using the following formula: in, For the first i Risk score of each monitored sample The larger the value, the more likely the monitoring sample needs to be retested or resampled. This represents a set of high-risk states, including boundary transition states, states suspected of increasing risk, and states with high uncertainty. Indicates risk status k Risk level weighting.

[0059] Calculate the uncertainty score , It consists of five interpretable technical quantities, and the formula is as follows: in, to The weights are the values ​​of each uncertainty factor; each weight can satisfy... + + + + =1; The score is for uncertainty.

[0060] The components are defined as follows: (1) Entropy is the membership degree of a state, used to measure the membership degree of monitored samples. i The formula for calculating the degree of ambiguity in attribution among multiple risk states is as follows: (2) Indicates monitoring samples i The average observation reliability is calculated using the following formula: (3) Indicates monitoring samples i The proportion of censored observations: (4) Indicates monitoring samples i The proportion of water quality indicators for which there are no applicable standard limits: (5) This indicates the magnitude of change in risk status at the same emission outlet or monitoring point between adjacent sampling times: in, For monitoring samples i The risk state membership vector, This represents the risk status membership vector of the monitoring samples of the same monitoring object at the previous moment. A closed-loop retest task is generated when any of the following conditions are met: ; in, As the risk trigger threshold, For uncertainty trigger threshold, The threshold for triggering state changes is set. When any condition is met, a closed-loop retest task is generated.

[0061] Preferably, the specific process of S7 is as follows: After generating the retest task, the candidate hardware platforms are scored; the candidate platforms include fixed online monitoring terminals, automatic samplers, drones, inspection robots, and unmanned surface vessels; for the candidate platforms Platform compatibility score for: in, Indicates the reachability of the platform to the target location; This indicates the platform's positioning and sampling stability in the current environment; This indicates that the platform targets water quality indicators. j Detection or sampling adaptability; This indicates the platform's ability to form a chain of evidence for retesting; Indicates the cost, time, or energy consumption of the task; to Select weights for the platform.

[0062] Select Adaptation Score The highest platform Perform a retest: in, O For the set of all candidate hardware platforms; when When using fixed online monitoring terminals, increase the sampling frequency of corresponding water quality indicators and perform automatic calibration checks; when When it is an automatic sampler, it generates the sampling bottle number, storage temperature, and delivery task; when When it is a drone, generate flight path, hovering point, shooting angle, and sampling coordinates; when When generating inspection routes for robots, the system generates factory routes, obstacle avoidance paths, and nearby sampling points; when... When it is an unmanned surface vessel, it generates surface routes, cross-sectional sampling points, and sampling depths; the risk identification results are directly converted into executable hardware control tasks.

[0063] This invention provides a method for intelligent monitoring and retesting of wastewater with standard detection limit sensing, which has the following beneficial effects: (1) This invention identifies the membership degree of the monitoring sample to "boundary transition state" and "suspected high risk state" through fuzzy risk state identification, and can discover potential risk trends before they are officially exceeded. By combining risk score, uncertainty score and state change magnitude, it automatically triggers retesting tasks, changing "passive response" to "active early warning", and significantly improving the timeliness of risk discovery. (2) This invention uniformly constructs water quality index-level observation records, distinguishing between measured values, values ​​below the detection limit, unmeasured values, and water quality indicators without applicable limits, and constructs standard relative levels, standard deviations, non-judgment background features, censored states, and detection limit intensity features respectively. This design avoids the mixing of different semantic data, retains effective information on water quality indicators below the detection limit and without standard limits, and improves the accuracy of state identification. (3) Based on risk membership, uncertainty score, state change magnitude, platform accessibility, stability, adaptability, evidence chain capability, and task cost, this invention automatically calculates the platform adaptability score and selects the optimal platform for retesting. The system prioritizes the allocation of limited sampling resources to areas with high risk, insufficient information, or state changes, thereby achieving intelligent scheduling and efficient utilization of sampling resources.

[0064] (4) This invention incorporates equipment reliability factors and platform confidence factors into the calculation of observation reliability. The contribution of low-reliability data to the state identification distance is automatically reduced, thereby reducing false alarms caused by equipment malfunctions, environmental interference, or unstable single sampling. At the same time, the system outputs formal exceedance judgments and potential risk states separately. It does not replace the judgments of statutory standards, but provides interpretable information such as state evolution, main contributing water quality indicators, censoring ratio, and retesting recommendations, which facilitates understanding and decision-making by regulatory personnel. Attached Figure Description

[0065] Figure 1 This is a flowchart of the standard detection limit sensing wastewater intelligent monitoring and retesting method of the present invention.

[0066] Figure 2 This is a flowchart of the standard detection limit sensing feature block construction process of the present invention.

[0067] Figure 3This is a flowchart of the fuzzy risk state identification and closed-loop retest scheduling process of the present invention.

[0068] Figure 4 This is a schematic diagram of the hardware collaborative deployment of the present invention. Detailed Implementation

[0069] The present invention will be further described in detail below with reference to specific embodiments and accompanying drawings, but the scope of protection of the present invention is not limited to the content described.

[0070] This invention provides a method for intelligent monitoring and retesting of wastewater based on standard detection limit sensing. This method utilizes online water quality index data, laboratory test report data, automatic sampling data, and data transmitted from retesting platforms such as drones, inspection robots, and unmanned surface vessels generated during wastewater discharge monitoring. It unifies monitoring information from different sources, in different formats, and with varying degrees of reliability into water quality index-level observation records. Furthermore, it constructs a standard detection limit sensing feature block to comprehensively express information such as relative standard level, standard deviation, non-judgment background, censoring status, detection limit intensity, and observation reliability.

[0071] This invention does not simply use fixed online monitoring terminals, automatic samplers, drones, inspection robots, and unmanned surface vessels as independent data acquisition devices in parallel. Instead, it uses them as data sources or retest execution platforms in this method. Specifically, fixed online monitoring terminals are used to acquire online water quality index data at discharge outlets or monitoring points; automatic samplers are used to collect samples after generating closed-loop retest tasks; drones, inspection robots, and unmanned surface vessels are used to perform on-site inspections, location verification, image evidence collection, mobile detection, or supplementary sampling. All data collected or transmitted by the above devices are entered into the water quality index-level observation record according to the unified steps of this method, and further participate in risk status identification and retest task generation.

[0072] This invention expands the monitoring of each water quality indicator from ordinary concentration values ​​to water quality indicator-level observation records. For water quality indicators with applicable standard limits, standard relative level characteristics and standard deviation characteristics are constructed. For water quality indicators without applicable standard limits, non-judgment background characteristics are constructed; for observations below the method detection limit, censoring status and detection limit intensity are retained. Simultaneously, observation reliability is calculated by combining equipment status and platform status. Through the above processing, this invention avoids mixing data with different semantic meanings such as unmeasured, below detection limit, no applicable limit, and truly low risk.

[0073] This invention further identifies fuzzy risk states based on the risk characterization at the monitoring sample level, outputting the membership degree of the monitoring sample to stable compliant states, boundary transition states, suspected elevated risk states, high uncertainty states, censored sensitive states, and abnormal states without applicable limits. Based on the risk identification results, the system calculates the monitoring sample's risk score, uncertainty score, and state change magnitude. When the risk score, uncertainty score, or state change magnitude meets the triggering conditions, the system generates a closed-loop retesting task and selects a suitable retesting platform to execute the retesting based on platform accessibility, sampling stability, water quality index adaptability, evidence chain capability, and task cost. After the retesting results are transmitted back, the corresponding monitoring sample's water quality index-level observation records, observation reliability, state membership degree, and risk output results are updated, thus forming a closed-loop monitoring process of "risk identification—hardware retesting—result transmission—state update."

[0074] The flowchart of a standard detection limit sensing method for intelligent monitoring and retesting of wastewater according to the present invention is as follows: Figure 1 As shown, the specific steps include the following: S1, Multi-source data acquisition S1.1 Acquiring Water Quality Index Data: Online water quality index data is obtained by continuously collecting and analyzing data through fixed online monitoring terminals (integrating water quality sensors and online analyzers) deployed at the discharge outlet; the water quality index data includes data on pH, water temperature, conductivity, dissolved oxygen, turbidity, oxidation-reduction potential, COD, TOC, ammonia nitrogen, total nitrogen, total phosphorus, and suspended solids.

[0075] S1.2 Obtain background data corresponding to water quality indicators By using drones, inspection robots, and unmanned surface vessels to monitor equipment status and sampling locations in real time, and combining this with monitoring reports, background data for each online water quality indicator can be obtained.

[0076] Background data includes the method detection limit, censoring flag (1 = below detection limit, 0 = not censored), whether it was actually observed (1 = yes, 0 = no), whether there is an applicable standard limit (1 = applicable standard limit, 0 = inapplicable standard limit), standard limit type, sensor or detection method status (calibration, drift, maintenance, or communication status), hardware platform status (positioning error, attitude stability, sampling depth, weather, etc.), sampling location, and sampling time. The existence of applicable standard limits and the type of standard limit are determined based on the local and industry standards corresponding to the discharge destination to determine whether each water quality indicator has an applicable limit, and the limit type is recorded. The limit type is upper limit, lower limit, or range.

[0077] S2. Construction of Unified Observation Records Each water quality indicator collected by S1 was converted into a unified water quality indicator-level observation record. The formula is: in, i An index for monitoring samples (a complete set of multi-dimensional data records generated from a specific monitoring activity); j An index for monitoring water quality indicators; This indicates the reported value or sensor measurement value; Indicates the detection limit of the method; Indicates deletion flag; This indicates whether the water quality indicator was actually observed; This indicates whether the water quality indicator has applicable standard limits under the current regulatory scenario; Indicates the standard limit type; Indicates the status of the sensor or detection method; Indicates the hardware platform status; Indicates the sampling location; Indicates the sampling time; For reported values ​​or sensor measurements And its corresponding background data.

[0078] The purpose of this unified record is to integrate online monitoring from fixed online monitoring terminals, drone inspections, robot sampling, unmanned surface vessel monitoring, and laboratory reports into the same data structure, so that subsequent risk identification no longer relies solely on pollutant concentration values, but also utilizes detection limits, standard applicability, equipment status, and sampling platform status.

[0079] S3. Construction of Standard Detection Limit Sensing Features S3.1 Construction of Water Quality Index Substitute Values like Figure 2 As shown, for each water quality index level observation record, input values ​​for continuous calculation are constructed. The censoring flag indicates that water quality indices below the detection limit are not discarded, and substitute values ​​are used. For continuous calculations, the formula is: in, =0 indicates that the corresponding water quality indicator was observed without censoring. =1 indicates that the corresponding water quality indicator was observed as a left-censored observation below the method detection limit; η The substitution factor is the detection limit. Substitution value. It is only used for constructing continuous feature spaces, and uncertainties below the detection limit are still preserved through censored state features and detection limit intensity features.

[0080] S3.2, Construction of Standard Relative Level and Deviation Characteristics When a water quality indicator has applicable standard limits, the relative level characteristics of the standard are constructed according to the type of limit for that water quality indicator. and standard deviation characteristics .in, Used to indicate the first The first monitoring sample The position of the observed value of each water quality indicator relative to the applicable standard limit. This is used to indicate the degree of deviation of the observed value from the applicable standard limit. Through this processing, upper limit type, lower limit type, and interval type water quality indicators can all be converted into unified standard relative level characteristics and standard deviation characteristics, which can be used for the subsequent construction of standard detection limit sensing feature blocks.

[0081] set up Indicates the first The first monitoring sample The input values ​​of each water quality indicator after being processed to the detection limit; Indicators of water quality The standard upper limit, Indicators of water quality The lower limit of the standard Indicators of water quality The standard range To prevent division by zero of positive numbers.

[0082] For upper limit water quality indicators, the following must be met: The standard relative level characteristics and standard deviation characteristics are constructed as follows: in, Indicates the position of the observation relative to the upper limit of the standard; when When, it indicates that the water quality indicator has not exceeded the upper limit; when When this occurs, it indicates that the water quality indicator exceeds the upper limit. This indicates the relative degree of deviation from the standard upper limit.

[0083] For lower limit water quality indicators, the following must be met: The standard relative level characteristics and standard deviation characteristics are constructed as follows: in, Indicates the position of the observed value relative to the lower limit of the standard; when When, it indicates that the water quality index is not lower than the lower limit; when When the water quality index is below the lower limit, it indicates that the water quality index is below the lower limit. This indicates the relative degree of deviation from the lower limit of the standard.

[0084] For interval-type water quality indicators, the following must be met: The standard relative level characteristics and standard deviation characteristics are constructed as follows: in, This indicates the relative position of the observed value within the standard interval. When the observed value is below the lower limit of the interval... Set to 0; when the observed value is higher than the upper limit of the interval, Take 1; when the observed value is within the standard interval, This indicates its normalized position within the interval. This indicates the degree to which the observed value deviates from the standard range. A deviation value will be generated if the value is below the lower limit or above the upper limit, and the deviation value will be 0 when the value is within the range.

[0085] To facilitate subsequent unified calculations, the system is based on water quality indicators. The standard limit type determines the relative level characteristics of the final standard used. and standard deviation characteristics : Through the above processing, water quality indicators of different standard limit types are all converted into a unified form. and .in, This indicates the relative level of observed water quality indicators under the constraints of corresponding standards. This indicates the degree of deviation of observed water quality index values ​​from the corresponding standard constraints. When constructing subsequent water quality index-level standard detection limit sensing feature blocks, this method will be used uniformly. and As input features.

[0086] S3.3 Construction of Non-determined Background Features For water quality indicators without applicable standard limits, construct non-judgment background features. The formula is: in, Indicators of water quality j Historical substitutes The median or a surrogate value from similar emission outlets the median; This indicates a robust scale; non-judgment background features are only used for latent state identification and not for formal overshoot judgment.

[0087] S3.4 Construction of censoring state and detection limit intensity characteristics The formulas for constructing censoring state features and detection limit intensity features are as follows: in, Indicate whether the observation is a censored observation; This indicates the intensity of the detection limit relative to the normal fluctuation scale of the corresponding water quality index. The constructed censoring state features and detection limit intensity features can distinguish between "true low-risk observations" and "uncertain observations limited by higher detection limits".

[0088] S3.5 Calculation of Observation Reliability Reliability The calculation formula is: in, Indicates the standard applicability factor; Indicates the censored state factor; Indicates the detection limit intensity factor; This indicates the reliability factor of fixed online equipment or laboratory testing methods; This represents the platform confidence factor for drones, inspection robots, or unmanned surface vessels.

[0089] The formula for determining the value of is: In the formula, The retention coefficient for water quality indicators without standard values ​​is between 0 and 1. The formula for determining the value of is: In the formula, The coefficient for retaining censored observations, with values ​​between 0 and 1; The formula for determining the value of is: In the formula, This is the detection limit intensity attenuation parameter. Even when both are "below the detection limit," the higher the detection limit (relative to the normal fluctuation scale), the greater the uncertainty of the observation, and the lower the reliability should be. For example, if the detection limit is 5 times the normal fluctuation scale (…), then… =5), the detection limit is comparable to the normal scale ( =1) is less reliable, and the higher the detection limit, the lower the impact of the observation on risk status identification.

[0090] Reliability factor of fixed online equipment or laboratory testing methods Used to indicate the first The first monitoring sample The reliability of observed values ​​for a water quality indicator at the level of fixed online equipment or laboratory testing methods. The determination is jointly made based on calibration effectiveness, drift degree, self-cleaning status, communication packet loss, flow cell or sampling pipeline condition, and maintenance status. The calculation formula is: in, This is the minimum reliability reservation value, and its value is greater than [value missing]. and less than ; to Assign weights to each device status deduction item, and satisfy the following: .

[0091] This is a deduction item for calibration validity, used to indicate whether the equipment or testing instrument has exceeded its calibration validity period. It is calculated as follows: in, This indicates the time interval since the last calibration. Indicators of water quality The allowable calibration validity period for the corresponding equipment or testing method. To prevent division by zero, this constant is used. This item is not deducted while the equipment is within its calibration validity period; however, the longer the period exceeds the calibration validity period, the greater the reduction in reliability.

[0092] This is a drift deduction item, used to indicate the degree of deviation of equipment verification values ​​or quality control values ​​from standard values. The calculation method is as follows: in, This indicates the equipment verification value, the quality control sample test value, or the standard solution test value. This indicates the corresponding standard or reference value. This indicates the allowable drift threshold. The greater the deviation, the worse the stability of the device or method, and the greater the reduction in reliability.

[0093] This is a deduction item for self-cleaning status, used to indicate the self-cleaning or backwashing status of fixed online equipment. The calculation method is as follows: For laboratory testing methods that do not have self-cleaning or backwashing units, Pick .

[0094] This is a deduction item for packet loss during communication, used to represent the communication stability during data upload. The calculation method is as follows: in, This indicates the packet loss rate during data collection or uploading. This indicates the maximum allowable packet loss rate. A higher packet loss rate indicates poorer data integrity and a greater reduction in reliability for that observation record. For records entered into offline laboratory reports and for which there are no communication upload anomalies, Pick .

[0095] This is a deduction item for the status of the flow tank or sampling pipeline, used to indicate whether the water sample flow path is stable during fixed online monitoring. The calculation method is as follows: For laboratory testing methods that do not involve flow cells or sampling tubing, Pick .

[0096] This is a deduction item for maintenance status, used to indicate whether the equipment has any unresolved faults or overdue maintenance. The calculation method is as follows: By using the above methods, the reliability factor of fixed online equipment or laboratory testing methods is determined. It is limited to a calculable value. When the equipment calibration is effective, drift is minimal, self-cleaning is normal, communication is stable, the flow path is normal, and maintenance is in good condition, near When equipment calibration is overdue, drift is significant, communication packet loss is severe, the flow pool is abnormal, or the maintenance status is abnormal, This reduces the contribution of the observation to subsequent risk status identification.

[0097] Platform confidence factor This factor is used to represent the sampling reliability of mobile retesting platforms, including drones, inspection robots, or unmanned surface vessels, when performing inspection, retesting, or sampling tasks. This factor is determined by platform positioning error, attitude or hovering stability, flight path deviation, sampling depth deviation, weather or water flow conditions, and sample preservation status. The calculation formula is: in, to The weight of the deduction item for the platform status is determined, and the following conditions are met: The calculation method for the positioning error deduction item is as follows: in, This indicates that the retesting platform is performing monitoring of samples. Positioning error when corresponding to a task; This indicates the allowable positioning error threshold. The larger the positioning error, the lower the platform's confidence level.

[0098] This is a deduction for attitude or hovering stability, calculated as follows: in, This indicates hovering fluctuations of drones, attitude fluctuations during robot sampling, or instantaneous attitude fluctuations during sampling of unmanned surface vessels. This represents the permissible stability threshold. Greater fluctuations indicate a more unstable sampling process and lower platform confidence.

[0099] The deduction for route or path offset is calculated as follows: in, This indicates the maximum deviation of the drone's flight path, the unmanned surface vessel's path, or the robot's travel path from the preset route. This indicates the allowed path offset threshold. The larger the path offset, the more inconsistent the actual retest location is with the planned location, and the lower the platform's confidence level.

[0100] The sampling depth deviation deduction item is calculated as follows: in, Indicates the actual sampling depth. This indicates the target sampling depth set for the task. This indicates the allowable sampling depth deviation threshold. A larger sampling depth deviation indicates a worse consistency between the collected sample and the target water layer, resulting in a lower platform confidence level.

[0101] Deductions are made for weather or water flow conditions, calculated as follows: in, Indicates the wind speed during task execution. Indicates the water flow velocity in the sampled water area. Indicates rainfall intensity; , , These represent the permissible wind speed threshold, water flow velocity threshold, and rainfall intensity threshold, respectively. The larger the deduction item is when wind speed, water flow, or rainfall conditions are more unfavorable for stable sampling.

[0102] This is a deduction item for sample storage status, calculated as follows: When the observation data is obtained not from mobile platforms such as drones, inspection robots, or unmanned surface vessels, but from fixed online monitoring terminals or laboratory test reports, the platform confidence factor... Desirable When the mobile retesting platform participates in sampling, retesting, or inspection, the platform confidence factor is calculated according to the formula. .

[0103] Through the above methods, the platform confidence factor This reflects the reliability of the mobile retesting platform during specific task execution. More accurate positioning, more stable attitude, smaller flight path deviation, closer sampling depth to the target value, more suitable weather and water flow conditions, and more standardized sample preservation all contribute to this reliability. The closer ;on the contrary, This factor is reduced. It is included in the observation reliability calculation along with the reliability factor of fixed online equipment or laboratory testing methods, resulting in different calculation weights for the same water quality index value under different equipment conditions and sampling conditions.

[0104] The same pollutant concentration will not be treated equally by the system under different hardware conditions. For example, data from a properly calibrated fixed online terminal has high reliability; however, when a drone samples under strong wind conditions or a robot has a large positioning deviation, the reliability of the corresponding observation decreases, thereby reducing the interference of abnormal hardware conditions on risk assessment.

[0105] S3.6 Finally, the detection limit sensing feature block for water quality index level standards is formed: Based on the standard detection limit sensing features constructed from S3.1-S3.5, the first... The first monitoring sample Each water quality indicator is used to construct a water quality indicator-level standard detection limit sensing feature block, and the formula is as follows: in, Indicates the first The first monitoring sample A standard detection limit sensing feature block for each water quality indicator; This indicates whether the water quality indicator has an applicable standard limit under the current regulatory scenario. This indicates that applicable standard limits apply. This indicates that there are no applicable standard limits.

[0106] when When enabled and That is, retaining the relative level and deviation information of the standard while turning off non-judgment background features, the detection limit sensing feature block of the water quality index level standard is represented as follows: when At that time, no formal judgment of exceeding the standard is made, the information on the relative level of the standard and the deviation of the standard is turned off, and the non-judgment background feature is enabled. The water quality index level standard detection limit sensing feature block is represented as follows: Through the above construction, water quality indicators with applicable standard limits are identified as risk states through standard relative levels and standard deviation characteristics; water quality indicators without applicable standard limits are identified as risk states through non-judgment background characteristics, but do not participate in formal exceedance judgment. Therefore, the standard detection limit sensing feature block can simultaneously express standard constraints, non-judgment background, censored states, detection limit intensity, and observation reliability, avoiding the conflation of upper limit type, lower limit type, interval type, and water quality indicators without applicable limits.

[0107] S4. Generation of Risk Characterization at the Monitoring Sample Level Based on the standard detection limit sensing feature blocks of various water quality indicators generated by S3, a monitoring sample-level risk characterization is constructed according to the preset set of monitoring water quality indicators. Let the preset set of monitoring water quality indicators be: in, J This indicates the total number of water quality indicators preset by the system. Indicates the first One water quality indicator is monitored. For the first... Each monitoring sample, according to The standard detection limit sensing feature blocks corresponding to each water quality indicator are arranged in a fixed order to form a risk characterization at the monitoring sample level: in, Indicates the first Risk characterization at the monitoring sample level for each monitoring sample. Indicates the first The first monitoring sample A standard detection limit sensing feature block for each water quality indicator.

[0108] For monitoring samples The water quality indicators actually observed in China are... And construct the corresponding according to S3. .

[0109] For monitoring samples Water quality indicators that were not actually observed in China led to... And set placeholder feature blocks : This placeholder feature block is only used to maintain consistency in the representation dimensions of different monitoring samples and does not participate in the distance, weight, and contribution calculations in subsequent risk status identification. When calculating the distance between a monitoring sample and a risk status, observation markers are used. Unobserved water quality indicators are masked, that is, when At that time, the feature block corresponding to the water quality index is not included in the distance calculation.

[0110] Therefore, monitoring sample-level risk characterization It is not a simple numerical concatenation in the ordinary sense, but a block structure representation composed of multiple water quality index-level standard detection limit sensing feature blocks arranged in a fixed order of water quality indicators. This representation not only ensures that different monitoring samples have a unified dimension, but also retains the standard applicability, non-judgment background, censoring state, detection limit intensity, and observation reliability information of each water quality indicator, providing a unified input for subsequent fuzzy risk state identification and closed-loop retesting tasks.

[0111] S5. Fuzzy Risk Status Identification and Risk Result Output Based on the risk characterization of the monitoring samples obtained in S4, the membership degree of the monitoring samples to different risk states is calculated, and the risk identification results are output. The risk identification results output in S5 serve as the input for the closed-loop retest task in S6.

[0112] The system first combines all water quality index-level feature blocks in the same monitoring sample into a monitoring sample-level risk characterization. Risk status The weighted distance between them is calculated using the following formula: in, For the first The monitoring sample and the first The square of the weighted distance between the risk states; For risk status index; The total number of water quality indicators to be monitored; Indicates risk status Water quality indicators State-specific weights; Representing state Water quality indicators The central feature. The smaller this distance value, the more central the monitored sample. Risk status The closer.

[0113] According to monitoring samples Weighted distance between each risk state Calculate monitoring samples For the Membership degree of a risk state Let the total number of risk states be... The fuzzy coefficient is ,and To avoid division by zero when the distance is 0, a distance correction constant is introduced. ,in The corrected distance is: Monitoring samples were calculated based on the corrected distance. Risk status Distance affinity: in, Indicates monitoring samples Risk status The closer they are. The smaller the distance, the more likely they are to be found. The larger the value, the closer the monitored sample is to that risk state; the greater the distance, The smaller the value, the further away the monitored sample is from the risk state.

[0114] Subsequently, the same monitoring sample In all Normalize the distance affinity at each risk state to obtain the membership degree: Will After expansion, the membership degree can also be written as: in, Indicates the first The monitoring sample belongs to the first The degree of risk status, ranging from 0 to 1. Since the distance affinity of the same monitoring sample across all risk statuses has been normalized, it satisfies: In one specific implementation, the risk status includes a stable compliance status, a boundary transition status, a suspected increased risk status, a high uncertainty status, a censorship-sensitive status, and an abnormal status with no applicable limits. . Indicates monitoring samples For the first The degree of membership in a risk state does not indicate the probability of exceeding the standard in a legal sense.

[0115] In a special case, if the monitored sample Distance to a certain risk state center A value of 0 indicates that the monitoring sample completely overlaps with the center of the risk state. The membership degree of the monitoring sample to the risk state is set to 1, and the membership degree to other risk states is set to 0. If there are multiple risk states with a distance of 0 at the same time, the membership degree is evenly distributed among these risk states.

[0116] The membership matrix is ​​formed by the following formula: in, Indicates the total number of monitored samples. K Represents the number of risk states, and the state membership matrix. The degree to which each monitored sample was assigned to each risk state was recorded.

[0117] After obtaining the state membership degree, the monitoring samples are determined. The dominant risk state is represented by the formula: in, Indicates monitoring samples The dominant risk status. The dominant risk status refers to the dominant risk status of the monitored sample. The state with the highest membership degree among all risk states.

[0118] Calculate the impact of each monitored water quality index on the monitored samples The contribution of risk identification results. For the first The first monitoring sample Each water quality indicator, and its contribution to water quality. The calculation formula is: in, Indicators of water quality For monitoring samples The contribution of risk identification results; Indicates the dominant risk status Water quality indicators State-specific weights; This is the standard deviation item; This is a non-judgment background anomaly. This indicates non-judgmental background anomalies for water quality indicators for which no applicable standard limits exist. This is a deleted status item; To detect the limit strength item; For unreliable observation terms, Indicates unreliable contribution from observations; to Calculate weights for contributions, satisfying the following conditions: when When, it indicates water quality indicators In monitoring samples This water quality indicator was not actually observed, and its contribution was... If the value is 0, it will not be included in the ranking of major contributing water quality indicators.

[0119] according to Monitoring samples from largest to smallest The observed water quality indicators were ranked, and one or more indicators with the highest contribution were selected as the main contributing water quality indicators. The main contributing water quality indicators are used to explain the main reasons for the formation of the risk status of the monitored samples and serve as the basis for selecting water quality indicators for retesting in subsequent closed-loop retesting tasks.

[0120] Therefore, the risk identification results output by S5 include: the state membership vector of the monitored sample, the dominant risk state, the main contributing water quality indicators, the contribution degree of each main contributing water quality indicator, and risk interpretation labels. The risk interpretation labels are generated from the dominant risk state, the main contributing water quality indicators, and the main sources of contribution, respectively generating state class labels, main explanation objects, and cause labels, as follows: 1. Determine the status category label based on the dominant risk status: When the dominant risk status is stable and compliant, a "stable and compliant" label is generated. When the dominant risk state is the boundary transition state, a "boundary transition" label is generated; When the dominant risk status is "suspected elevated risk", a "suspected elevated risk" label is generated. When the dominant risk state is a high uncertainty state, a "high uncertainty" label is generated; When the dominant risk status is censorship-sensitive, a "censorship-sensitive" label is generated; When the dominant risk status is an anomaly with no applicable limits, a "no applicable limits anomaly" label is generated.

[0121] 2. Identify one or more water quality indicators with the highest contribution (with equal and highest contribution) as the main objects of interpretation.

[0122] 3. Determine the cause label based on the highest percentage contribution item in the water quality indicator contribution formula: If the contribution mainly comes from the standard deviation, the reason label is "standard deviation contribution"; If the contribution mainly comes from non-judgment background anomalies, the reason label is "background anomaly contribution"; If the contribution mainly comes from the censored state item or the detection limit strength item, the reason label is "detection limit uncertain contribution"; If the contribution mainly comes from unreliable observations, the reason label is "insufficient observation reliability".

[0123] The risk identification results output by the fuzzy risk status identification serve as input for the S6 closed-loop retest task. When a monitoring sample has a high degree of membership to a suspected elevated risk status, the system prioritizes retesting; when a monitoring sample has a high degree of membership to a high uncertainty status, a censoring-sensitive status, or an abnormal status without applicable limits, the system prioritizes high-reliability sampling, on-site inspection, or laboratory verification.

[0124] S6, Closed-loop retest task generation The system takes the risk identification results output by S5 as input, and combines the state membership degree of the monitoring sample, the dominant risk state, the main contributing water quality indicators, the contribution degree of water quality indicators, the risk interpretation label, and the state changes of the same monitoring object between adjacent sampling times to calculate the risk score, uncertainty score, and state change amplitude of the monitoring sample, and determines whether to generate a closed-loop retest task based on this. The system calculates the risk score of the monitoring sample based on the risk state membership degree, using the following formula: in, For the first i Risk score of each monitored sample The larger the value, the more likely the monitoring sample needs to be retested or resampled. This represents a set of high-risk states, including boundary transition states, states suspected of increasing risk, and states with high uncertainty. Indicates risk status k Risk level weighting.

[0125] Calculate the uncertainty score , It consists of five interpretable technical quantities, and the formula is as follows: in, to The weights for each uncertainty factor are set according to the regulatory scenario or equipment configuration; each weight can meet the requirements. + + + + =1. Scoring for uncertainty, The larger the value, the more unstable the monitoring conclusions of the sample, and the more necessary it is to confirm through retesting or resampling. The definitions of each component are as follows: (1) Entropy is the membership degree of a state, used to measure the membership degree of monitored samples. iThe formula for calculating the degree of ambiguity in attribution among multiple risk states is as follows: in, The larger the value, the closer the monitored sample is to the boundary area of ​​multiple risk states, and the less clear the state boundaries are.

[0126] (2) Indicates monitoring samples i The average observation reliability is calculated using the following formula: in, This indicates the total number of monitored water quality indicators; Indicators of water quality j Whether it was actually observed; This indicates the reliability of the observed value of the water quality index, then 1 This indicates the unreliability of the average observation. If the sensor is not calibrated, the UAV positioning error is large, or the robot's sampling attitude is unstable, then... Reduce, and thus make Increase.

[0127] (3) Indicates monitoring samples i The proportion of censored observations: in, =1 indicates a water quality indicator j The observed value is below the method detection limit, which indicates a censored observation; =0 indicates a non-censored observation. The larger the value, the more data in the monitoring sample are affected by the detection limit, and the higher the uncertainty of the monitoring results.

[0128] (4) Indicates monitoring samples i The proportion of water quality indicators for which there are no applicable standard limits: in, =1 indicates a water quality indicator j Under the current regulatory framework, applicable standard limits apply. =0 indicates that there is no applicable standard limit for this water quality indicator. The larger the value, the more water quality indicators in the monitoring sample that cannot be directly judged as exceeding the standard but still have background change information.

[0129] (5) This indicates the magnitude of change in risk status at the same emission outlet or monitoring point between adjacent sampling times: in, For monitoring samples i The risk state membership vector, This is the risk status membership vector of the monitoring samples of the same monitoring object at the previous moment. The larger the value, the more significant the change in the monitoring status within a short period of time, and the system should prioritize retesting or on-site inspection.

[0130] A closed-loop retest task is generated when any of the following conditions are met: in, As the risk trigger threshold, For uncertainty trigger threshold, The threshold for triggering state changes is set. When any one of the three conditions is met, the system generates a closed-loop retest task.

[0131] S7. Selection of Retesting Equipment After generating the retest task, the system scores the candidate hardware platforms. Candidate platforms include fixed online monitoring terminals, automatic samplers, drones, inspection robots, and unmanned surface vessels. For the candidate platforms... Platform compatibility score for: in, Indicates the reachability of the platform to the target location; This indicates the platform's positioning and sampling stability in the current environment; This indicates that the platform targets water quality indicators. j Detection or sampling adaptability; This indicates the platform's ability to form a chain of evidence for retesting; Indicates the cost, time, or energy consumption of the task; to Select weights for the platform.

[0132] Select Adaptation Score The highest platform Perform a retest: in, O For the set of all candidate hardware platforms; when When a fixed online monitoring terminal is used, the system increases the sampling frequency of the corresponding water quality indicators and performs automatic calibration checks; when When used as an automatic sampler, the system generates a sampling bottle number, storage temperature, and delivery task; when When it is a drone, the system generates flight path, hovering point, shooting angle, and sampling coordinates; when When inspecting robots, the system generates factory routes, obstacle avoidance paths, and nearby sampling points; when When operating as an unmanned surface vessel, the system generates surface routes, cross-sectional sampling points, and sampling depths. Thus, the risk identification results are directly translated into executable hardware control tasks.

[0133] S8, Perform Retest Drones, robots, unmanned surface vessels, fixed online terminals, or automated samplers perform retesting or resampling.

[0134] The flowchart for fuzzy risk state identification and closed-loop retest scheduling is as follows: Figure 3 As shown, after the retest results are transmitted back, the system updates the water quality index-level observation records, observation reliability, state membership, and risk output results for the corresponding monitoring samples. This forms a closed-loop monitoring process of "risk identification - hardware retest - result transmission - state update".

[0135] Based on the above system and method, the present invention is further described as follows: A schematic diagram of the hardware collaborative deployment method for implementing the present invention is shown below. Figure 4 As shown.

[0136] I. Deployment of Fixed Online Monitoring System for Enterprise Emission Outlets In one specific implementation, online water quality index data collected by fixed online monitoring terminals can be obtained from the enterprise's main discharge outlet, key workshop discharge outlet, or the outlet of the industrial park's wastewater treatment facility. When retesting is required, an automatic sampler can be invoked to collect samples. The fixed online monitoring terminal is used to acquire online water quality index data from one or more of the following: pH, water temperature, conductivity, dissolved oxygen, turbidity, redox potential, COD, TOC, ammonia nitrogen, total nitrogen, total phosphorus, and suspended solids. The automatic sampler is used to collect samples according to the sampling time, sampling location, and water quality indexes determined by the retesting task after a closed-loop retesting task is generated.

[0137] The multi-parameter water quality sensor can collect pH, water temperature, conductivity, dissolved oxygen, turbidity, and redox potential; the online analyzer can collect COD, TOC, ammonia nitrogen, total nitrogen, total phosphorus, or suspended solids; the automatic sampler can collect samples within a specified time window according to the model trigger command, and record the sampling bottle number, sampling volume, storage temperature, and delivery status.

[0138] Data collected or transmitted by fixed online monitoring terminals, automatic samplers, and retesting platforms can be uploaded to the data processing platform via wired or wireless communication, and serve as the data source for S1 multi-source data acquisition and S8 retesting result updates.

[0139] II. Collaborative Retesting of Unmanned Aerial Vehicles, Robots, and Unmanned Surface Vessels When the fuzzy risk status identification module detects a suspected increased risk or high uncertainty at a discharge outlet, the closed-loop sampling scheduling module generates a task. If the target is an open water surface, shoreline, or inaccessible discharge outlet, the system selects a drone to perform inspection, collecting visible light, multispectral, thermal infrared images, or short-term water quality probe data. If the target is located in a water surface diffusion area, the system selects an unmanned surface vessel to perform cross-sectional patrols and fixed-point sampling. If the target is located in a culvert, pipe network, narrow ditch, or complex area inside the plant, the system selects an inspection robot to perform close-range sampling or video verification.

[0140] Upon completion of the task, the execution platform transmits back location information, sampling time, equipment status, images, sensor data, and sample number. Based on the transmitted results, the system updates the observation reliability and the risk characterization at the sample level, and recalculates the state membership.

[0141] III. Construction of Water Quality Indicator-Level Characteristics Suppose a company's discharge outlet generates a monitoring sample on a certain day. This sample includes water quality indicators such as pH, COD, ammonia nitrogen, total phosphorus, suspended solids, and fecal coliforms. For COD, if the applicable standard has an upper limit... Then the system calculates the standard relative level and standard deviation For pH, if the standard is a range limit, the system calculates its position within the range and the deviation from the lower limit or the upper limit.

[0142] If a laboratory reports an observation as below the detection limit, the system does not treat this record as an exact value, but instead records the input surrogate value and the censoring status. and detection limit intensity If the same water quality indicator is collected by a drone water quality probe and the positioning error is large or the sampling attitude is unstable, the platform's confidence level will be... reduce.

[0143] IV. Risk Status Identification and Interpretation After the system is running, each monitoring sample outputs a membership vector. In this embodiment, the membership of a certain monitoring sample is 0.35 for stable compliance state, 0.45 for boundary transition state, 0.15 for suspected elevated risk state, and 0.05 for high uncertainty state. This result indicates that the monitoring sample is not clearly exceeding the standard, but is closer to the boundary transition state, and should be retested in conjunction with the main contributing water quality indicators.

[0144] The system further outputs the main contributing water quality indicators. For example, if the state weights show that COD, ammonia nitrogen, and total phosphorus contribute significantly to the state, the closed-loop sampling scheduling module will prioritize listing these water quality indicators as retest water quality indicators; if the risk mainly comes from abnormal water quality indicators with high censoring intensity or no applicable limit, the system will mark the task type as "verification retest" or "background anomaly investigation".

[0145] V. Closed-loop retest triggering Online monitoring data from a certain emission outlet showed that ammonia nitrogen was close to the limit, while COD was below the detection limit but the detection limit intensity was high. The system calculated that the membership degree of this monitoring sample to the boundary transition state was 0.62, and the membership degree to the high uncertainty state was 0.31. Because the risk score exceeded the threshold, the system generated a retest task.

[0146] If the discharge outlet is located in a ditch within the plant area, the system selects an inspection robot to approach and collect samples; if the discharge outlet is near an open area of ​​the receiving water body, the system selects an unmanned surface vessel for cross-sectional patrol; if abnormal shoreline image evidence is required, the system simultaneously dispatches a drone to take pictures; if the reliability of the fixed online terminal decreases, the system activates an automatic sampler to send samples to the laboratory for verification. After the retest results are returned, the system updates the risk status of the monitored sample and generates a regulatory alert.

[0147] This embodiment demonstrates that the risk status identification results in this application are not only used for display, but are directly converted into executable hardware control tasks.

[0148] Based on the above technical solution, the present invention will be further illustrated by the following embodiments: Example 1 This embodiment collects data from routine wastewater monitoring reports in southeastern Guizhou. The data includes 238 archived reports, data from 69 monitoring agencies or units, 122 discharge outlet-level time units, and 246 valid discharge outlet-date monitoring samples, spanning from January 4, 2023 to April 14, 2025. 25 valid water quality indicators were collected, resulting in 2984 water quality indicator-level observation records. Of these, 517 were left-censored observations (17.33%), and 15.92% had no applicable limits.

[0149] The testing method is as follows: First, the reports and monitoring records are converted into unified water quality index-level observation records; then, K-means, Gaussian mixture model, ordinary fuzzy C-means, two-class weighted fuzzy clustering method and the standard detection limit-aware fuzzy risk identification method of this application are compared; finally, water quality indicators such as Silhouette coefficient, Davies-Bouldin index, Xie-Beni index and fuzzy partitioning coefficient are used to evaluate the clustering quality, and the results are analyzed in conjunction with the state interpretation and retest task generation.

[0150] The above results show that, on the verification data, compared with the ordinary fuzzy C-mean, the method of this application improves the Silhouette coefficient from 0.1428 to 0.2002, an increase of approximately 40.2%; the Davies-Bouldin index decreases from 3.3903 to 2.3757, a decrease of approximately 29.9%; and the Xie-Beni index decreases from 6.4025 to 5.5028, a decrease of approximately 14.1%. This indicates that the method of this invention performs better than other comparative methods in the above indicators.

[0151] In the expression layer ablation test, the Silhouette coefficient of the original normalized input was 0.1112, and the Xie-Beni index was 9.9670; the Silhouette coefficient of the complete method was 0.2002, and the Xie-Beni index was 5.5028. Compared to the original input, the Silhouette coefficient improved by approximately 80.0%, and the Xie-Beni index decreased by approximately 44.8%. This indicates that the detection limit, censoring status, and standard applicability should not be considered merely as data notes, but should be involved in risk status identification.

[0152] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent monitoring and retesting of wastewater with standard detection limit sensing, characterized in that: Includes the following steps: S1. Obtain online water quality index data through fixed online monitoring terminals; monitor equipment status and sampling location in real time through drones, inspection robots and unmanned surface vessels, and obtain background data for each online water quality index data in conjunction with monitoring reports; S2. Convert each water quality index data and corresponding background data obtained in S1 into a unified water quality index-level observation record; S3. Based on the unified observation record of S2, each water quality indicator is expanded into a detection limit sensing feature block that includes the standard relative level, deviation from the standard, non-judgment background, censoring status, detection limit intensity and observation reliability. S4. Based on the standard detection limit sensing feature blocks of each water quality indicator generated in S3, the standard detection limit sensing feature blocks of all water quality indicators of the same monitoring sample are connected end to end in sequence to form a long vector representing the entire monitoring sample, which is combined to form a monitoring sample-level risk characterization. S5. Based on the monitoring sample-level risk characterization generated in S4, fuzzy risk state identification is performed and risk results are output by calculating the weighted distance between the monitoring sample and each preset risk state and the membership vector matrix of the monitoring sample to each risk state. S6. Calculate the risk status change range, monitoring sample risk score and uncertainty score of the same emission outlet or the same monitoring point between adjacent sampling times based on the risk status membership degree. When the risk status change range or monitoring sample risk score or uncertainty score of the same emission outlet or the same monitoring point between adjacent sampling times exceeds the corresponding trigger threshold, generate a closed-loop retest task. S7. After generating the retest task, calculate the platform adaptation score for all candidate retest devices, and select the platform with the highest score to perform the retest. S8. Perform retesting through the platform selected in S7. After the retesting results are returned, update the water quality index level observation records, observation reliability, status membership degree and output risk results of the corresponding monitoring samples, forming a closed-loop monitoring system of risk identification, hardware retesting, result return and status update.

2. The method for intelligent monitoring and retesting of wastewater with standard detection limit sensing according to claim 1, characterized in that: The fixed online monitoring terminal integrates a water quality sensor and an online analyzer; The water quality data include data on pH, water temperature, conductivity, dissolved oxygen, turbidity, oxidation-reduction potential, COD, TOC, ammonia nitrogen, total nitrogen, total phosphorus, and suspended solids. The background data includes method detection limits, censoring flags, whether it was actually observed, whether it has applicable standard limits, standard limit type, sensor or detection method status, hardware platform status, sampling location, and sampling time. Among them, whether it has applicable standard limits and standard limit type are determined based on the local standards and industry standards corresponding to the discharge destination to determine whether each water quality indicator has applicable limits, and the limit type is recorded. The limit type is upper limit, lower limit, and interval.

3. The method for intelligent monitoring and retesting of wastewater with standard detection limit sensing according to claim 2, characterized in that: The specific process of S2 is as follows: Each water quality indicator collected in S1 is converted into a unified water quality indicator-level observation record. The formula is: in, i For monitoring sample index; j An index for monitoring water quality indicators; This indicates the reported value or sensor measurement value; Indicates the detection limit of the method; The censoring flag is indicated by 1 = below the detection limit and 0 = not censored. This indicates whether the water quality indicator was actually observed; 1 = yes, 0 = no. This indicates whether the water quality indicator has an applicable standard limit under the current regulatory scenario, 1 = applicable standard limit, 0 = no applicable standard limit; Indicates the standard limit type; Indicates the status of the sensor or detection method; Indicates the hardware platform status; Indicates the sampling location; Indicates the sampling time; For reported values ​​or sensor measurements And its corresponding background data.

4. The method for intelligent monitoring and retesting of wastewater with standard detection limit sensing according to claim 3, characterized in that: The specific process of S3 is as follows: S3.1 Construction of Water Quality Index Substitute Values For each water quality index-level observation record, input values ​​for continuous calculation are constructed. Censorship is indicated by water quality indices below the detection limit not being discarded; instead, alternative water quality indices are used. For continuous calculations, the formula is: in, =0 indicates that the corresponding water quality indicator was observed without censoring. =1 indicates that the corresponding water quality indicator was observed as a left-censored observation below the method detection limit; η The substitution factor for the detection limit; S3.2 Construction of Standard Relative Level and Deviation Characteristics According to water quality indicators The standard limit type determines the standard relative level characteristics used. and standard deviation characteristics : In the formula, For upper limit type water quality indicators, the position of the observed value relative to the upper limit of the standard; For lower limit water quality indicators, the observed value is positioned relative to the lower limit of the standard. For interval-type water quality indicators, this refers to the relative position of the observed value within the standard interval; For upper limit type water quality indicators, this refers to the relative deviation of the observed value from the upper limit of the standard; For lower limit water quality indicators, this refers to the relative deviation of the observed value from the lower limit of the standard. For interval-type water quality indicators, the degree to which the observed value deviates from the standard interval is considered. A deviation value will occur if the value is below the lower limit or above the upper limit, and the deviation value will be 0 when the value is within the interval. S3.3 Construction of Non-determinative Background Features For water quality indicators without applicable standard limits, construct non-judgment background features. The formula is: in, Indicators of water quality j Historical substitutes The median or a surrogate value from similar emission outlets the median; Indicates a robust scale; S3.4 Construction of censoring state and detection limit intensity characteristics The formulas for constructing censoring state features and detection limit intensity features are as follows: in, Indicate whether the observation is a censored observation; This indicates the strength of the detection limit relative to the normal fluctuation scale of the corresponding water quality index; S3.5 Calculation of Observation Reliability Reliability The calculation formula is: in, Indicates the standard applicability factor; Indicates the censored state factor; Indicates the detection limit intensity factor; This indicates the reliability factor of fixed online equipment or laboratory testing methods; Platform confidence factor representing drones, inspection robots, or unmanned surface vessels; S3.6 Finally, the detection limit sensing feature block for water quality index level standards is formed: Based on the standard detection limit sensing features constructed from S3.1-S3.5, the first... The first monitoring sample Each water quality indicator is used to construct a water quality indicator-level standard detection limit sensing feature block, and the formula is as follows: in, Indicates the first The first monitoring sample A standard detection limit sensing feature block for each water quality indicator; This indicates whether the water quality indicator has an applicable standard limit under the current regulatory scenario. This indicates that applicable standard limits apply. This indicates that there are no applicable standard limits.

5. The method for intelligent monitoring and retesting of wastewater with standard detection limit sensing according to claim 4, characterized in that: S4 includes the following process: Based on the standard detection limit sensing feature blocks of various water quality indicators generated by S3, the preset set of monitored water quality indicators is set as follows: in, J This indicates the total number of water quality indicators preset by the system. Indicates the first One water quality indicator is monitored; for the first Each monitoring sample, according to The standard detection limit sensing feature blocks corresponding to each water quality indicator are arranged in a fixed order to form a risk characterization at the monitoring sample level: in, Indicates the first Risk characterization at the monitoring sample level for each monitoring sample; Indicates the first The first monitoring sample A standard detection limit sensing feature block for each water quality indicator.

6. The method for intelligent monitoring and retesting of wastewater with standard detection limit sensing according to claim 5, characterized in that: The specific process of S5 is as follows: in, For the first The monitoring sample and the first The square of the weighted distance between the risk states; For risk status index; The total number of water quality indicators to be monitored; Indicates risk status Water quality indicators State-specific weights; Representing state Water quality indicators The central feature; Let the total number of risk states be The fuzzy coefficient is ,and To avoid division by zero when the distance is 0, a distance correction constant is introduced. ,in The corrected weighted distance is: For the same monitoring sample In all Normalize the distance affinity at each risk state to obtain the membership degree: in, Indicates the first The monitoring sample belongs to the first The degree of risk status; Indicates monitoring samples Risk status The degree of closeness; The membership matrix is ​​formed by the following formula: in, Indicates the total number of monitored samples; K Represents the number of risk states, and the state membership matrix. The degree to which each monitored sample was assigned to each risk status was recorded; After obtaining the state membership degree, the monitoring samples are determined. The dominant risk state is represented by the formula: in, Indicates monitoring samples The dominant risk status; the dominant risk status refers to the monitoring sample The state with the highest membership degree among all risk states; Calculate the impact of each monitored water quality index on the monitored samples The contribution of risk identification results; for the first The first monitoring sample Each water quality indicator, and its contribution to water quality. The calculation formula is: in, Indicators of water quality For monitoring samples The contribution of risk identification results; Indicates the dominant risk status Water quality indicators State-specific weights; This is the standard deviation item; This is a non-judgment background anomaly. This indicates non-judgment background anomalies for water quality indicators for which there are no applicable standard limits. This is a deleted status item; To detect the limit strength item; For unreliable observation terms, Indicates unreliable contribution from observations; to Calculate weights for contributions, satisfying the following conditions: when At that time, the corresponding water quality index contribution If the value is 0, it will not be included in the ranking of major contributing water quality indicators; according to Monitoring samples from largest to smallest The observed water quality indicators were ranked, and one or more water quality indicators with the highest contribution were selected as the main contributing water quality indicators. The output risk identification results include: the state membership matrix of the monitoring sample, the dominant risk state, the main contributing water quality indicators, the contribution degree of the main contributing water quality indicators, and the risk interpretation label; among them, the risk interpretation label is generated by the dominant risk state, the main contributing water quality indicators, and the main contribution sources, respectively generating state category label, main interpretation object, and cause label.

7. The method for intelligent monitoring and retesting of wastewater with standard detection limit sensing according to claim 6, characterized in that: S6 includes the following process: The risk score of the monitoring sample is calculated based on the membership degree of the risk status, using the following formula: in, For the first i Risk score of each monitored sample The larger the value, the more likely the monitoring sample needs to be retested or resampled. This represents a set of high-risk states, including boundary transition states, states suspected of increasing risk, and states with high uncertainty. Indicates risk status k Risk level weighting; Calculate the uncertainty score , It consists of five interpretable technical quantities, and the formula is as follows: in, to The weights are the weights of each uncertainty factor; each weight can satisfy... + + + + =1; Score for uncertainty; Entropy is the membership degree of a state. Indicates monitoring samples i The average observation reliability; Indicates monitoring samples i The proportion of censored observations; Indicates monitoring samples i The proportion of water quality indicators for which there are no applicable standard limits; The formula for representing the change in risk status of the same emission outlet or monitoring point between adjacent sampling times is as follows: in, For monitoring samples i The risk state membership vector, This represents the risk status membership vector of the monitoring samples of the same monitoring object at the previous moment. A closed-loop retest task is generated when any of the following conditions are met: ； in, As the risk trigger threshold, For uncertainty trigger threshold, The threshold for triggering state changes is set. When any condition is met, a closed-loop retest task is generated.

8. The method for intelligent monitoring and retesting of wastewater with standard detection limit sensing according to claim 7, characterized in that: The specific process of S7 is as follows: After generating the retest task, the candidate hardware platforms are scored; the candidate platforms include fixed online monitoring terminals, automatic samplers, drones, inspection robots, and unmanned surface vessels; for the candidate platforms Platform compatibility score for: in, Indicates the reachability of the platform to the target location; This indicates the platform's positioning and sampling stability in the current environment; This indicates that the platform targets water quality indicators. j Detection or sampling adaptability; This indicates the platform's ability to form a chain of evidence for retesting; Indicates the cost, time, or energy consumption of the task; to Select weights for the platform; Select Adaptation Score The highest platform Perform a retest: in, O For the set of all candidate hardware platforms; when When using fixed online monitoring terminals, increase the sampling frequency of corresponding water quality indicators and perform automatic calibration checks; when When it is an automatic sampler, it generates the sampling bottle number, storage temperature, and delivery task; when When it is a drone, generate flight path, hovering point, shooting angle, and sampling coordinates; when When generating inspection routes for robots, the system generates factory routes, obstacle avoidance paths, and nearby sampling points; when... When it is an unmanned surface vessel, it generates surface routes, cross-sectional sampling points, and sampling depths; the risk identification results are directly converted into executable hardware control tasks.

Images