A method and system for real-time monitoring of groundwater pollution concentration

Abstract

The application relates to the technical field of concentration monitoring, in particular to a groundwater pollution concentration real-time monitoring method and system. The method comprises the following steps: acquiring water level and concentration sequences; performing baseline decomposition on the two sequences to obtain a short-time fluctuation sequence, calculating a threshold value through a standard deviation, and obtaining a water level fluctuation point set and a concentration fluctuation point set; determining a correlation weight based on the time difference between the water level and the concentration, and matching based on the correlation weight; calculating a short-time fluctuation response coefficient based on the corresponding amplitude and the correlation weight after successful matching; determining a water dynamic correlation component estimation value by correcting the water level and the concentration based on the short-time fluctuation response coefficient; determining a drift increment based on the regression slope after baseline difference fitting and an analysis window, and updating a zero-point drift value; after temperature compensation of real-time concentration, using the updated zero-point drift value to perform addition and calibration to obtain concentration calibration data, and completing real-time monitoring. The application improves the reliability of monitoring data.

Description

Technical Field

[0001] This application relates to the field of concentration monitoring technology, specifically to a method and system for real-time monitoring of groundwater pollution concentration. Background Technology

[0002] Groundwater environment monitoring systems widely employ long-term deployed submersible multi-parameter sensors to acquire key hydrogeological indicators such as conductivity, dissolved oxygen, and water level in real time. In actual operation and maintenance, electrochemical water quality sensors (especially conductivity probes) experience slow, monotonous zero-point drift over time due to the accumulation of biofilms or polarization effects on the electrode surface. This systematic error manifests as a continuous increase in concentration readings over time, easily confused with the actual concentration increase caused by the natural diffusion of solutes in groundwater. Existing monitoring data processing methods typically rely on periodic on-site manual calibration or simple thresholding. The former is costly to maintain, while the latter struggles to distinguish between slowly accumulating drift errors and real environmental changes, often leading to false alarms or data distortion.

[0003] In hydrogeology, the static water level in underground monitoring wells is not absolutely constant, but rather exhibits minute, instantaneous fluctuations influenced by changes in atmospheric pressure, solid tides, or surrounding pumping activities. According to solute transport theory, the concentration of dissolved substances in an aquifer is driven by changes in water level (e.g., water displacement or pressure changes), theoretically resulting in a corresponding dilution or concentration response. However, existing data processing methods often analyze water level and water quality data in isolation, failing to utilize the aforementioned correlation as a verification basis. More importantly, due to the complexity of the underground medium, there is an uncertain time lag in the transmission of water level fluctuations to concentration changes, and significant environmental noise interference makes it difficult to determine the correlation between the two through simple linear correlation analysis. Therefore, how to pinpoint the causal relationship between water level and concentration from noisy monitoring data without increasing hardware costs, and thereby distinguish between sensor drift and actual pollution, is a key technical problem that needs to be solved to improve the confidence of monitoring data. Summary of the Invention

[0004] To address the technical problem of inaccurate concentration monitoring, this application provides a method and system for real-time monitoring of groundwater pollution concentration, the specific technical solution of which is as follows: Firstly, this application proposes a method for real-time monitoring of groundwater pollution concentration, which includes the following steps: A preset analysis window is used to collect temperature and concentration data. Temperature compensation is applied to the concentration data based on temperature to obtain a concentration sequence. Water level data is also collected to form a water level sequence. Low-pass decomposition is performed on the water level sequence and concentration sequence to obtain the water level baseline and concentration baseline. The difference between the baseline and the corresponding water level and concentration is then calculated to obtain the corresponding short-term fluctuation sequence. The extraction thresholds for water level and concentration are calculated based on the standard deviation of the corresponding short-term fluctuation sequence and compared with the water level and concentration to obtain the water level fluctuation point set and concentration fluctuation point set. The correlation weight is calculated based on the time difference between water level and concentration and the preset conduction delay time; the correlation weight is used to match the water level fluctuation point set and the concentration fluctuation point set to obtain the linkage matching pair set; the hydraulic quiescent period is determined based on the logarithm of the successful matching, and the short-time fluctuation response coefficient is calculated based on the amplitude and correlation weight corresponding to the linkage matching pair set. Based on the water level and baseline changes at each moment within the analysis window, the short-time fluctuation response coefficient is used for correction, and then the concentration at the baseline point is combined to determine the estimated value of the hydrodynamic correlation component; the regression slope after subtracting the estimated value of the hydrodynamic correlation component and the concentration baseline and the analysis window are used to determine the drift increment, and the preset zero-point drift value is updated. After real-time concentration temperature compensation, the updated zero-point drift value is used for calibration to obtain concentration calibration data, thus completing real-time monitoring.

[0005] In the aforementioned scheme, this application introduces sparse feature extraction and bipartite graph matching algorithms. Under conditions of environmental noise and uncertainties in transmission hysteresis, this invention can accurately identify causal water level-concentration linkage pairs from discrete fluctuation events. Compared to traditional cross-correlation analysis, this method effectively avoids local noise interference by utilizing the principle of global optimality, ensuring that the calculated short-time fluctuation response coefficients accurately reflect the current monitoring environment characteristics, unaffected by the sensor's own zero-point state, thus providing an objective data reference for drift identification. Secondly, the differential analysis mechanism based on signal component reconstruction established in this application utilizes the characteristic differences (linearity and rate) between sensor drift and actual pollution in the time domain evolution to achieve accurate classification of the two. This not only enables the system to automatically deduct accumulated drift errors without human intervention, ensuring the long-term continuity of monitoring data, but also effectively prevents the risk of over-calibration or under-reporting due to misjudgment, significantly reducing operation and maintenance costs and improving the reliability of monitoring data.

[0006] In one embodiment, the concentration sequence is obtained by temperature compensation based on temperature: , This represents the original conductivity sequence after time synchronization. This represents the original water temperature sequence after time synchronization. Indicates the temperature compensation coefficient. This represents the conductivity sequence after temperature compensation; the conductivity sequence is the same as the concentration sequence.

[0007] In one embodiment, the short-term fluctuation sequence is the corresponding sequence minus the corresponding baseline.

[0008] In one embodiment, the method for calculating the extraction thresholds for water level and concentration based on the standard deviation of the corresponding short-term fluctuation sequence, and comparing them with the water level and concentration to obtain the water level fluctuation point set and the concentration fluctuation point set is as follows: The extreme points in the short-term water level fluctuation sequence that are greater than the water level extraction threshold are used to form a water level fluctuation point set; The extreme points in the short-term concentration fluctuation sequence that are greater than the concentration extraction threshold are used to form a concentration fluctuation point set; The expression for extracting the threshold is: , , This represents the preset minimum noise floor. This represents the standard deviation of a short-term water level fluctuation series. The standard deviation of a short-term concentration fluctuation series is represented by... Represents the confidence coefficient. This indicates the water level extraction threshold. This indicates the concentration extraction threshold.

[0009] In one embodiment, the method for calculating the association weight based on the time difference between water level and concentration and a preset conduction delay time is as follows: For water level and concentration points, determine their sign. If the positive and negative directions are opposite, the association weight is 0; if the positive and negative directions are the same, the expression for the association weight is: , This represents the time difference between the i-th water point and the j-th concentration point. Indicates the conduction delay time. Indicates the width of the time window. This represents an exponential function with the natural constant as its base. This represents the association weight between the elements corresponding to the i-th water point and the j-th concentration point.

[0010] In one embodiment, the method for determining the hydraulic quiescent period based on the logarithm of successful matches and calculating the short-term fluctuation response coefficient based on the amplitude and correlation weights corresponding to the linked matching pair set is as follows: If the number of successfully matched pairs is less than the preset minimum sample threshold, then during the hydraulic quiescent period, the short-time fluctuation response coefficient is 0. Conversely, if it is not in the hydraulic quiescent period, the expression for the short-time fluctuation response coefficient is: , This represents the association weight of the kth successfully matched pair. This represents the water level fluctuation amplitude of the kth successfully matched logarithm. This represents the concentration fluctuation amplitude of the kth successfully matched log. This represents the short-time fluctuation response coefficient.

[0011] In one embodiment, the method for determining the estimated hydrodynamic components based on the water level and baseline changes at each moment within the analysis window, corrected using the short-time fluctuation response coefficient, and then combined with the concentration at the baseline point, is as follows: The starting time within the analysis window is recorded as the reference point, and the difference between the water level at each time point and the water level at the reference point is recorded as the water level change. The product of the water level change and the short-time fluctuation response coefficient is recorded as the first product. The first product is added to the concentration at the reference point to obtain the estimated value of the hydrodynamic correlation component.

[0012] In one embodiment, the method for determining the drift increment by using the regression slope and analysis window after subtracting the estimated hydrodynamic components from the concentration baseline, and updating the preset zero-point drift value, is as follows: The drift increment is the product of the regression slope and the length of the analysis window; The sum of the drift increment, corrected by the damping coefficient, and the preset zero-point drift value is used as the updated zero-point drift value.

[0013] In one embodiment, the concentration calibration data is the difference between the real-time concentration under standard conditions and the updated zero-point drift value.

[0014] On the other hand, this application also provides a real-time groundwater pollution concentration monitoring system, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any of the above-described real-time groundwater pollution concentration monitoring methods.

[0015] The beneficial effects of this application are as follows: This application, by introducing sparse feature extraction and bipartite graph matching algorithms, enables the precise identification of causal water level-concentration linkage pairs from discrete fluctuation events, even under conditions of environmental noise and transmission hysteresis uncertainty. Compared to traditional cross-correlation analysis, this method effectively avoids local noise interference by utilizing the principle of global optimality, ensuring that the calculated short-time fluctuation response coefficients accurately reflect the current monitoring environment characteristics, unaffected by the sensor's own zero-point state, thus providing an objective data reference for drift identification. Secondly, the differential analysis mechanism based on signal component reconstruction established in this application utilizes the characteristic differences (linearity and rate) between sensor drift and actual pollution in the time domain evolution to achieve accurate classification of the two. This not only enables the system to automatically deduct accumulated drift errors without human intervention, ensuring the long-term continuity of monitoring data, but also effectively prevents the risk of over-calibration or under-reporting due to misjudgment, significantly reducing operation and maintenance costs and improving the reliability of monitoring data. Attached Figure Description

[0016] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of a method for real-time monitoring of groundwater pollution concentration provided in one embodiment of this application. Detailed Implementation

[0018] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a real-time monitoring method and system for groundwater pollution concentration proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0020] An embodiment of a method and system for real-time monitoring of groundwater pollution concentration: The following description, in conjunction with the accompanying drawings, details the specific scheme of the real-time monitoring method and system for groundwater pollution concentration provided in this application.

[0021] Please see Figure 1 The illustration shows a flowchart of a method and system for real-time monitoring of groundwater pollution concentration according to an embodiment of this application. The method includes the following steps: Step S001: Set the analysis window, collect temperature and concentration, perform temperature compensation on the concentration based on temperature to obtain the concentration sequence, and collect water level to form a water level sequence.

[0022] In the first analysis cycle after the monitoring system is first started or reset (i.e., the cold start phase), since historical operating parameters are not yet stored in memory, the system first performs parameter initialization operations.

[0023] The system sets the initial value of the global variable: the transmission delay time is determined based on the medium type of the aquifer where the monitoring well is located. The time is set to a preset empirical constant; in this embodiment, the gravel layer is set to 0 minutes and the clay layer to 30 minutes; the current accumulated zero-point drift value is set to... Set to 0.

[0024] To ensure the stability of subsequent algorithms, the system sets a warm-up period, which includes... There are one analysis window; in this embodiment, three windows are set. During the warm-up period, the system only performs data acquisition, feature extraction, and trial calculation and update of response coefficients, and does not perform drift detection and data subtraction operations until... Once the parameters have converged and stabilized, the system will automatically enter the formal monitoring mode.

[0025] Since submersible monitoring equipment typically contains multiple independent sensing units; in this embodiment, this includes a water level gauge, a conductivity probe, and a temperature probe. The sampling frequencies of each unit may have slight deviations. To ensure that the variables in subsequent analysis have a strict correspondence at the same time, the system first constructs a data structure of length [length missing]. A sliding data analysis window is required, which needs to cover at least one complete solid tide cycle and perform synchronization operations on the data within the window; in this embodiment, It lasts for 24 hours.

[0026] Specifically, the system acquires the original static water level sequence, original water temperature sequence, and original conductivity sequence through a sensing unit. Using a fixed sampling interval as a reference, linear interpolation is employed to resample these three sequences onto a unified time axis t. In this embodiment, the sampling interval is 1 minute. Subsequently, to eliminate non-concentration interference from water temperature fluctuations on the conductivity readings, the system utilizes the synchronized water temperature sequence... For conductivity sequence Perform standard temperature compensation.

[0027] Preferably, in this embodiment, the specific expression for temperature compensation is: , This represents the original conductivity sequence after time synchronization. This represents the original water temperature sequence after time synchronization. Indicates the temperature compensation coefficient. This represents the conductivity sequence after temperature compensation. In this embodiment, the temperature compensation coefficient is set to 0.02.

[0028] Using the above formula, when the water temperature is above 25℃, the denominator is greater than 1, and the original conductivity reading is corrected downwards, thus restoring the conductivity value that is determined only by the solute concentration.

[0029] Since conductivity is determined solely by concentration, the compensated conductivity sequence is denoted as the concentration sequence; for ease of reference, the original static water level sequence after time synchronization is denoted as the water level sequence, and the original water temperature sequence after time synchronization is denoted as the water temperature sequence.

[0030] Thus, the water level sequence and concentration sequence were obtained.

[0031] Step S002: Decompose the two sequences to obtain the baseline, then subtract the sequences to obtain the short-term fluctuation sequence, calculate the extraction threshold through its standard deviation, and obtain the water level fluctuation point set and the concentration fluctuation point set.

[0032] In groundwater monitoring scenarios, the zero-point drift of sensors typically manifests as a low-frequency component that changes slowly over time, while the response caused by hydrodynamic forces (such as solid tides and pressure effects) manifests as a high-frequency fluctuation component that changes rapidly. To separate these two types of characteristics, the system uses low-pass filtering technology to perform orthogonal decomposition on the sequence.

[0033] Specifically, the system sets the cutoff frequency of the low-pass filter to the length of the analysis window. The reciprocal of the water level sequence. and concentration sequence The input to this filter produces a smooth low-frequency curve, which is defined as the water level baseline. and concentration baseline These two baseline sequences preserve the possible drift trends of the sensors and the seasonal water level changes of the aquifer. The obtained baselines are normalized using a maximum-minimum normalization algorithm, mapping their values ​​to the range of 0-1.

[0034] Furthermore, by subtracting the corresponding long-term evolution baseline from the corresponding sequence, the high-frequency residual signal with zero mean is extracted to obtain the corresponding short-term fluctuation sequences, which are denoted as the water level short-term fluctuation sequence and the concentration short-term fluctuation sequence, respectively. The short-term sequences eliminate long-term baseline shifts and only contain fluctuation characteristics caused by instantaneous hydrodynamic effects.

[0035] In order to accurately capture statistically significant hydrodynamic events from continuous fluctuation sequences and transform them into discrete point sets suitable for subsequent graph theory algorithms, the system adopts an adaptive threshold strategy for feature extraction.

[0036] First, calculate the standard deviation of the short-term water level fluctuation sequence within the current window. and the standard deviation of the short-term concentration fluctuation series To prevent background white noise from being misidentified as significant fluctuations during the hydrodynamically stable period (when the signal is extremely stable and the variance is very small), the system presets a minimum noise floor. In this embodiment, the value is taken as 3 times the sensor resolution. The water level extraction threshold is calculated separately. Extraction threshold of concentration .

[0037] The expressions for the extraction thresholds of water level and concentration are: , , This represents the preset minimum noise floor. This represents the standard deviation of a short-term water level fluctuation series. The standard deviation of a short-term concentration fluctuation series is represented by... Represents the confidence coefficient. This indicates the water level extraction threshold. This represents the concentration extraction threshold. In this embodiment, the confidence coefficient is set to 3.

[0038] The above expression selects the larger of the statistical threshold and the noise basis as the final threshold, ensuring the robustness of the extraction. Subsequently, the short-term fluctuation sequence is traversed to identify all local extrema where the absolute amplitude exceeds the corresponding threshold, including peaks and troughs. These extracted extrema form a fluctuation point set, resulting in the water level fluctuation point set for the short-term water level fluctuation sequence. For short-time concentration fluctuation sequences, a set of concentration fluctuation points is obtained. Each element in the set contains the time and amplitude of the fluctuation point. These two sets of points will serve as input for the next stage of the graph matching algorithm.

[0039] Thus, the water level fluctuation point set and the concentration fluctuation point set were obtained.

[0040] Step S003: Determine the correlation weight based on the time difference between water level and concentration, and use it for matching; after successful matching, calculate the short-time fluctuation response coefficient based on the corresponding amplitude and correlation weight.

[0041] Through the above steps, the water level fluctuation point set and the concentration fluctuation point set were obtained. The core purpose of this step is to use graph theory algorithms to lock the unique causal correspondence between water level and concentration fluctuations under the interference of environmental noise and time lag, and to calculate the short-time fluctuation response coefficient that is not affected by sensor drift, as a physical benchmark for subsequent drift determination.

[0042] To quantify the likelihood of a causal relationship between any water level fluctuation and any concentration fluctuation, a weighted correlation matrix is ​​first constructed. This matrix combines proximity in the time dimension with polarity consistency in the amplitude dimension, providing a quantitative basis for subsequent graph matching.

[0043] Specifically, the system reads the propagation delay time updated in the previous analysis cycle. (If it is the first cycle, use the default values ​​initialized in step S001). Assume the water level fluctuation point set contains... The concentration fluctuation points are concentrated at several water points. Each concentration point. The system constructs a dimension... The correlation matrix is ​​such that for any element in the matrix, it represents the correlation weight between the water point and the concentration point in the corresponding row and column.

[0044] First, a judgment is made based on the polarity of the fluctuations in water level and concentration for each element. If the polarities are consistent, the correlation weight is calculated based on the difference between the time difference and the conduction delay time between the two points. .

[0045] Specifically, check whether the polarities of the fluctuations of the two are consistent. According to hydrodynamic principles, dilution or concentration effects should maintain a consistent direction. If the amplitudes of the water point and concentration point have opposite signs (i.e., one is a positive fluctuation and the other is a negative fluctuation), then it is determined that the two cannot be linked, and the correlation weight is directly set. .

[0046] If the polarities of the two points are the same, calculate the time difference between the water point and the concentration point, and use a Gaussian kernel function to measure this time difference relative to the conduction delay time. The degree of matching is used as the association weight, where the time window of the Gaussian kernel function is 10 in this embodiment.

[0047] Preferably, in this embodiment, the expression for the association weight is: , This represents the time difference between the i-th water point and the j-th concentration point. Indicates the conduction delay time. Indicates the width of the time window. This represents an exponential function with the natural constant as its base. This represents the association weight between the elements corresponding to the i-th water point and the j-th concentration point.

[0048] When the time difference between two points The closer to the conduction delay time When the exponent term is closer to 0, the calculated weight is higher. The closer the value is to 1, the higher the probability that the two points are linked; conversely, if the time difference deviates significantly from the empirical value, the weight will rapidly decay and approach 0. The final matrix fully describes the association confidence of all potential point pairs within the current window.

[0049] Due to environmental noise, relying solely on local threshold judgments or nearest-neighbor matching can easily lead to multiple solutions (e.g., a water level point may be temporally adjacent to multiple concentration points). To eliminate this ambiguity, this embodiment employs the bipartite graph maximum weight matching algorithm (Kuhn-Munkres algorithm, or KM algorithm for short) to find the globally optimal correspondence. Its inputs are the set of water level fluctuation points and the set of concentration fluctuation points, along with the association weights of water level points and concentration points in the association matrix. The output is a set of non-conflicting matching pair indices, denoted as the linked matching pair set, where K pairs are successfully matched.

[0050] To prevent parameter errors caused by forced calculations during hydrodynamically stable periods (i.e., without significant fluctuations) or when data quality is extremely poor, the system executes a quiescent period detection logic at this step.

[0051] If the number of successfully matched pairs K is less than the preset minimum sample threshold, the current window is determined to be in a hydraulic quiescent period. In this embodiment, the minimum sample threshold is 3 pairs.

[0052] During the hydraulic quiescent period, the system considers the current data insufficient to support parameter updates, and therefore executes a protection strategy; the short-term fluctuation response coefficient is 0. If there is no fluctuation response coefficient in the previous analysis period, the marking period is invalid and no drift calibration is performed; the conduction delay time remains unchanged.

[0053] If it is not during the hydraulic quiescent period, then it is an effective linkage period. Calculate the average time difference of all matching pairs in the linkage matching pair set, and use this average to adjust the memory... Updates are performed to achieve adaptive tracking of aquifer conductivity characteristics.

[0054] After confirming the effective linkage period, the system uses the selected high-confidence matching data to calculate the short-term fluctuation response coefficient. This coefficient characterizes the statistical proportion of concentration change caused by a unit change in water level under the current geological environment.

[0055] Specifically, the calculation is performed using the weighted least squares method. For each pair in the linkage matching set, its correlation weight, water level fluctuation amplitude, and concentration fluctuation amplitude are extracted for calculation. The formula for calculating the short-time fluctuation response coefficient is as follows: , This represents the association weight of the kth successfully matched pair. This represents the water level fluctuation amplitude of the kth successfully matched logarithm. This represents the concentration fluctuation amplitude of the kth successfully matched log. This represents the short-time fluctuation response coefficient.

[0056] In the above formula, the numerator is the weighted covariance, and the denominator is the weighted variance. The introduction of correlation weights ensures that point pairs with high confidence (i.e., precise time matching) dominate the calculation, further reducing the impact of random errors. It is worth noting that to prevent calculation anomalies, the system checks the value of the denominator before performing division. If the denominator is less than the preset minimum calculation precision value... (For example If the current condition is not met, the calculation is considered invalid, and the system automatically switches to the hydraulic quiescent period protection logic described above, stopping the current update. It is worth noting that the short-time fluctuation response coefficient during the hydraulic quiescent period and when the condition is invalid is recorded as 0.

[0057] Thus, the short-time fluctuation response coefficient was obtained.

[0058] Step S004: After correcting the water level with the short-time fluctuation response coefficient and then correcting the concentration, determine the estimated value of the hydrodynamic related components; determine the drift increment based on the regression slope and analysis window after fitting the baseline difference, and update the zero-point drift value.

[0059] Using the two point sets obtained from the above steps and the short-time fluctuation response coefficient as input, based on the signal superposition model assumption, the signal component containing only hydrodynamic influence is reconstructed, and the cumulative drift error of the sensor is separated through difference analysis.

[0060] Within the preset analysis window, it is assumed that the effects of water displacement or pressure changes caused by water level variations on concentration statistically satisfy a linear superposition relationship. That is, the observed concentration signal can be regarded as a linear superposition of sensor drift components, water level linkage components, and external pollution components, and the response coefficient of the water level linkage component can be identified by the parameter identification through the short-term fluctuation characteristics of the previous step.

[0061] To assess the current operating status of the sensor, the system first reconstructs the concentration change trajectory based on the above assumptions, assuming that it is only affected by hydrodynamic forces.

[0062] Specifically, taking the start time of the analysis window as the reference point, for any time within the analysis window, the system first calculates the change in water level baseline relative to the start time, which is recorded as the water level change. The calculation method is to set the difference between the water level at that time and the water level at the reference point.

[0063] Then, the water level change is projected onto the concentration domain using the short-time fluctuation response coefficient, and the estimated values ​​of the hydrodynamic correlation components are calculated. The specific expression is: , Indicates the concentration at the baseline point. Represents the short-time fluctuation response coefficient. This represents the water level change at time c in the analysis window. This represents the estimated value of the hydrodynamic correlation components at time c in the analysis window. Indicates the reference point.

[0064] The estimated values ​​of hydrodynamic correlation components represent the changes in background concentration caused by changes in water level, inferred from current hydrogeological characteristics. This is relevant when the system is in a quiescent or invalid state. That is, it is assumed that there are no observable hydrodynamic effects during this period.

[0065] Furthermore, the measured concentration baseline is subtracted from the estimated hydrodynamic components, and the differences at all times are recorded as a baseline deviation sequence. This eliminates natural background changes caused by water level fluctuations and retains only anomalous components caused by sensor drift, external contamination injection, or equipment failure.

[0066] Subsequently, linear least squares regression analysis was performed on the baseline deviation sequence to obtain the goodness of fit and regression slope. A better linear fit closer to 1 indicates that the deviation sequence changes more linearly, consistent with the monotonic cumulative characteristics caused by sensor electrode aging or slow biofilm growth. The regression slope characterizes the rate of deviation increase over time. A value greater than 0 indicates a continuous increase in the measured value relative to the estimated value (positive drift); conversely, a value less than 0 indicates a negative drift.

[0067] When the sensor is in an effective linkage period, and the linear fit goodness of fit is greater than a preset linear threshold, and the absolute value of the regression slope is less than a preset maximum drift rate threshold, the current deviation is considered sensor drift; otherwise, the anomaly is considered a non-drift event. In this embodiment, the linear threshold is 0.85, and the maximum drift rate threshold is the maximum aging rate set by the sensor hardware manual.

[0068] It is worth noting that being within the linkage validity period means that the current hydrodynamic model is reliable and the baseline reconstruction is effective; the goodness of linear fit is high because it takes into account the difference that drift usually shows a linear gradual change, while pollution usually shows a nonlinear abrupt change or fluctuation.

[0069] After determining sensor drift through the above steps, a global calibration parameter update operation is performed. To avoid system output oscillations caused by single calculation errors, the system employs a damped smoothing strategy to update the current cumulative zero-point drift value in memory. The product of the regression slope and the length of the analysis window is used as the drift increment, and the zero-point drift value is updated based on this drift increment. The specific expression is: , This represents the current cumulative zero-point drift value. This represents the drift increment of the analysis window. Indicates the damping coefficient. This represents the new zero-point drift value. In this embodiment, the damping coefficient is 0.2. This coefficient limits the magnitude of a single correction, ensuring a smooth change in the calibration curve and preventing data jumps caused by occasional noise.

[0070] This completes the update of the zero-point drift value.

[0071] Step S005: After real-time concentration temperature compensation, use the updated zero-point drift value to perform a calibration to obtain concentration calibration data and complete real-time monitoring.

[0072] The system collects concentration and water temperature data in real time, and uses temperature compensation to correct the concentration under standard conditions. The difference between the real-time concentration under standard conditions and the updated zero-point drift value is then used to obtain the concentration calibration data. This data eliminates baseline offsets identified by the system as being caused by non-environmental factors (i.e., sensor aging, scaling), thus restoring the true water concentration value.

[0073] The acquired concentration calibration data is sent to the monitoring center database via a remote telemetry unit (RTU) for users to query or trigger subsequent pollution warnings.

[0074] This completes the real-time monitoring of pollution concentration.

[0075] Based on the same inventive concept as the above method, this embodiment of the invention also provides a real-time monitoring system for groundwater pollution concentration, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-described methods for real-time monitoring of groundwater pollution concentration.

[0076] It should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

[0077] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for real-time monitoring of groundwater pollution concentration, characterized in that, The method includes the following steps: A preset analysis window is used to collect temperature and concentration data. Temperature compensation is applied to the concentration data based on temperature to obtain a concentration sequence. Water level data is also collected to form a water level sequence. Low-pass decomposition is performed on the water level sequence and concentration sequence to obtain the water level baseline and concentration baseline. The difference between the baseline and the corresponding water level and concentration is then calculated to obtain the corresponding short-term fluctuation sequence. The extraction thresholds for water level and concentration are calculated based on the standard deviation of the corresponding short-term fluctuation sequence and compared with the water level and concentration to obtain the water level fluctuation point set and concentration fluctuation point set. The correlation weight is calculated based on the time difference between water level and concentration and the preset conduction delay time; the correlation weight is used to match the water level fluctuation point set and the concentration fluctuation point set to obtain the linkage matching pair set; the hydraulic quiescent period is determined based on the logarithm of the successful matching, and the short-time fluctuation response coefficient is calculated based on the amplitude and correlation weight corresponding to the linkage matching pair set. Based on the water level and baseline changes at each moment within the analysis window, the short-time fluctuation response coefficient is used for correction, and then the concentration at the baseline point is combined to determine the estimated value of the hydrodynamic correlation component; the regression slope after subtracting the estimated value of the hydrodynamic correlation component and the concentration baseline and the analysis window are used to determine the drift increment, and the preset zero-point drift value is updated. After real-time concentration temperature compensation, the updated zero-point drift value is used for calibration to obtain concentration calibration data, thus completing real-time monitoring.

2. The method for real-time monitoring of groundwater pollution concentration as described in claim 1, characterized in that, The concentration sequence is obtained by temperature compensation based on temperature: , This represents the original conductivity sequence after time synchronization. This represents the original water temperature sequence after time synchronization. Indicates the temperature compensation coefficient. This represents the conductivity sequence after temperature compensation; the conductivity sequence is the same as the concentration sequence.

3. The method for real-time monitoring of groundwater pollution concentration as described in claim 1, characterized in that, The short-term fluctuation sequence is the corresponding sequence minus the corresponding baseline.

4. The method for real-time monitoring of groundwater pollution concentration as described in claim 1, characterized in that, The method for calculating the extraction thresholds for water level and concentration based on the standard deviation of the corresponding short-time fluctuation sequence, and comparing them with the water level and concentration to obtain the water level fluctuation point set and concentration fluctuation point set is as follows: The extreme points in the short-term water level fluctuation sequence whose absolute values ​​are greater than the water level extraction threshold are used to form a water level fluctuation point set; The extreme points in the short-term concentration fluctuation sequence whose absolute values ​​are greater than the concentration extraction threshold are used to form a concentration fluctuation point set; The expression for extracting the threshold is: , , This represents the preset minimum noise floor. This represents the standard deviation of a short-term water level fluctuation series. The standard deviation of a short-term concentration fluctuation series is represented by... Represents the confidence coefficient. This indicates the water level extraction threshold. This indicates the concentration extraction threshold.

5. The method for real-time monitoring of groundwater pollution concentration as described in claim 1, characterized in that, The method for calculating the correlation weight based on the time difference between water level and concentration and the preset conduction delay time is as follows: For water level and concentration points, determine their sign. If the positive and negative directions are opposite, the association weight is 0; if the positive and negative directions are the same, the expression for the association weight is: , This represents the time difference between the i-th water point and the j-th concentration point. Indicates the conduction delay time. Indicates the width of the time window. This represents an exponential function with the natural constant as its base. This represents the association weight between the elements corresponding to the i-th water point and the j-th concentration point.

6. The method for real-time monitoring of groundwater pollution concentration as described in claim 1, characterized in that, The method for determining the hydraulic quiescent period based on the logarithm of successful matches, and calculating the short-time fluctuation response coefficient based on the amplitude and correlation weights corresponding to the linked matching pairs, is as follows: If the number of successfully matched pairs is less than the preset minimum sample threshold, then during the hydraulic quiescent period, the short-time fluctuation response coefficient is 0. Conversely, if it is not in the hydraulic quiescent period, the expression for the short-time fluctuation response coefficient is: , This represents the association weight of the kth successfully matched pair. This represents the water level fluctuation amplitude of the kth successfully matched logarithm. This represents the concentration fluctuation amplitude of the kth successfully matched log. This represents the short-time fluctuation response coefficient.

7. The method for real-time monitoring of groundwater pollution concentration as described in claim 1, characterized in that, The method for determining the estimated hydrodynamic components based on the water level and baseline changes at each moment within the analysis window, corrected using the short-time fluctuation response coefficient, and combined with the concentration at the baseline point, is as follows: The starting time within the analysis window is recorded as the reference point, and the difference between the water level at each time point and the water level at the reference point is recorded as the water level change. The product of the water level change and the short-time fluctuation response coefficient is recorded as the first product. The first product is added to the concentration at the reference point to obtain the estimated value of the hydrodynamic correlation component.

8. The method for real-time monitoring of groundwater pollution concentration as described in claim 1, characterized in that, The method for determining the drift increment by using the regression slope and analysis window after subtracting the estimated hydrodynamic components from the concentration baseline, and updating the preset zero-point drift value, is as follows: The drift increment is the product of the regression slope and the length of the analysis window; The sum of the drift increment, corrected by the damping coefficient, and the preset zero-point drift value is used as the updated zero-point drift value.

9. The method for real-time monitoring of groundwater pollution concentration as described in claim 1, characterized in that, The concentration calibration data is the difference between the real-time concentration under standard conditions and the updated zero-point drift value.

10. A real-time monitoring system for groundwater pollution concentration, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for real-time monitoring of groundwater pollution concentration as described in any one of claims 1-9.

Images