Integrated seepage flow intelligent monitoring method and monitor

Abstract

The application discloses an integrated seepage flow intelligent monitoring method and a monitor, and the method comprises the following steps: acquiring original water level distance generated based on a single-pass propagation mechanism and synchronously collecting current temperature; performing temperature compensation on the original water level distance based on the current temperature to obtain compensated water level elevation; calculating the water head on the weir based on the compensated water level elevation and preconfigured structure parameters, and converting to obtain seepage flow; utilizing multi-level detection criteria to perform real-time state monitoring on the seepage flow, entering an active confirmation mode when any level criterion is triggered, performing abnormality determination through the time sequence evaluation index of the high-frequency re-sampling sequence by improving the sampling frequency, and obtaining an abnormality confirmation flag; determining a final alarm level based on a trigger combination state and the abnormality confirmation flag, and scheduling a communication link to output a message. The application moves the abnormality confirmation logic to the sensor end, is helpful to identify transient interference and real danger, and is favorable to improving the accuracy and early warning timeliness of seepage flow monitoring.

Description

Technical Field

[0001] This invention relates to the field of dam and engineering safety monitoring technology, and in particular to an integrated intelligent seepage flow monitoring method and monitor. Background Technology

[0002] Seepage flow is a direct and sensitive physical parameter representing the safety status of hydraulic structures such as dams, dikes, and tailings ponds. Accurate seepage flow monitoring can reflect the evolution of the seepage field inside the structure in real time, providing basic data support for assessing the integrity of seepage barriers, obtaining hydraulic gradient change characteristics, and early detection of structural hazards such as piping. It plays an important technical role in the safety assessment and hazard early warning system of water conservancy projects.

[0003] Currently, seepage monitoring in hydraulic structures primarily employs a water level-flow conversion scheme based on weirs and flumes. In actual engineering deployments, level gauges such as magnetostrictive sensors are typically used to obtain the head above the weir, and the collected distance and temperature data are then transmitted back to a host computer system. The host computer system performs preliminary compensation on the data based on a preset linear temperature coefficient and calculates the final seepage flow rate by combining it with a set piecewise constant flow coefficient. Regarding anomaly monitoring, existing systems mainly rely on a monitoring center platform. The platform software compares the returned seepage flow data with manually set alarm upper and lower limits using absolute thresholds, or counts multiple instances of exceeding limits to trigger the backend early warning process.

[0004] However, in long-term, wide-temperature-varying outdoor environments, physical measurement systems face complex challenges from nonlinear interference and high-frequency environmental noise. On the one hand, basic physical calibration and flow conversion models are prone to numerical steps and cumulative biases under complex boundary conditions, leading to a decrease in the continuity and fidelity of the underlying measurement data. On the other hand, the traditional "front-end centralized acquisition - back-end threshold judgment" architecture heavily relies on stable communication links and struggles to intelligently separate occasional transient interference from the gradual trends of the initial stages of real emergencies in a short period, easily resulting in false alarms or missed alarms. Therefore, it is necessary to study a monitoring method that can improve the smoothness of underlying data calculation and the robustness of front-end perception and judgment under complex operating conditions. Summary of the Invention

[0005] Purpose of the invention: To provide an integrated intelligent monitoring method and device for seepage flow to solve the existing technical problems.

[0006] Technical solution: An integrated intelligent monitoring method for seepage flow, comprising:

[0007] Obtain the original water level distance based on the one-way propagation mechanism, and simultaneously collect the current temperature;

[0008] Temperature compensation is applied to the original water level distance based on the current temperature to obtain the compensated water level elevation.

[0009] The water head above the weir is calculated based on the compensated water level elevation and the pre-configured structural parameters, and the seepage flow rate is obtained by converting the water head above the weir.

[0010] The seepage flow is monitored in real time using pre-configured multi-level detection criteria. When any level criterion is triggered, the active confirmation mode is entered. Anomalies are determined by increasing the sampling frequency and extracting time-series evaluation indicators of high-frequency resampling sequences, and an anomaly confirmation flag is obtained.

[0011] The final alarm level is determined by the trigger combination state and abnormal confirmation flag based on the multi-level detection criteria, and the communication link is scheduled to output data packets according to the final alarm level.

[0012] An integrated intelligent seepage flow monitor includes a main control module, a magnetostrictive sensor assembly, a temperature sensor, and a communication module;

[0013] The main control module includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor;

[0014] When the processor executes the computer program, it implements the method described in any of the embodiments. The magnetostrictive sensor assembly, temperature sensor, and communication module are respectively connected to the main control module.

[0015] Beneficial effects: This invention moves the anomaly confirmation logic forward to the sensor end. The judgment mechanism that combines high-frequency resampling and three-dimensional time series evaluation indicators in the active confirmation mode helps to distinguish between transient interference and real danger, which is conducive to improving the accuracy of seepage monitoring and the timeliness of early warning. Attached Figure Description

[0016] Figure 1 This is a flowchart of the integrated intelligent seepage flow monitoring method of the present invention.

[0017] Figure 2 This is a flowchart illustrating the process of adjusting the water level elevation at the distance from the original water level to the temperature-compensated water level in this invention.

[0018] Figure 3 This is a flowchart for determining the comprehensive compensation factor for temperature deviation in this invention.

[0019] Figure 4 This is a flowchart illustrating the calculation of seepage flow rate from the head above the weir according to the present invention.

[0020] Figure 5 This is a flowchart illustrating the process of extracting time-series evaluation indicators from high-frequency resampling sequences to characterize fluctuation amplitude and evolution trend, as described in this invention. Detailed Implementation

[0021] Example 1: This example provides an integrated intelligent monitoring method for seepage flow, such as... Figure 1As shown, the method includes the following steps:

[0022] Step 101: Obtain the original water level distance generated based on the one-way propagation mechanism, and simultaneously collect the current temperature;

[0023] In this embodiment, after the sensor is powered on, the main control module loads the factory-pre-written parameter set from its internal non-volatile memory. Specifically, the parameter set may include the sensor's unique identifier, reference temperature, reference wave velocity, threshold upper and lower limits, maximum rate of change, statistical deviation multiple, and communication parameters. The main control module performs a self-test to verify that all modules communicate normally and that the power supply voltage meets the operating requirements, and then enters the cyclic acquisition state. In each acquisition cycle, the main control module triggers the current pulse excitation circuit to apply a current pulse to the magnetostrictive waveguide wire. At this time, the electromagnetic pulse propagates along the waveguide wire at near the speed of light, generating a torsional wave at the position of the float magnetic ring through the Widman effect. Since the torsional wave propagates unidirectionally towards the head-end pickup at the speed of sound, the time difference recorded by the timer is the one-way propagation time. The main control module uses this one-way propagation time and the pre-configured reference wave velocity to calculate the original water level distance. The specific calculation formula is as follows:

[0024] L_raw=v_0*Δ_t;

[0025] Where L_raw is the original water level distance, v_0 is the reference wave velocity, and Δ_t is the one-way propagation time. Simultaneously with acquiring the original water level distance, the main control module reads the current temperature from the temperature sensor, which serves as the input data for subsequent temperature compensation.

[0026] Step 102: Perform temperature compensation on the original water level distance based on the current temperature to obtain the compensated water level elevation;

[0027] To address the nonlinear coupling error caused by the temperature dependence of torsional wave velocity and the thermal expansion effect of the structure, the main control module uses a comprehensive compensation factor to scale and correct the original water level distance.

[0028] Specifically, the main control module calculates the temperature deviation between the current temperature and the pre-configured reference temperature, and determines a comprehensive compensation factor based on this temperature deviation. The construction details of the comprehensive compensation factor will be further elaborated in subsequent embodiments.

[0029] The main control module multiplies the comprehensive compensation factor by the original water level distance to obtain the corrected water level distance. It then calculates the difference between the preset sensor installation reference elevation and the corrected water level distance to obtain the compensated water level elevation, providing reference data for subsequent flow conversion.

[0030] Step 103: Calculate the head above the weir based on the compensated water level elevation and the pre-configured structural parameters, and calculate the seepage flow rate based on the head above the weir.

[0031] When calculating the head above the weir, the main control module subtracts the pre-configured weir crest elevation from the compensated water level elevation to obtain the head above the weir. If the head above the weir is less than or equal to zero, the seepage flow is directly output as zero. If the head above the weir is greater than zero, the main control module calls the corresponding hydraulic formula based on the pre-configured structural parameters. Specifically, the pre-configured structural parameters may include the weir type identifier and corresponding geometric parameters pre-written at the factory.

[0032] For example, if the weir type is identified as a triangular thin-walled weir, the corresponding formula including the weir crest angle is used; if the weir type is identified as a rectangular thin-walled weir, the corresponding formula including the weir width is used. After determining the weir type, the main control module can use a preset piecewise constant flow coefficient method for conversion. In this method, the head range above the weir is divided into multiple intervals, each interval corresponding to a fixed flow coefficient. The main control module selects the corresponding constant flow coefficient according to the interval to which the current head above the weir belongs, and substitutes it into the hydraulic formula to calculate the seepage flow.

[0033] Step 104: Real-time status monitoring of seepage flow is performed using pre-configured multi-level detection criteria, and the trigger combination status of each level of criteria is recorded. When any level of criterion is triggered, the active confirmation mode is entered. Anomalies are determined by increasing the sampling frequency and extracting the time-series evaluation index of the high-frequency resampling sequence, and an anomaly confirmation flag is obtained.

[0034] The pre-configured multi-level detection criteria can include threshold detection, rate detection, and statistical deviation detection. During real-time monitoring, the main control module continuously substitutes the calculated seepage flow rate into the multi-level detection criteria. If the absolute value, rate of change, or statistical deviation of the seepage flow rate exceeds the corresponding preset threshold, it indicates a possible abnormal operating condition, and any level of criterion is triggered. After triggering, the main control module enters an active confirmation mode. In this mode, the main control module switches the sampling frequency from the normal operating cycle (e.g., 10 minutes to 30 minutes) to a high-frequency confirmation cycle (e.g., 5 seconds to 20 seconds), continuously executing steps 101 to 103 above within a short period of time to construct a high-frequency resampling sequence composed of multiple dense sampling data.

[0035] The main control module analyzes the high-frequency resampling sequence and extracts time-series evaluation indicators reflecting its numerical fluctuation amplitude and temporal evolution trend. Based on these indicators, the main control module can perform comprehensive judgments such as majority voting. If the percentage of anomalies detected during continuous resampling reaches a preset proportion, it is determined to be a genuine anomaly, and a confirmation flag is output; otherwise, it is determined to be transient interference. Through this closed-loop confirmation mechanism, false alarms caused by water surface fluctuations, etc., can be autonomously identified and filtered at the sensor end.

[0036] Step 105: Determine the final alarm level based on the trigger combination status and abnormal confirmation flag of the multi-level detection criteria, and schedule the communication link to output data packets according to the final alarm level.

[0037] After receiving a confirmed anomaly flag, the main control module extracts the state combinations of various detection criteria that trigger the active confirmation mode. Based on preset mapping rules, the combined states are converted into the final alarm level. For example, only exceeding the statistical deviation limit corresponds to the attention level, while exceeding the absolute threshold corresponds to the warning or alarm level.

[0038] After determining the final alarm level, the main control module schedules different communication links according to that level. For example, under normal or alert levels, the main control module can schedule low-power communication links (such as LoRa channels) to report data periodically; under alarm levels, the main control module will wake up high-speed cellular network links (such as 4G channels) for immediate transmission. The output data packets include fields such as sensor identifier, timestamp, seepage flow rate, compensated water level elevation, current temperature, and battery voltage, delivering the complete monitoring results to the backend system.

[0039] Example 2: In one possible implementation, temperature compensation is performed on the original water level distance based on the current temperature to obtain the compensated water level elevation, such as... Figure 2 and Figure 3 As shown, it includes the following steps:

[0040] Step 201: Calculate the temperature deviation between the current temperature and the pre-configured reference temperature; determine the comprehensive compensation factor based on the temperature deviation. The comprehensive compensation factor is used to characterize the nonlinear coupling error jointly constituted by the temperature dependence of the torsional wave velocity and the thermal expansion effect of the structure; use the comprehensive compensation factor to scale and correct the original water level distance, and convert the corrected distance into the compensated water level elevation based on the pre-configured sensor installation reference elevation.

[0041] Wherein, the current temperature is the real-time ambient temperature value T read by the sensor's built-in temperature probe during the current acquisition cycle, and the reference temperature is the baseline temperature value T_0 pre-stored in non-volatile memory, which is set to 20℃ in this embodiment. The main control module obtains the temperature deviation through subtraction. The specific calculation formula is as follows:

[0042] Δ_T = T - T_0;

[0043] Where Δ_T is the temperature deviation, T is the current temperature, and T_0 is the reference temperature.

[0044] Specifically, the propagation speed of the torsional wave in the iron-nickel alloy waveguide wire drifts with changes in the material's elastic modulus and density caused by temperature. Simultaneously, the external support structure of the sensor experiences mechanical zero-point displacement due to thermal expansion and contraction. These physical factors exhibit nonlinear coupling and superposition over a wide temperature range. This embodiment introduces a comprehensive compensation factor to proportionally scale the original measurement distance. The scaled distance, combined with the sensor's installation reference elevation, is used to determine the current compensated water level elevation through subtraction. The specific calculation formula is as follows:

[0045] H_w = H_ref - L_comp;

[0046] Where H_w is the compensated water level elevation, H_ref is the sensor installation reference elevation, and L_comp is the corrected water level distance.

[0047] Step 202: Obtain the pre-configured first-order temperature coefficient and second-order temperature coefficient; multiply the temperature deviation by the first-order temperature coefficient as the linear superposition effect; multiply the square of the temperature deviation by the second-order temperature coefficient as the nonlinear coupling effect; sum the preset benchmark constant, the linear superposition effect, and the nonlinear coupling effect to obtain the comprehensive compensation factor.

[0048] In this embodiment, the comprehensive compensation factor K_T is constructed as a second-order polynomial function of the temperature deviation. The main control module extracts the first-order temperature coefficient c_1 and the second-order temperature coefficient c_2 from the parameter set. The first-order temperature coefficient c_1 is mainly used to compensate for the linear components of wave velocity variation and structural linear expansion, while the second-order temperature coefficient c_2 is used to correct higher-order nonlinear terms. During calculation, the temperature deviation is first multiplied by c_1 to obtain the linear superposition effect, and then the square of the temperature deviation is multiplied by c_2 to obtain the nonlinear coupling effect. Finally, the preset reference constant 1 is added to the two effect quantities to determine the final comprehensive compensation factor. The specific calculation formula is as follows:

[0049] K_T = 1 + c_1 * Δ_T + c_2 * Δ_T^2;

[0050] Where K_T is the comprehensive compensation factor, c_1 is the first-order temperature coefficient, c_2 is the second-order temperature coefficient, and Δ_T is the temperature deviation.

[0051] After obtaining the comprehensive compensation factor, the main control module multiplies it by the original water level distance to achieve real-time dynamic elimination of ranging errors. The specific calculation formula is as follows:

[0052] L_comp = L_raw * K_T;

[0053] Where L_comp is the corrected water level distance, L_raw is the original water level distance, and K_T is the comprehensive compensation factor.

[0054] Step 203: In a preset temperature control environment, traverse multiple set calibration temperature points and collect the true value and original measurement value of the sensor at each known standard distance; based on the temperature deviation formed by each calibration temperature point and the reference temperature, use the ratio of the true value to the original measurement value as the optimization target, and perform polynomial regression fitting using the least squares method; solve and output the first-order temperature coefficient and the second-order temperature coefficient according to the regression fitting results, and persistently store them as pre-configured parameters.

[0055] Specifically, before the sensor leaves the factory, it needs to be placed in a temperature-controlled environmental chamber to perform a full-range calibration protocol. For example, seven calibration temperature points can be set: -10, 0, 10, 20, 30, 40, and 50. At each calibration temperature point, after the environment reaches thermal equilibrium, multiple known standard distance true values ​​covering the low, medium, and high ranges of the measurement range are selected for actual measurement, and the original measurement value corresponding to each point is recorded. By establishing a proportional sample set consisting of true values ​​and original measurement values, with temperature deviation as the independent variable, the second-order regression equation is fitted using the least squares method to solve for the optimal first-order temperature coefficient c_1 and second-order temperature coefficient c_2.

[0056] In some alternative implementations, if the range of ambient temperature fluctuations is narrow, a single linear temperature coefficient can be used for compensation. This method corrects the distance by setting a single linear compensation factor, without involving calculations of second-order or higher-order terms. The specific calculation formula is as follows:

[0057] L_comp = L_raw * (1 + γ * Δ_T);

[0058] Where L_comp is the corrected water level distance, L_raw is the original water level distance, γ is the single linear temperature coefficient, and Δ_T is the temperature deviation.

[0059] This alternative has lower computational overhead and is suitable for deployment scenarios where processor computing power is limited.

[0060] Example 3: In one possible implementation, the seepage flow rate is calculated based on the head above the weir, such as... Figure 4 As shown, it includes the following steps:

[0061] Step 301: Obtain the pre-divided continuous head interval consisting of multiple head boundary points, and the calibration flow coefficient corresponding to each head boundary point; determine the target head interval to which the head above the weir belongs.

[0062] In practical engineering, the flow coefficient is not a strict constant, especially in the low head section, where viscosity and surface tension effects can have a nonlinear impact on flow conversion.

[0063] Therefore, in this embodiment, the full-range head is divided into multiple continuous head intervals by pre-determining the actual flow rate or consulting an empirical correction table.

[0064] Specifically, the pre-divided intervals consist of a series of discrete head boundary points, each associated with a physically calibrated calibration flow coefficient. During each acquisition cycle, the main control module acquires the current head above the weir and performs a numerical comparison with the pre-stored set of head boundary points to determine the specific target head interval in which the current head above the weir is located.

[0065] Step 302: Extract the upper and lower boundary heads corresponding to the target head interval, as well as the upper boundary calibration flow coefficient corresponding to the upper boundary head and the lower boundary calibration flow coefficient corresponding to the lower boundary head; use the upper and lower boundary heads, the upper boundary calibration flow coefficient, and the lower boundary calibration flow coefficient to perform piecewise linear interpolation calculation on the current weir head to obtain the continuously changing current flow coefficient, so as to reduce the step in seepage flow calculation when the interval is crossed;

[0066] After determining the target head interval, the main control module extracts the boundary characteristic parameters of that interval from non-volatile memory. The upper and lower boundary heads define the hydraulic span of the current interval, while the upper and lower boundary calibration flow coefficients are empirical values ​​of the actual flow rates at both ends of that span. The main control module uses a first-order linear interpolation algorithm, combined with the current head above the weir, to calculate the current flow coefficient. The specific calculation formula is as follows:

[0067] C_d(h)=C_d(h_lower)+(C_d(h_upper)-C_d(h_lower)) / (h_upper-h_lower)*(h-h_lower);

[0068] Where C_d(h) is the current flow coefficient, h is the current head above the weir, h_lower is the lower boundary head, h_upper is the upper boundary head, C_d(h_lower) is the lower boundary calibration flow coefficient, and C_d(h_upper) is the upper boundary calibration flow coefficient.

[0069] By performing this interpolation calculation, the calculated current flow coefficient becomes a continuous piecewise linear function of the weir head, thus ensuring that the final output seepage flow changes smoothly and continuously with the weir head. This reduces potential artificial numerical steps that may occur when the weir head crosses different interval boundaries, preventing the backend anomaly detection module from misinterpreting these numerical steps as abnormal changes in seepage flow. Furthermore, when the range is at its limit, if the current weir head is less than the minimum boundary head, the current flow coefficient is directly taken from the calibration flow coefficient corresponding to the minimum boundary; if the current weir head is greater than the maximum boundary head, the current flow coefficient is directly taken from the calibration flow coefficient corresponding to the maximum boundary, thus avoiding distortion caused by far-end linear extrapolation.

[0070] Step 303: Read the factory-pre-written weir type identifier, and when the weir type identifier indicates a triangular thin-walled weir or a rectangular thin-walled weir, call the preset geometric feature parameters that match the corresponding weir type respectively, and perform the corresponding hydraulic formula calculation on the water head above the weir in combination with the current flow coefficient to complete the conversion of seepage flow.

[0071] When deploying seepage monitoring equipment, the actual physical geometry of the weir plate being measured may vary. The main control module automatically matches the corresponding hydraulic calculation logic by reading the weir type identifier from the parameter set. For example, when the weir type identifier indicates that the current deployment is a triangular thin-walled weir, the main control module extracts the preset geometric feature parameter specifically the weir crest angle. Then, it uses a hydraulic reference formula that includes gravitational acceleration, the tangent function of the weir crest angle, and the 5 / 2 power of the head above the weir, multiplying it by the current flow coefficient to calculate the seepage flow. When the weir type identifier indicates a rectangular thin-walled weir, the extracted preset geometric feature parameter specifically the weir width. Then, it calls a hydraulic reference formula that includes gravitational acceleration, the weir width, and the 3 / 2 power of the head above the weir, multiplying it by the current flow coefficient to solve for the seepage flow. Through the physical model calling mechanism, the same algorithm architecture is compatible and adaptable to different hardware molds.

[0072] In some alternative implementations, if the backend monitoring system is not sensitive to false alarms from sudden anomalies, a piecewise constant flow coefficient method can be used for conversion. This method configures a single, constant calibration flow coefficient for each head interval. After determining the target head interval, this constant coefficient is directly substituted into the hydraulic formula to calculate the seepage flow, without the need for linear interpolation. This alternative method can reduce the floating-point operation overhead of the main control module.

[0073] A polynomial temperature compensation mechanism with two-factor coupling and a piecewise linear interpolation flow coefficient conversion algorithm are adopted. The former effectively corrects the nonlinear measurement deviation caused by wave velocity drift and structural thermal expansion, while the latter ensures a continuous transition of the flow coefficient at interval boundaries, which helps to reduce the numerical calculation abruptness when the head crosses the segment boundaries and reduces spurious abrupt changes caused by algorithmic discontinuities.

[0074] Example 4: In one possible implementation, pre-configured multi-level detection criteria are used to perform real-time status monitoring of seepage flow, such as... Figure 5 As shown, it includes at least one of the following detection methods:

[0075] Step 401: Compare the seepage flow rate with the preset absolute limit value to obtain the absolute threshold detection result; calculate the rate of change of the seepage flow rate between adjacent cycles and compare it with the preset maximum allowable rate of change to obtain the rate of change detection result.

[0076] Additionally, the sliding window statistical deviation detection method extracts recent sliding window statistical features of seepage flow and evaluates the degree of deviation of the current value from the sliding baseline to obtain the sliding window statistical deviation detection results.

[0077] In this embodiment, after receiving the seepage flow rate calculated for the current acquisition cycle, the main control module inputs it into a multi-level parallel detection branch. The first level is absolute threshold detection, where the main control module determines whether the current seepage flow rate is greater than a preset alarm upper limit or less than a preset alarm lower limit. The second level is rate of change detection, where the main control module calculates the difference between the current seepage flow rate and the seepage flow rate of the previous cycle, divides this difference by the normal acquisition cycle to obtain the real-time rate of change, and then determines whether the absolute value of the real-time rate of change is greater than a preset maximum allowable rate of change. The third level is sliding window statistical deviation detection, where the main control module maintains a sliding window with a fixed number of samples, calculates the sliding average and sliding standard deviation of historical seepage flow rates within the window. Then, it calculates the absolute difference between the current seepage flow rate and the sliding average, and determines whether this absolute difference is greater than the product of a preset statistical deviation multiple and the sliding standard deviation. The preset statistical deviation multiple is typically set to 3.0. If the judgment result of any of the above three levels of detection is true, a trigger command is generated, indicating that the system may have intercepted a sudden piping failure or a partial sensor malfunction.

[0078] Step 402: Extract the statistical features from the seepage flow sliding window maintained during the most recent normal acquisition cycle before triggering the active confirmation mode, and save it as a frozen baseline; during the operation of the active confirmation mode, pause the update of the seepage flow sliding window, and use the frozen baseline as a unified reference benchmark for sliding window statistical deviation detection to prevent high-frequency re-acquisition data from polluting the normal statistical benchmark.

[0079] Upon receiving the trigger command, the main control module extracts the latest calculated moving average and moving standard deviation before the current moment and stores them in a separate register area to form a frozen baseline. If the standard deviation of the frozen baseline is less than a preset minimum protection value, it is set to that minimum protection value to ensure the numerical stability of subsequent division operations. After entering the active confirmation mode, due to the significant increase in the acquisition frequency, if the sliding window continues to be updated at this stage, the abnormal data acquired at high frequencies will quickly fill the entire window. This will cause the calculated moving standard deviation to increase sharply, and the moving average to shift rapidly in the abnormal direction, causing the sliding window statistical deviation detection condition to quickly fail and generating the risk of missed detections. Therefore, during the active confirmation period, the main control module cuts off the data push operation of the historical window and forces all single high-frequency re-acquisition data to be calculated for deviation using the latched frozen baseline. By performing the baseline freezing operation, the statistical reference system under normal operating conditions is protected, ensuring the consistency of anomaly identification sensitivity during the confirmation period.

[0080] Upon receiving the trigger command, the main control module locks the frozen baseline and latches the seepage flow calculated from the last normal acquisition cycle before the current moment as the reference flow before the trigger, storing it in an independent register for subsequent rate over-limit ratio calculation.

[0081] Step 403: At the trigger time, the acquisition period is switched to the preset confirmation sampling period, and multiple acquisitions and calculations are performed continuously to construct a high-frequency resampling sequence; this preset confirmation sampling period has a higher frequency than the current sampling period.

[0082] After generating the trigger command, the main control module updates the trigger interval of its internal timer from the long period of normal operation to the short period of the confirmation sampling period. For example, if the normal acquisition period is 600 seconds, the confirmation sampling period can be set to 10 seconds. The total number of continuous acquisitions, N_confirm, can be set to 20. Those skilled in the art can determine the specific number of acquisitions based on the required statistical test confidence level and equipment power consumption budget, using conventional engineering experience. Under this high-frequency cycle, the main control module continuously executes the specified number of magnetostrictive waveguide excitations, temperature readings, compensation calculations, and flow conversions. The instantaneous seepage flow obtained from each calculation is stored in a memory array in chronological order, constructing an equally spaced high-frequency resampling sequence, providing a data foundation for subsequent extraction of comprehensive decision features.

[0083] Time series evaluation indicators for characterizing fluctuation amplitude and evolution trend are extracted from high-frequency resampling sequences, specifically including:

[0084] Step 404: Perform detection on each single re-sampling data in the high-frequency re-sampling sequence using multi-level detection criteria, and calculate the proportion of abnormal detections to obtain the binary confirmation rate; calculate the overall magnitude of the deviation of the high-frequency re-sampling sequence from the normal range defined by the pre-configured detection limits at each level to obtain the aggregated severity index; extract the change characteristics of the high-frequency re-sampling sequence on the time axis to obtain the trend consistency index; and use the binary confirmation rate, aggregated severity index, and trend consistency index together to form a time series evaluation index.

[0085] For high-frequency resampling sequences, the main control module extracts feature vectors from three dimensions: frequency, amplitude, and trend. In the frequency dimension, the main control module re-executes the aforementioned three-level detection logic for each seepage flow sample in the high-frequency resampling sequence; a single sample is considered detected if it triggers any of the first-level criteria. The total number of detected samples is divided by the total number of sampling attempts to calculate the binary confirmation rate. In the amplitude and trend dimensions, the main control module extracts aggregated severity and trend consistency indicators, respectively. These three indicators are combined to generate a time-series evaluation index containing multi-dimensional judgment information. Traditional methods rely solely on binary voting for judgment, which easily loses information about the differences in physical destructive force between minor and significant exceedances. This embodiment improves the reliability of the system in identifying real deterioration risks in complex noise backgrounds by constructing a three-dimensional confirmation framework composed of frequency, severity, and trend.

[0086] Step 405: For each resampled data in the high-frequency resampled sequence, calculate the over-limit ratio compared to the preset absolute alarm limit, the rate over-limit ratio calculated based on the corresponding time span, and the deviation factor of the statistical standard deviation compared to the frozen baseline; extract the maximum deviation value among the over-limit ratio, rate over-limit ratio, and statistical standard deviation deviation factor as the single severity, and use the average single severity of multiple resampled data as the aggregated severity index; perform linear regression fitting on the high-frequency resampled sequence as an equally spaced time series to extract the regression slope, and use the total number of collections of the high-frequency resampled sequence and the standard deviation of the frozen baseline to convert the regression slope into a dimensionless trend parameter, and output a trend consistency index reflecting the continuous increase or continuous decrease state based on the comparison result of the trend parameter and the preset trend significance threshold.

[0087] Specifically, the main control module performs normalized severity calculation on the j-th sampled data of the high-frequency resampling sequence. The limit exceedance ratio is the absolute limit exceedance value of the difference between the current flow rate and the alarm upper and lower limits, divided by the normal range span. The rate exceedance ratio is calculated by subtracting the current resampling flow rate from the last normal flow rate before triggering, dividing by the cumulative time span from the confirmation to the current sampling point, and finally dividing by the maximum allowable rate of change. Since the time span expands with increasing j, this rate calculation method can effectively suppress measurement differential noise in high-frequency sampling. The statistical standard deviation factor is the absolute difference between the current flow rate and the average value of the frozen baseline, divided by the product of the statistical deviation factor and the standard deviation of the frozen baseline.

[0088] Specifically, when the traffic volume Q_j of the j-th resampling exceeds the alarm limit Q_high, the over-limit ratio s_1j = (Q_j - Q_high) / (Q_high - Q_low);

[0089] When Q_j is less than the alarm lower limit Q_low, s_1j=(Q_low-Qj) / (Q_high-Q_low);

[0090] When Q_j is within the alarm upper and lower limits, s_1j=0.

[0091] Where Q_high is the preset alarm upper limit, Q_low is the preset alarm lower limit, and (Q_high-Q_low) is the alarm range.

[0092] The main control module selects the maximum value among the above three ratios as the single severity of the j-th sample, and the calculation formula is as follows:

[0093] S_j=max(s_1j,s_2j,s_3j);

[0094] Where S_j represents the single severity score, s_1j represents the rate of exceedance, s_2j represents the rate of exceedance, and s_3j represents the fold deviation of the statistical standard deviation. The aggregate severity index is calculated by summing the S_j values ​​of all samples and dividing by the total number of data collections.

[0095] When calculating the trend consistency index, the main control module uses the sampling sequence number as the independent variable and the resampled flow rate as the dependent variable to perform linear equation fitting, calculating the regression slope characterizing the flow rate change at each sampling interval. Subsequently, the main control module converts this into a dimensionless trend parameter using the following formula:

[0096] τ=(b*(N_confirm-1)) / σ_frozen;

[0097] Where τ is a dimensionless trend parameter, b is the regression slope, N_confirm is the total number of consecutive data collections, and σ_frozen is the standard deviation of the frozen baseline.

[0098] The main control module compares τ with a preset trend significance threshold (e.g., 2.0). If τ is greater than or equal to the threshold, the trend consistency index is assigned a value of +1; if τ is less than or equal to a negative number of the threshold, it is assigned a value of -1; if it is between the two, it is assigned a value of 0.

[0099] Step 406: Perform comprehensive voting based on the timing evaluation index to determine whether the trigger is a transient disturbance. If it is confirmed as a real anomaly, output an anomaly confirmation flag. If it is confirmed as a transient disturbance, output an anomaly confirmation flag.

[0100] The main control module compares the calculated binary confirmation rate with a preset confirmation threshold (e.g., 0.7). If the binary confirmation rate is greater than or equal to the confirmation threshold, it indicates that most of the resampling data exhibits abnormal characteristics. The main control module determines that the trigger is not a random fluctuation but a real abnormal state, thus generating an anomaly confirmation flag indicating a true logical state. The above comprehensive voting process uses the binary confirmation rate as the primary criterion for confirming an anomaly, while the aggregated severity index and trend consistency index serve as auxiliary quantitative outputs of the anomaly characteristics. The three indicators play different roles in the comprehensive voting framework: the binary confirmation rate is responsible for the binary decision in the frequency dimension, while the aggregated severity index and trend consistency index are responsible for characterizing the magnitude characteristics and temporal evolution of the anomaly, respectively. Together, they constitute a complete judgment output from confirmation to qualitative assessment of the anomaly event, and the aggregated severity index and trend consistency index are simultaneously sent to the backend hierarchical module.

[0101] Step 407: If the trigger is determined to be transient interference, the current interference features are extracted based on the high-frequency resampling sequence, a transient event log containing the current interference features is generated and stored, and the monitor is controlled to exit the active confirmation mode and the sampling frequency is restored to the normal acquisition cycle.

[0102] If the binary confirmation rate is strictly less than the confirmation threshold, the main control module determines that the original triggering action was a transient jump caused by accidental factors such as water surface wave fluctuations or bubbles passing over the float. At this time, the main control module blocks the external transmission of alarm signals, extracts the peak value and occurrence time of the high-frequency resampling sequence, encapsulates it into a transient event log, and stores it in the internal file system. Simultaneously, a reset command is sent to the hardware control logic to shut down the high-frequency clock cycle, restore the normal working cycle, and balance the relationship between detection sensitivity and device communication and computing power consumption.

[0103] In some alternative implementations, to reduce the computational load on low-cost microcontrollers, the extraction of aggregate severity and trend consistency metrics can be omitted. This alternative, after constructing the high-frequency resampling sequence, determines the authenticity of anomalies solely based on a single comparison of the binary confirmation rate and the confirmation threshold. While this alternative eliminates the floating-point operations of the dimensionless formula, it cannot provide fine-grained dynamic adjustments to subsequent alarm levels. It is suitable for basic monitoring scenarios with extremely high power consumption requirements and low engineering risks.

[0104] The multi-level anomaly detection and proactive confirmation logic is moved forward to the front-end measurement node. By introducing a high-frequency resampling mechanism based on frozen statistical benchmarks and establishing a three-dimensional indicator evaluation system covering confirmation frequency, normalized severity, and temporal evolution trend, it is beneficial to filter false triggers caused by accidental environmental factors such as water surface fluctuations or bubbles, and at the same time, it is beneficial to capture the continuous deterioration characteristics in the early stage of seepage damage, thereby reducing false alarm and missed alarm rates.

[0105] Example 5: In one possible implementation, the multi-dimensional evaluation-driven dynamic rating and adaptive scheduling strategy for communication links includes the following steps:

[0106] Step 501: When an abnormal confirmation flag is obtained and confirmed to be true, extract the trigger combination state of the multi-level detection criteria that trigger the active confirmation mode; according to the trigger combination state in the multi-level detection criteria where a single-level criterion is triggered independently or multiple-level criteria are triggered synchronously, map the current state to a preset basic alarm level; output the basic alarm level as the final alarm level, or use the basic alarm level as the basis for subsequent dynamic level upgrades.

[0107] Specifically, after receiving a confirmed anomaly flag, the main control module backtracks and extracts the Boolean trigger states of each level of detection criteria before triggering high-frequency resampling. Given that the multi-level detection criteria include absolute threshold detection, rate of change detection, and sliding window statistical deviation detection, the main control module is configured with a discrete mapping table covering all trigger combinations. For example, if and only if the sliding window statistical deviation detection is triggered independently, or if and only if the rate of change detection is triggered independently, both of which indicate that the seepage flow has deviated or is rapidly adjusting but the absolute value has not yet exceeded the limit, the main control module maps the basic alarm level to level 1. When the absolute threshold detection is triggered, or when the rate of change detection and the sliding window statistical deviation detection are triggered simultaneously and synchronously, the main control module maps the basic alarm level to level 2. When the absolute threshold detection and the rate of change detection are triggered simultaneously and synchronously, indicating that the seepage flow has not only exceeded the limit but is also rapidly deteriorating, the main control module directly maps the basic alarm level to level 3. Under normal circumstances without any triggers, the basic alarm level is level 0.

[0108] Step 502: Obtain the aggregated severity index used to characterize the magnitude of the anomaly and the trend consistency index used to characterize the evolution trend generated during the extraction of time series evaluation indicators in the active confirmation mode; when the aggregated severity index reaches the preset severity upgrade threshold and the trend consistency index indicates that the anomaly shows a continuous deterioration trend, perform a level upgrade operation on the basic alarm level to obtain the final alarm level; otherwise, directly confirm the basic alarm level as the final alarm level.

[0109] To overcome the limitation of static threshold mapping in preventing early intervention in the deterioration phase, this embodiment introduces dynamic feedforward control logic. The main control module acquires a time-series evaluation index including the aforementioned aggregated severity index and trend consistency index, and sets a preset severity escalation threshold. The preset severity escalation threshold can be set to 2.0. The main control module executes a logical judgment: when the aggregated severity index is greater than or equal to the severity escalation threshold, and the trend consistency index is equal to +1, the main control module determines that the current abnormal potential is severe and the direction is continuously deteriorating. Under this condition, the main control module performs a mathematical operation of incrementing the basic alarm level by 1 to complete the escalation operation.

[0110] The specific calculation formula is: E_final=min(E_base+1,3);

[0111] Where E_final is the final alarm level, E_base is the base alarm level, min is the comparison function that takes the smaller value, and 3 is the preset upper limit of the alarm level.

[0112] Correspondingly, if the above judgment conditions are not met simultaneously, the main control module maintains the original judgment conclusion.

[0113] The final alarm level is E_final = E_base; where E_base is the base alarm level.

[0114] Step 503: When the final alarm level indicates a normal state or a low-level alert state, a preset low-power communication link is used to report data packets at a regular cycle or a shortened encrypted transmission cycle; when the final alarm level indicates a high-level warning state, the high-speed cellular network link is activated first to send data packets immediately; when the final alarm level indicates the highest level alarm state, the high-speed cellular network link and the low-power communication link are activated simultaneously, and data packets are forcibly pushed using a dual-channel redundancy method.

[0115] Based on the calculated final alarm level, the main control module performs hardware power switching and channel scheduling for the various communication radio frequency modules within the system. When the final alarm level is 0, corresponding to the normal state, the main control module maintains low-power communication link operation and sends data packets according to the default regular cycle.

[0116] When the final alarm level is 1, corresponding to the alert state, the main control module still uses a low-power communication link, but the transmission cycle is shortened proportionally. When the final alarm level is 2, corresponding to the warning state, the main control module prioritizes powering the high-speed cellular network link and establishing a connection, and immediately sends data packets.

[0117] When the final alarm level reaches 3, corresponding to the alarm status, the main control module simultaneously activates the low-power communication link and the high-speed cellular network link, sending the same data packet concurrently through dual channels to prevent information loss due to congestion at a single node. In addition to encapsulating basic data such as the sensor's unique identifier, timestamp, seepage flow rate, compensated water level elevation, current temperature, and battery voltage in the data packet, the main control module also aggregates severity indicators and trend consistency indicators as additional fields for joint reporting.

[0118] Step 504: The low-power communication link is specifically configured as a LoRa communication channel, and the high-speed cellular network link is specifically configured as a 4G or 5G communication channel; the monitor also maintains a working near-field Bluetooth channel for on-site mobile terminals to directly access and read data packets, and configures a wired serial port channel as a backup link for data packet transmission.

[0119] In specific engineering deployments, the main control module is connected to various physical transceiver hardware components. The low-power communication link is specifically constructed from LoRa radio frequency hardware based on spread spectrum modulation technology, used to meet the requirements of long-cycle, low-power, long-distance transmission. The high-speed cellular network link is specifically constructed from 4G or 5G baseband modem chips that support broadband networks.

[0120] In addition, the main control module's printed circuit board integrates a Bluetooth antenna and a physical serial interface. The near-field Bluetooth channel operates as a wireless debugging interface, used to complete firmware updates or parameter rewriting without physical contact. The wired serial port channel serves as a fallback link, ensuring that the data acquisition terminal can directly retrieve data records from the memory via a wired physical cable when external environmental factors cause all RF wireless networks to be interrupted.

[0121] In some alternative implementations, if the communication power supply in the monitoring scenario is extremely limited, the system can be configured with only a single LoRa communication channel. Under this alternative, as the final alarm level evolves from low to high, the main control module replaces hardware switching between different network links by continuously increasing the number of repetitive broadcasts by the RF chip within a single transmission cycle. This ensures information delivery rate while reducing the overall communication hardware power consumption and manufacturing cost of the device.

[0122] Example 6: Based on Example 1 above, this example further provides an implementation method for cross-sensor linkage verification in large-scale distributed deployment scenarios.

[0123] Step 601: Generate an abnormal event message based on the abnormal confirmation flag and the current seepage flow rate, and broadcast the abnormal event message to the nearby sensors within the physical space range through the communication link to trigger the nearby sensors to switch to encrypted sampling mode;

[0124] In this embodiment, a neighboring sensor refers to other monitoring nodes that are physically within the effective coverage radius of the local node's single-hop communication and are grouped into the same network communication group. Specifically, the effective coverage radius of single-hop communication is typically between 1 km and 3 km. Each monitor preloads its local neighbor node addressing list during factory configuration or installation and commissioning. When the local main control module internally analyzes and obtains a confirmed anomaly confirmation flag, the main control module extracts the local sensor's unique identifier, specific message type, timestamp, current final alarm level, current seepage flow, and flow change direction flag, and encapsulates the above data according to a fixed protocol to generate an anomaly event message. Subsequently, the main control module schedules the communication link to broadcast the anomaly event message to the external space. Upon receiving the anomaly event message, the internal control logic of nearby sensors in dormant or conventional low-frequency acquisition states is triggered, forcing them to interrupt their original timed sleep cycle and autonomously switching their acquisition frequency from a long period to a high frequency state, entering an encrypted sampling mode.

[0125] Step 602: Receive the synchronous change trend transmitted back by the nearby sensor in encrypted sampling mode;

[0126] After entering encrypted sampling mode, the proximity sensor performs multiple consecutive water level measurements and seepage flow calculations. Subsequently, the proximity sensor extracts the average value and regression slope of this high-frequency sampling sequence to generate a synchronous trend reflecting the seepage flow status of its local area, and transmits this data back to the local monitor that initially broadcast the data via a communication link. During this period, the main control module of the local monitor immediately starts an internal waiting timer after sending an abnormal event message. The timeout threshold for this waiting timer can be set to 30 seconds.

[0127] During the timer's operation, the main control module opens the RF receiving window to continuously receive the synchronous change trends transmitted back from each neighboring sensor. Furthermore, if the main control module fails to receive any transmission message from a neighboring sensor after the timer reaches its timeout threshold, it executes a retransmission mechanism, rebroadcasts the abnormal event message, and resets the waiting timer. If there is still no response after the second timeout, the main control module determines that the neighboring node communication has failed and appends an explanation field indicating incomplete linkage verification to the locally generated data packet. This communication failure event is handled independently, without blocking or affecting the local monitor's own classification and alarm transmission processes.

[0128] Step 603: Perform cross-sensor consistency judgment on the current abnormal state and the synchronous change trend to distinguish whether the current abnormality is caused by a regional group seepage abnormality or by a single hardware failure of a local sensor.

[0129] The main control module performs a spatial cross-comparison between the locally confirmed abnormal change direction and all received valid synchronous change trends. If, among the extracted synchronous change trends, at least one neighboring sensor reports a significant numerical change in the same direction as the local node, such as a surge in the seepage flow of the local node and a synchronous increase in the seepage flow of at least one neighboring node, the main control module determines that the current abnormal state is not an isolated event, but rather originates from a regional group seepage anomaly.

[0130] In this situation, the main control module maintains or even escalates the current final alarm level. Conversely, if all nearby sensors that successfully transmit synchronous change trends indicate that the seepage flow at their respective measuring points is within a stable fluctuation range, i.e., no abnormal changes in the same direction are detected, the main control module determines that the current abnormal state has a very high degree of spatial isolation.

[0131] Based on these spatial characteristics, the main control module identified that the probability of this anomaly stemmed from a single hardware failure of a local sensor, such as mechanical jamming of the float or data processing distortion caused by waveguide wire deformation. Accordingly, the main control module downgraded the local alarm level or forcibly added a flag to the output data message suggesting on-site verification of the sensor hardware status, thereby assisting the system in effectively filtering false hydrological alarms caused by equipment failure.

[0132] A dynamic grading model based on multi-dimensional feature feedforward and a cross-node linkage verification mechanism were constructed. This mechanism enables early dynamic escalation of alarm responses and autonomous scheduling of low-power and high-speed redundant physical channels accordingly. Furthermore, by collecting synchronous evolution trends from neighboring monitoring points through local broadcasting, it can effectively identify local mechanical jamming faults and regional group seepage anomalies without the need for a host computer, thus improving the survivability of the monitoring network and the reliability of hazard assessment under harsh communication conditions.

[0133] According to one aspect of this application, an integrated intelligent seepage flow monitor is provided, including a main control module, a magnetostrictive sensor assembly, a temperature sensor, and a communication module;

[0134] The main control module includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor;

[0135] When the processor executes a computer program, it implements the method described in any of the embodiments.

Claims

1. An integrated intelligent monitoring method for seepage flow, characterized in that, include: Obtain the original water level distance based on the one-way propagation mechanism, and simultaneously collect the current temperature; Temperature compensation is applied to the original water level distance based on the current temperature to obtain the compensated water level elevation. The head above the weir is calculated based on the compensated water level elevation and the pre-configured structural parameters, and the seepage flow is obtained by converting the head above the weir. The seepage flow is monitored in real time using pre-configured multi-level detection criteria. When any criterion is triggered, the active confirmation mode is entered. Anomalies are determined by increasing the sampling frequency and extracting time-series evaluation indicators of high-frequency resampling sequences, and an anomaly confirmation flag is obtained. The final alarm level is determined by the trigger combination state and abnormal confirmation flag based on the multi-level detection criteria, and the communication link is scheduled to output data packets according to the final alarm level. When any level criterion is triggered, the system enters active confirmation mode. Anomalies are determined by increasing the sampling frequency and extracting time-series evaluation metrics from the high-frequency resampled sequence, resulting in anomaly confirmation flags, including: At the trigger moment, the acquisition cycle is switched to a higher frequency confirmation sampling cycle, and multiple acquisitions and calculations are performed continuously to construct a high-frequency resampling sequence; Time series evaluation indicators for characterizing fluctuation amplitude and evolution trend are extracted from high-frequency resampling sequences; A comprehensive vote is performed based on the time-series evaluation index to determine whether the trigger is a transient disturbance, and an anomaly confirmation flag is output when it is confirmed as a real anomaly. Time series evaluation indicators for characterizing fluctuation amplitude and evolution trend are extracted from high-frequency resampling sequences, specifically including: The multi-level detection criteria were used to perform detection on each single re-sampling data in the high-frequency re-sampling sequence, and the proportion of abnormal detections was statistically analyzed to obtain the binary confirmation rate. The overall magnitude of the deviation of the high-frequency resampling sequence from the normal range is calculated to obtain the aggregation severity index; Extracting the variation characteristics of high-frequency resampling sequences on the time axis to obtain a trend consistency index; The time series evaluation index is composed of binary confirmation rate, aggregation severity index and trend consistency index.

2. The method according to claim 1, characterized in that, Temperature compensation is applied to the original water level distance based on the current temperature to obtain the compensated water level elevation, including: Calculate the temperature deviation between the current temperature and the pre-configured reference temperature; The comprehensive compensation factor is determined based on the temperature deviation. The comprehensive compensation factor is used to characterize the nonlinear coupling error jointly constituted by the temperature dependence of the torsional wave velocity and the thermal expansion effect of the structure. The original water level distance is scaled and corrected using a comprehensive compensation factor, and the corrected distance is converted into a compensated water level elevation.

3. The method according to claim 2, characterized in that, The comprehensive compensation factor is determined based on temperature deviation, specifically as follows: Obtain the pre-configured first-order and second-order temperature coefficients; The product of the temperature deviation and the first-order temperature coefficient is used as the linear superposition effect quantity. The product of the square of the temperature deviation and the second-order temperature coefficient is used as the nonlinear coupling effect quantity. The comprehensive compensation factor is obtained by summing the preset benchmark constant, the linear superposition effect, and the nonlinear coupling effect.

4. The method according to claim 1, characterized in that, The seepage flow rate is calculated based on the head above the weir, including: Obtain a pre-divided continuous head interval consisting of multiple head boundary points, and the calibration flow coefficient corresponding to each head boundary point; Determine the target head range to which the water head above the weir belongs; The current flow coefficient is determined based on the calibrated flow coefficients corresponding to the target head interval and its endpoints. Hydraulic calculations are performed on the head above the weir using the current flow coefficient, and the seepage flow is then calculated.

5. The method according to claim 1, characterized in that, Real-time status monitoring of seepage flow is performed using pre-configured multi-level detection criteria, including at least one of the following detection methods: Absolute threshold detection compares the seepage flow rate with a preset absolute limit. Calculate the rate of change of seepage flow between adjacent cycles and compare it with the preset maximum allowable rate of change. It also includes the extraction of recent sliding window statistical features of seepage flow and the detection of sliding window statistical bias to assess the degree to which the current value deviates from the sliding baseline.

6. The method according to claim 1, characterized in that, The construction of the high-frequency resampling sequence also includes: Extract the statistical features of the sliding window from the most recent normal collection cycle update before the active confirmation mode is triggered, and save it as the frozen baseline; During active confirmation mode operation, the sliding window updates are paused, and the frozen baseline is used as a unified reference benchmark for sliding window statistical bias detection to prevent high-frequency resampling data from contaminating the normal statistical benchmark.

7. The method according to claim 1, characterized in that, The aggregation severity index and trend consistency index are extracted using the following methods: For each resampled data in the high-frequency resampled sequence, calculate its over-limit ratio relative to the absolute alarm limit, its rate over-limit ratio calculated in combination with the corresponding time span, and its statistical standard deviation multiple relative to the frozen baseline. The maximum deviation among the over-limit ratio, rate over-limit ratio and the statistical standard deviation is extracted as the single severity, and the average single severity of multiple re-sampling data is used as the aggregate severity index. The high-frequency resampled sequence is used as an equally spaced time series to perform linear regression fitting to extract the regression slope. The regression slope is converted into a dimensionless trend parameter. Based on the comparison result of the trend parameter and the preset trend significance threshold, a trend consistency index reflecting the continuous increase or decrease state is output.

8. An integrated intelligent seepage flow monitor, characterized in that, It includes a main control module, a magnetostrictive sensor assembly, a temperature sensor, and a communication module; The main control module includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor; The processor executes the method as described in any one of claims 1 to 7 when executing a computer program.

Images