A sealed pressure measuring port device with replaceable sensor cavity and a corridor monitoring method

Abstract

This invention discloses a sealed pressure measuring hole device with replaceable sensor cavity and a corridor monitoring method. The device includes a primary sleeve fixed to the corridor, which is sequentially connected to a secondary sleeve and a multi-port pipe. The primary sleeve is a hollow tube with a pressure measuring hole inside. A filter geotextile is wrapped around the threaded end at the front, and a seepage-proof sealing structure is provided at the rear. A front flange with threaded holes is located at the tail end. The secondary sleeve is also hollow and can communicate with the primary sleeve and the multi-port pipe, and can accommodate monitoring sensors. The multi-port pipe interface is equipped with a sealed wire-passing plug, and the remaining interfaces are connected to external monitoring instruments. The sensor data line is led out through the sealed interface to an external processing platform. The corridor monitoring method is implemented based on the above device, enabling long-term monitoring of seepage pressure in dam corridors. This invention is applicable to various dam corridor projects, allowing for the stable deployment of sensors such as piezometers to obtain reliable data, solving problems related to sealing and monitoring instability, and improving monitoring accuracy and maintainability.

Description

Technical Field

[0001] This invention relates to the field of safety monitoring of dam bodies in water conservancy projects, and in particular to a sealed pressure measuring hole device with a replaceable sensor cavity and a corridor monitoring method. Background Technology

[0002] Earth-rock dams, concrete dams, and other types of dam engineering are widely used in reservoirs, dams, and other facilities, primarily for water storage, power generation, and flood control. As dams age, especially under the influence of water pressure and climate change, seepage may occur in the galleries and sidewalls within the dam body, becoming a significant factor affecting dam safety. During long-term operation, especially when structural defects such as galleries, cracks, or voids exist, the seepage pressure of water flowing through the dam can cause deformation and instability, and in severe cases, even lead to dam failure. Therefore, timely and accurate monitoring of the seepage pressure inside the dam body, particularly in the galleries and sidewalls, is crucial for preventing seepage risks and assessing dam safety.

[0003] Seepage hole monitoring is an important method for monitoring seepage pressure in dams, consisting of a water-stopping sleeve, a water filter, a single osmometer, and supporting pipelines. During construction, holes are drilled in high-seepage areas, and an integrated sleeve structure is lowered into the holes. The water filter corresponds to the target monitoring layer, and its outer perimeter is filled with quartz sand filter media and cement mortar water-stopping material to achieve stratified water collection and isolation from non-monitoring layers. The water-stopping sleeve is made of solid metal tubing, forming a sealed pressure-bearing chamber inside, with an internal osmometer sensing the seepage pressure. Data acquisition largely relies on periodic manual on-site readings. The analog signal from the osmometer is read using a portable instrument, manually converted into pressure values, and then recorded and analyzed. Some upgrade solutions only add a simple data acquisition instrument to achieve timed storage but cannot transmit in real time. Flow monitoring requires the additional installation of a measuring weir at the orifice, and the data is obtained through manual observation or independent measurement equipment, independent of pressure monitoring data.

[0004] Traditional methods for monitoring seepage holes have several shortcomings and fail to meet the safety monitoring needs of dams: First, the integrated casing design requires the entire casing to be pulled out for repairs when the sensor fails, resulting in high maintenance costs and long construction periods. This can also cause secondary disturbances to the dam structure, damaging the original water-stopping sealing performance and leaving safety hazards. Second, the monitoring dimensions are limited, only collecting seepage pressure data and lacking auxiliary parameters such as temperature, water quality, and flow velocity. This makes it impossible to cross-verify multiple parameters to eliminate interference, resulting in low data accuracy and difficulty in accurately determining the seepage channel and type. Third, integration and sealing are poor. Flow and pressure monitoring require separate equipment deployment, and the lack of standardized sealing for cable leads and external connections makes them prone to water and sand infiltration. Reliance on manual or near-real-time data collection leads to poor timeliness and an inability to detect sudden seepage and provide effective early warnings. Fourth, adaptability and durability are insufficient. Installation after drilling damages the existing dam structure, and the complex quartz sand filling process can easily lead to cracking and interlayer flow during long-term service, affecting the accuracy of the data.

[0005] To address these issues, scholars and engineers both domestically and internationally have proposed various improvement schemes, such as deploying multiple sensors inside the dam body and utilizing fiber optic sensing technology to monitor pressure changes. However, existing technologies still face challenges such as high equipment costs, installation difficulties, and data transmission delays, failing to effectively meet the needs for real-time monitoring of seepage pressure on the sidewalls of earth-rock dam galleries.

[0006] Therefore, there is an urgent need for an accurate, reliable and convenient monitoring device for real-time monitoring of seepage pressure changes in earth-rock dams, concrete dams and other types of dam projects, especially the sidewalls of the gallery. Summary of the Invention

[0007] To address the aforementioned problems in the existing technology, this invention provides a sealed pressure measuring hole device with a replaceable sensor cavity and a gallery monitoring method, which can monitor the seepage pressure changes on the sidewall of the earth-rock dam gallery in real time and effectively improve the accuracy of dam seepage monitoring, thereby better assessing the stability of the dam and solving the problems of low monitoring accuracy and slow response in the existing technology.

[0008] To achieve the above objectives, the technical solution of the present invention is as follows:

[0009] It includes two aspects, the first aspect:

[0010] A sealed pressure measuring hole device with a replaceable sensor cavity includes a first-order sleeve, which is fixedly installed in a corridor, and the first-order sleeve, a second-order sleeve, and a multi-port pipe are connected in sequence.

[0011] The first-order sleeve is a hollow tubular structure with a pressure measuring hole inside; the outer wall of the front end of the first-order sleeve is threaded, the end of the thread is wrapped with geotextile for filtration, and a seepage-proof sealing structure is provided behind the thread; the tail end of the first-order sleeve is provided with a front flange, the front flange is provided with a screw hole, and a gap for bolt fixing is left between the seepage-proof sealing structure and the front flange.

[0012] The second-order sleeve is a hollow tubular structure with a rear flange fixed in the middle. The rear flange divides the tube body of the second-order sleeve into a front guide tube and a sensing tail tube. The outer diameter of the front guide tube is smaller than the inner diameter of the pressure measuring hole, and it is embedded in the pressure measuring hole to connect the first-order sleeve and the second-order sleeve. The rear flange has screw holes, and the screw holes correspond one-to-one with the screw holes on the front flange. The first-order sleeve and the second-order sleeve are fastened together by bolts and nuts passing through the corresponding screw holes.

[0013] The sensing tail tube of the two-stage bushing is sealed and connected to the multi-port pipe. In addition to the connection port with the sensing tail tube, the multi-port pipe has at least two interfaces, at least one of which is fitted with a sealed wire-passing plug. The remaining interfaces are selectively connected to external monitoring instruments or fitted with ordinary sealed plugs. The sealed wire-passing plug has a sealed wire-passing interface. The monitoring sensor installed inside the sensing tail tube has its data transmission line connected to a data cable. The data cable is sealed and passes through the sealed wire-passing interface and is led out through the sealed wire-passing interface to connect to an external data processing platform.

[0014] Furthermore, the external monitoring instruments include an external pressure gauge and a flow meter; the external pressure gauge is connected to a reserved interface of the multi-port pipe for monitoring the internal seepage pressure of the pipeline; the flow meter is connected in sequence through a connecting rod, a valve and the multi-port pipe for measuring the amount of water leakage in the pipeline; the monitoring sensors include an osmometer, a temperature sensor and a water quality sensor.

[0015] Furthermore, the multi-port pipe has a four-way pipe structure; of the four ports of the multi-port pipe, two ports are respectively sealed and connected to the second-order sleeve and the connecting rod, one port is connected to the external monitoring instrument, and the other port is equipped with the sealing wire plug.

[0016] Furthermore, an anti-seepage sealing ring is provided on the side of the anti-seepage sealing structure near the thread, and a sealing gasket is installed between the mating surfaces of the front flange and the rear flange.

[0017] Furthermore, the outer wall of the pre-conduit is wrapped with a filter geotextile, forming a double filtration structure with the geotextile at the threaded end of the first-order sleeve. The multi-port pipe is equipped with a built-in filter screen to filter impurities in the water flow, ensuring the cleanliness of the inside of the pressure measuring hole system and preventing impurities from affecting the accuracy of the osmotic pressure measurement.

[0018] The second aspect:

[0019] A method for monitoring a corridor using a sealed pressure measuring port device with a replaceable sensor cavity includes the following steps:

[0020] Step S1, Monitoring data acquisition and preprocessing:

[0021] Utilizing a multi-sensor system integrated into the sealed pressure measuring device, long-term continuous synchronous data acquisition is performed, collecting osmotic pressure data P(t), water temperature data T(t), water quality data W(t), and flow rate data Q(t). Consistency verification and compensation calibration of the sensor data are then performed, including:

[0022] The permeation pressure P(t) is calibrated in place using an external pressure gauge connected to a multi-port pipe, and flow correction is performed by combining valve opening and casing pipe diameter structural parameters; temperature compensation is performed on the pressure data based on water temperature data collected by a temperature sensor to calibrate the pressure P. cal(t);

[0023] And for P cal Missing values ​​were removed, time alignment was performed, and normalization was applied to Q(t), W(t).

[0024] Step S2, Baseline Switching and Drift Identification:

[0025] When the monitoring sensor is replaced or reinstalled, baseline parameters are established for the new sensor to complete the pressure sequence baseline alignment.

[0026] During operation, statistical analysis of relevant difference sequences can determine sensor zero-point drift or baseline jump, thereby increasing the acquisition frequency or prompting maintenance; enabling data availability and comparability to be quickly restored without prolonged downtime after sensor replacement.

[0027] Step S3, EMD decomposition and PCA feature fusion:

[0028] Using the calibrated or aligned pressure sequence as the main sequence, and introducing auxiliary sequences of water temperature, flow rate, and water quality, a multi-dimensional monitoring sequence is constructed.

[0029] Empirical Mode Decomposition (EMD) is performed on the pressure sequence to decompose it into K intrinsic mode function (IMF) components and residual terms, thus separating the high-frequency disturbances and low-frequency trend terms in the non-stationary signal.

[0030] Extract the EMD component features, including the energy, main period, amplitude statistics, and trend term change rate of each IMF, and combine them with the synchronous features of temperature, flow rate, and water quality to form the feature vector X(t);

[0031] Principal component analysis (PCA) is used to reduce the dimensionality and remove redundancy from the eigenvector X(t) to obtain a low-dimensional comprehensive feature Z(t). This integrates the linkage information of "osmotic pressure-temperature-flow rate-water quality" at the feature level, enhancing the adaptability to seasonal temperature influences and changes in operating conditions.

[0032] Step S4, LSTM time series prediction modeling and online update:

[0033] Using the low-dimensional comprehensive feature Z(t) as input, an LSTM prediction model is established to predict the infiltration pressure at future times and obtain the predicted value.

[0034] A sliding time window is used for model training and online updates: Under long-term continuous monitoring conditions, the model parameters are periodically retrained or fine-tuned using the latest data; when monitoring sensor replacement or baseline switching events occur, after baseline alignment is completed, short window data is used to quickly and adaptively update the model to ensure the continuity of predictions after the replacement.

[0035] Calculate the predicted residual e(t) = P cal (t)-Ppred (t) or P base (t)-P pred (t), and use the residual sequence as the basis for subsequent adaptive estimation of risk threshold;

[0036] Step S5: Obtain dynamic risk thresholds and graded early warnings using the confidence interval method.

[0037] Within the sliding time window, the residual sequence e(t) is statistically analyzed, and the residual standard deviation σ is calculated. e And construct the prediction confidence interval according to the set 1-α confidence level, the calculation formula is as follows:

[0038] P upper (t)=P pred (t)+z （1-α / 2） ·σ e

[0039] P lower (t)=P pred (t)-z （1-α / 2） ·σ e

[0040] Among them, z （1-α / 2） P represents the quantile of the standard normal distribution. upper (t) represents the upper limit of the predicted confidence interval pressure, P lower (t) Predict the lower limit of the confidence interval pressure;

[0041] When the calibrated or aligned pressure exceeds the upper or lower limit of the predicted confidence interval, it is considered an abnormal deviation; a tiered warning system is set based on the magnitude and duration of the deviation: deviation exceeding 1 times σ e Level 1 warning, exceeding 2 times σ e Level II warning, exceeding 3 times σ e This may be accompanied by an abnormal increase in flow rate Q(t) or a sudden change in water quality W(t), triggering a Level III warning.

[0042] The trend prediction output by the linkage step S4 uses "deviation between predicted value and actual measurement + acceleration of predicted trend" to suppress false alarms caused by single-point spikes, reduce the false alarm rate and improve the ability to identify sudden leakage conditions in advance.

[0043] Step S6, Long-term monitoring and model-data collaborative maintenance:

[0044] Establish a dual evaluation mechanism for data quality and model quality: assess the completeness, accuracy, and consistency of raw and calibration data; and evaluate the rolling error of the prediction model.

[0045] When the preset trigger conditions are met, perform in-situ verification, mark the data, and correct the data.

[0046] Missing data is imputed using time-aligned interpolation, and abnormal segments are marked and isolated to ensure the quality of training samples and avoid model drift caused by abnormal data.

[0047] Step S7, Risk Assessment and Decision Support:

[0048] A mechanism for quantitatively assessing and classifying leakage risks, and generating and pushing out explanations and reports;

[0049] The system integrates multi-dimensional monitoring data to quantitatively assess and classify leakage risks, and clarifies the criteria for determining risk escalation; it outputs information related to abnormal events and collaborative evidence, distinguishing between different types of disturbances and the development of leakage channels; and it regularly generates comprehensive seepage monitoring reports and pushes them for archiving, providing support for dam safety operation and maintenance and reinforcement decisions.

[0050] Furthermore, in step S1, the in-situ verification involves using an external pressure gauge connected to a multi-port pipe to verify the osmotic pressure P(t) and obtain the reference pressure P. ref (t), calculate the consistency deviation ΔP(t) = P(t) - P ref (t); When |ΔP(t)| exceeds the preset allowable deviation, zero-point or proportional correction is selected or combined to correct the pressure data. If the correction still does not meet the verification requirements, the pressure data is marked as abnormal.

[0051] The flow correction is completed by combining the valve opening degree and the casing diameter structural parameters. The flow data is controlled by the valve to control the on and off, and Q(t) is collected under the flow measurement condition with the valve open.

[0052] The pressure data is temperature-compensated based on water temperature data collected by a temperature sensor to eliminate the influence of water temperature changes on the measurement accuracy of the osmometer. A linear regression model is used for modeling, and the calculation formula is as follows:

[0053] P cal (t)=P mea (t)+k×(T(t)-T0)

[0054] Among them, P cal (t) represents the osmotic pressure after calibration, in kPa; P mea (t) represents the measured osmotic pressure in kPa; k is the temperature correction factor in kPa / ℃; T(t) represents the measured water temperature in ℃; T0 is the standard reference temperature, with a value of 20℃.

[0055] Furthermore, in step S2, the establishment of baseline parameters completes the pressure sequence baseline alignment. When the monitoring sensor in the sensing tailpipe of the second-sequence bushing is replaced or reinstalled, a quick reset is achieved using the detachable structure of the device flange and the sealing wire plug. Within the stable operating window after the system resumes operation, the reference value P from the external pressure gauge is used.ref (t) Establish baseline parameters for the new sensor and calculate the baseline bias ΔP0 = mean(P(t) - P ref (t) and baseline alignment of the pressure sequence: P base (t)=P(t)-ΔP0;

[0056] The formula for calculating the difference sequence is as follows, based on the statistics of the relevant difference sequences:

[0057] ΔP(t)=P base (t)-P ref (t).

[0058] Furthermore, in step S3, the Empirical Mode Decomposition (EMD) satisfies the following calculation formula:

[0059] P base (t)=∑(k=1…K)c k (t)+r(t)

[0060] Among them, c k (t) represents the k-th order intrinsic mode function (IMF) component; r(t) represents the residual term after EMD decomposition; K represents the effective number of IMFs obtained after EMD decomposition; P base (t) represents the aligned pressure sequence;

[0061] Principal component analysis (PCA) was used, and the calculation formula is as follows:

[0062] Z(t) = U T ·(X(t)-μ)

[0063] Where X(t) is the combined eigenvector at time t, μ is the mean of the eigenvectors, U is the principal component loading matrix selected in descending order of contribution rate, and Z(t) is the principal component eigenvector obtained after PCA dimensionality reduction.

[0064] Furthermore, in step S6, the preset triggering condition is:

[0065] (1) The in-situ verification deviation |ΔP(t)| continuously exceeds the preset allowable deviation threshold δ and the duration exceeds the preset duration τ;

[0066] (2) Predicted residual e(t) = P base (t)-P pred (t) Standard deviation σ within the sliding time window e Exceeding the preset threshold;

[0067] (3) The flow rate Q(t) changes abruptly when the valve switches from closed to open or when the valve opening changes.

[0068] The interpolation padding is calculated using the following formula:

[0069] P fill =P i-1 +(P i+1 -P i-1 ) / (t i+1 -t i-1 )×(t fill -t i-1 );

[0070] Among them, P fill To supplement the data, t fill t represents the collection time for missing data. i-1 t i+1 This represents the acquisition time of adjacent valid data.

[0071] The advantages of this invention compared to traditional technologies are as follows: This invention solves the problems of poor sealing, difficult maintenance, and data gaps in traditional monitoring by integrating a sealing structure and modular design. It further realizes real-time monitoring of seepage pressure on the sidewall of earth-rock dam gallery, integrates multi-parameter monitoring and analysis such as flow rate, improves the accuracy, continuity and intelligence of dam seepage monitoring, and has strong engineering adaptability.

[0072] 1. Convenient operation and maintenance with continuous data: The split structure eliminates the need for overall pipe removal and maintenance. If a sensor fails, only the secondary sleeve needs to be replaced. Monitoring is uninterrupted throughout the process, reducing operation and maintenance costs and time, and avoiding secondary disturbances to the dam body.

[0073] 2. Precise and efficient multi-parameter monitoring: It integrates multiple sensors for seepage pressure, temperature, flow rate, and water quality, and uses multi-parameter linkage compensation to eliminate interference. Combined with cross-analysis of flow rate and pressure, it can accurately determine the leakage channel and type, and quickly capture sudden leakage signals.

[0074] 3. Excellent integration and wide adaptability: Multifunctional pipe fittings allow for flexible switching between pressure and flow measurement modes without the need for separate equipment; standardized sealing structure prevents water seepage and sand ingress; pre-embedded sleeves are compatible with various dam bodies and new construction and reinforcement projects, simplifying construction and preventing interlayer flow.

[0075] 4. Dual-mode update for accurate monitoring: By updating online in two modes, the model mismatch problem caused by sensor replacement and baseline drift is overcome, and the robustness of the prediction model in long-term continuous monitoring scenarios is improved.

[0076] 5. Data-driven scientific decision-making: Integrating multiple indicators such as flow rate and pressure to conduct quantitative risk assessment, verifying the effectiveness of seepage prevention and reinforcement, and equipped with automatic data processing and report generation functions, providing a scientific basis for dam safety operation and maintenance and decision-making. Attached Figure Description

[0077] Figure 1 This is a structural diagram of the sealing pressure measuring hole device in this invention;

[0078] Figure 2 This is a schematic diagram of a sequential sleeve in this invention;

[0079] Figure 3 This is a cross-sectional view of the first-order sleeve in this invention;

[0080] Figure 4 This is a schematic diagram of the two-stage sleeve in this invention;

[0081] Figure 5 This is a schematic diagram of the bolt and nut in this invention;

[0082] Figure 6 This is a schematic diagram of the connecting rod in this invention;

[0083] Figure 7 This is a flowchart of the corridor monitoring method in this invention.

[0084] Among them, 1-corridor; 2-first-order sleeve; 201-proof sealing structure; 202-front flange; 203-pressure measuring hole; 204-screw hole; 205-thread; 206-proof sealing ring; 3-second-order sleeve; 301-rear flange; 302-bolt; 303-nut; 304-sensor tailpipe; 305-screw hole; 306-front guide pipe; 4-multi-port pipe; 5-connecting rod; 6-external pressure gauge; 7-sealing wire plug; 8-data cable; 9-valve. Detailed Implementation

[0085] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments disclosed herein will be described in further detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0086] Example 1:

[0087] This embodiment provides a sealed pressure measuring port device with a replaceable sensor cavity, such as... Figures 1-6 As shown, a first-order sleeve 2 is pre-embedded during the pouring of corridor 1, or it is installed by drilling and filling after the corridor is formed; the gap between the first-order sleeve 2 and the wall of the corridor 1 is filled with water-stopping and seepage-proof materials, including cement-based grouting materials, polymer grouting materials, and composite filling materials. Ordinary silicate cement grout is preferred. Low-pressure grouting is used to ensure that the grout body is dense and to block the seepage flow channel.

[0088] The first-order sleeve 2 is a hollow tubular structure with a through pressure measuring hole 203 inside. Its front end outer wall is machined with threads 205, which can enhance the bonding strength with the water-stopping and seepage-proof material and prevent the sleeve from shifting. The front end of the first-order sleeve 2 (threaded front end) is pre-wrapped with permeable geotextile and fastened with stainless steel cable ties to prevent it from falling off. This can filter dam particles to avoid clogging the pressure measuring hole and ensure smooth water seepage.

[0089] The seepage-proof sealing structure 201 of the first-order sleeve 2 is installed close to the outer wall of the corridor 1, and a seepage-proof sealing ring 206 is pre-installed on the side near the thread 205. The sealing ring is tightly fitted to the corridor wall by grouting pressure or installation fastening force to form a seepage barrier. The thread 205 section of the first-order sleeve 2 is completely buried in the hole of the corridor 1, and only the seepage-proof sealing structure 201 and the subsequent pipe body are exposed on the outer wall of the corridor.

[0090] The first sleeve 2 is equipped with a front flange 202 at the tail end. The front flange 202 has a bolt hole 204. A gap (greater than 3cm) is left between the anti-seepage sealing structure 201 and the front flange 202 to serve as the operating gap for bolt fixing.

[0091] The secondary sleeve 3 is a one-piece metal pipe with a rear flange 301 fixed in its middle. The rear flange 301 divides the pipe body of the secondary sleeve 3 into two parts: a front guide tube 306 and a sensing tail tube 304. The outer diameter of the front guide tube 306 is smaller than the inner diameter of the pressure measuring hole 203 inside the primary sleeve 2, forming a clearance fit structure. The outer wall of the front guide tube 306 is firmly wrapped with a corrosion-resistant and highly permeable geotextile, forming a second filtration structure. During installation, the front guide tube 306 is directly embedded into the pressure testing hole 203, forming a coaxial sealed connection structure between the first-order sleeve 2 and the second-order sleeve 3, ensuring smooth water flow. Combined with the geotextile at the front end of the first-order sleeve 2, it achieves dual filtration, effectively intercepting soil impurities of different particle sizes and preventing clogging of the monitoring equipment. The rear flange 301 has screw holes 305, which correspond one-to-one with the screw holes 204 on the front flange 202 at the tail end of the first-order sleeve 2. The front flange 202 and the rear flange 301... A sealing gasket is installed between the mating surfaces of the flanges 1. The sealing gasket is used to squeeze and seal to eliminate the flange mating gap, preventing external water seepage or internal water leakage. The first-order sleeve 2 and the second-order sleeve 3 are fastened together by bolts 302 passing through bolt holes 204 and 305 and nuts 303, achieving a rigid sealing connection between the two sleeves. An osmotic pressure gauge, a temperature sensor and a water quality sensor are fixedly installed inside the sensing tail tube 304. The sensors are fixed to the inner wall of the sensing tail tube 304 by brackets and are evenly distributed according to the monitoring requirements.

[0092] The end of the second-order sleeve 3 furthest from the first-order sleeve 2 is threadedly connected to the multi-way pipe 4 and internally connected. In this embodiment, the multi-way pipe 4 adopts a four-way pipe structure. Two ports of the four-way pipe are respectively sealed and connected to the second-order sleeve 3 and the connecting rod 5 to ensure smooth water flow in the pipeline. The other two ports are connected to an external pressure gauge 6 to monitor the seepage pressure data inside the pipeline in real time. The other port is equipped with a sealed cable plug 7. The data cable 8 of the sensor inside the second-order sleeve 3 passes through the sealed interface reserved on the plug 7 to achieve sealed passage and protection of the data cable 8. The far end of the data cable 8 is connected to an external data processing platform to complete the transmission and analysis of monitoring data. The end of the connecting rod 5 furthest from the multi-way pipe 4 is connected in series with a valve 9 and then connected to a flow meter. The valve 9 controls the opening and closing of the monitoring pipeline to meet the operational needs of maintenance, debugging, and selection of the measurement flow rate.

[0093] All external interfaces of the multi-port pipe 4 adopt a sealing method of threaded seal + sealing gasket.

[0094] In one embodiment of the present invention, the screw holes 204 are evenly distributed along the circumference of the front flange 202, and there are 4 of them, which correspond to the positions of the screw holes 305 on the rear flange 301.

[0095] In one embodiment of the present invention, the multi-port pipe 4 adopts a structure of two tee fittings joined together. The interfaces of this combined multi-port pipe 4 are functionally allocated: one is sealed and connected to the second-order sleeve 3, and the other is sealed and connected to the connecting rod 5; the remaining two sets of independent interfaces, one set of interfaces is connected to the external pressure gauge 6, and the other set of interfaces is equipped with a sealed cable plug 7, through which the data cable of the sensor inside the second-order sleeve 3 passes through the sealed interface reserved on the plug 7.

[0096] In one embodiment of the present invention, the bodies of the first-order sleeve 2 and the second-order sleeve 3 are made of stainless steel, alloy or other corrosion-resistant materials to ensure the long-term durability of the device in harsh environments; the seepage-proof sealing structure 201 adopts a double-layer sealing design, with a high-strength wear-resistant sealing gasket on the outer layer and a corrosion-resistant rubber material on the inner layer, and the seepage-proof sealing structure 201 also includes a heat-resistant material interlayer between the double-layer sealing structures, which can effectively isolate the influence of high-temperature environment and ensure the stable operation of the device under extreme temperature conditions.

[0097] In one embodiment of the present invention, the pressure gauge interface between the sensing tail tube 304 and the multi-port tube 4 adopts an integrated welding / threaded connection design, which facilitates installation and maintenance, while ensuring the long-term stability and accuracy of the measurement data.

[0098] In one embodiment of the present invention, a snap-fit ​​filter screen is built into the interface end of the multi-port pipe 4 near the second-order sleeve 3 to filter impurities in the water flow, ensure the cleanliness of the inside of the pressure measuring device, and avoid impurities affecting the accuracy of the osmotic pressure measurement.

[0099] In one embodiment of the present invention, the osmometer is an electronic pressure sensor, which is connected to an external data processing platform via a wired connection through an interface on the connecting rod 5 to support real-time data transmission; the connecting rod 5 has quick connection and disassembly functions, which facilitates sensor replacement and system maintenance, and is compatible with various monitoring data transmission methods, thereby improving system compatibility and maintenance efficiency.

[0100] Example 2:

[0101] This embodiment provides an installation method for a sealed pressure measuring hole device with a replaceable sensor cavity, based on the sealed pressure measuring hole device with a replaceable sensor cavity described in Embodiment 1, including the following steps:

[0102] Step 1, Preparation:

[0103] 1.1 Select an installation location on the side wall of the dam gallery, determine the installation depth and measurement point location, and ensure that the device can cover the key area for seepage pressure monitoring;

[0104] 1.2 Prepare the necessary equipment and materials for installation, including first-order sleeve 2, second-order sleeve 3, multi-port pipe 4, connecting rod 5, osmotic pressure gauge, temperature sensor, water quality sensor, pressure gauge, sealing ring 206, bolts, nuts, etc.

[0105] Step 2, Installing the first sleeve 2:

[0106] 2.1 Drill holes in the side wall of the dam gallery (or embed them in advance), with the hole diameter matching the outer diameter of the first-order sleeve 2, to ensure the stability and sealing of the installation hole.

[0107] 2.2 Insert the first-order sleeve 2 vertically into the predetermined hole, so that the threaded section 205 is completely embedded in the hole, with only the anti-seepage sealing structure 201 and the subsequent pipe body exposed on the outer wall of the corridor. Low-pressure grouting is used to ensure that the grout body is dense, to block the seepage flow channel, and to ensure the water tightness between the first-order sleeve 2 and the concrete structure.

[0108] 2.3 Check whether the seepage prevention and sealing structure 201 on the outer wall of the first-order sleeve 2 is intact, and ensure that the seepage prevention and sealing ring 206 is tightly fitted to the corridor wall without displacement or damage, to ensure sealing performance.

[0109] Step 3, Installation of second-order sleeve 3:

[0110] 3.1 Install the second-order sleeve 3 inside the first-order sleeve 2 and connect it to the rear flange 301 through the screw hole 305.

[0111] 3.2 Tighten the bolts 302 and nuts 303 on the rear flange 301 to ensure that the second-order sleeve 3 is firmly fixed.

[0112] 3.3 Ensure that there are no foreign objects inside the sensing tail tube 304 of the second-order sleeve 3, and prepare for the installation of the sensor.

[0113] Step 4, install the multi-port pipe 4 and connecting rod 5:

[0114] 4.1 Connect the multi-port pipe 4 to the sensing tail pipe 304 end of the second-order sleeve 3 with a threaded seal, install a sealing gasket and tighten it to ensure a secure connection and internal sealing without leakage.

[0115] 4.2 Connect the designated interface of the multi-port pipe 4 to the external pressure gauge 6 with a threaded seal to ensure a secure and leak-free connection. The pressure gauge can monitor the internal permeation pressure data of the pipeline in real time.

[0116] 4.3 Connect the connecting rod 5 to the multi-port pipe 4, and ensure that the interface on the multi-port pipe 4 used to connect the monitoring test data cable is correctly connected for subsequent data acquisition and transmission.

[0117] Step 5, Installation of the osmometer and data transmission system:

[0118] 5.1 Install the osmometer, water quality sensor, and temperature sensor onto the bracket inside the sensing tailpipe 304 within the secondary sleeve 3, ensuring the sensors are securely installed and evenly distributed according to monitoring requirements; after consolidating the sensor data cables, pass them through the sealed interface of the sealed cable plug 7 on the multi-port pipe 4 to connect to the external data processing platform, ensuring the cable interface is sealed and leak-free.

[0119] 5.2 Debug the osmometer to ensure that it can monitor osmotic pressure data in real time and transmit data with the external data processing platform.

[0120] 5.3 Complete sensor connection and system testing to ensure the accuracy and real-time performance of data transmission.

[0121] Step 6, Inspection and installation of the built-in filter in multi-port pipe 4:

[0122] 6.1 Install a filter screen inside the multi-port pipe 4 to ensure that impurities in the water flow do not enter the pressure measurement system and affect the measurement accuracy.

[0123] 6.2 Check whether the filter screen is firmly fixed to ensure that it is not easily moved and can effectively filter impurities in the water flow.

[0124] Step 7, sealing and final inspection of the pressure testing system:

[0125] 7.1 Check whether the flange joint between the first-order sleeve 2 and the second-order sleeve 3 is tight and sealed, and ensure that the front flange 202 and the rear flange 301 at the end of the first-order sleeve 2 fit tightly, the bolts and nuts are not loose, and the gasket is not misaligned or damaged.

[0126] 7.2 After all components are installed, a final inspection is conducted to confirm that all sealing rings 206 and interface parts are installed correctly.

[0127] 7.3 Check whether the connection between the external pressure gauge and the multi-port pipe 4 is secure to ensure stable data transmission and normal operation of the device.

[0128] Example 3:

[0129] This embodiment provides a corridor monitoring method for a sealed pressure measuring hole device with a replaceable sensor cavity, such as... Figure 7 As shown, the sealed pressure measuring port device based on the replaceable sensor cavity described in Embodiment 1 includes the following steps:

[0130] Step S1, Monitoring data acquisition and preprocessing:

[0131] 1.1 Data Acquisition:

[0132] Data acquisition relies on the multi-sensor system integrated into the device and follows a process of "synchronous acquisition - real-time transmission - preliminary verification - in-situ verification" to ensure the integrity and availability of long-term continuous data. The specific steps are as follows:

[0133] 1.1.1 Data Acquisition Trigger: Set the acquisition frequency (once every 5 minutes under normal operating conditions, and once per minute during flood season or abnormal dam conditions) to trigger the osmotic pressure gauge, temperature sensor, and water quality sensor to synchronously acquire osmotic pressure P(t), water temperature T(t), and water quality W(t); open valve 9 as needed to enter the flow measurement condition to acquire flow rate Q(t), avoiding time sequence deviations from affecting the accuracy of subsequent feature fusion and prediction modeling.

[0134] 1.1.2 Real-time data transmission: The collected data is converted into digital signals by the sensor module and transmitted to the terminal acquisition unit through a sealed cable. Anti-interference coding is adopted to reduce the impact of the electromagnetic environment of the corridor on the data.

[0135] 1.1.3 Preliminary validity verification: The terminal unit performs range verification and mutation verification on the data, removes obviously invalid values, marks interrupted data as abnormal and triggers re-collection, and stores the original data and verification log for easy traceability.

[0136] 1.1.4 In-situ consistency verification: Obtain the reference pressure P by connecting the multi-port pipe 4 to the external pressure gauge 6. ref Given the osmotic pressure P(t) collected by the osmometer, the consistency deviation ΔP(t) is calculated using the following formula:

[0137] ΔP(t) = P(t) - P ref (t) (1)

[0138] When |ΔP(t)| exceeds the preset allowable deviation threshold δ, the pressure data is zeroed / proportionally corrected or marked as abnormal, providing reliable input for subsequent modeling. This step utilizes the device's "multi-port pipe + external pressure gauge" structure to achieve online verification, which can reduce false alarms and missed alarms caused by single sensor drift.

[0139] 1.2 Data Compensation Calibration and Standardization Preprocessing:

[0140] To address errors in multi-sensor systems, compensation calibration and sequence preprocessing are performed to generate input data that can be used for prediction models.

[0141] 1.2.1 Pressure Data Temperature Compensation: Based on water temperature data collected by a temperature sensor, the influence of water temperature changes on the measurement accuracy of the osmometer is eliminated. A linear regression model is used for modeling, and the calculation formula is as follows:

[0142] P cal (t)=P mea (t)+k×(T(t)-T0) (2)

[0143] Among them, P cal (t) represents the osmotic pressure (kPa) after calibration, P mea (t) represents the measured osmotic pressure (kPa), k represents the temperature correction factor (kPa / ℃), T(t) represents the measured water temperature (℃), and T0 represents the standard reference temperature (preferably 20℃).

[0144] 1.2.2 Flow data correction: Flow data is controlled by valve 9. Q(t) is collected under the flow measurement condition with the valve open, and the flow is corrected by combining the valve opening degree, sleeve diameter and other structural parameters to obtain Qcal(t), which is used for linkage analysis and modeling input with pressure-water temperature-water quality.

[0145] 1.2.3 Time Alignment and Normalization: For P cal (t), Q cal Missing values ​​were removed from W(t), time alignment was performed, and normalization was performed to form a multi-dimensional monitoring sequence, providing a unified data benchmark for subsequent empirical mode decomposition (EMD), principal component analysis (PCA) dimensionality reduction, and long short-term memory network (LSTM) prediction.

[0146] Step S2, Baseline Switching and Drift Identification:

[0147] 2.1 Baseline Establishment and Alignment: When the monitoring sensor inside the sensing tailpipe 304 of the second-order sleeve 3 is replaced or reinstalled, a quick reset is achieved using the detachable flange structure and the sealing wiring plug; within the stable operating window of the system after resumption of operation (valve closed or maintaining a fixed opening, with minimal flow variation), the reference value P of the external pressure gauge is used. ref(t) Establish baseline parameters for the new sensor and calculate the baseline bias ΔP0 using the following formula:

[0148] ΔP0=mean(P(t)-P ref (t)) (3)

[0149] The pressure sequence was then baseline aligned to obtain the baseline-aligned pressure sequence P. base (t), the calculation formula is as follows:

[0150] P base (t)=P(t)-ΔP0 (4)

[0151] 2.2 Drift Identification and Triggering Strategy: During operation, the difference sequence ΔP is analyzed. base (t)=P base (t)-P ref (t) is used for sliding statistics; when ΔP base (t) When the data continuously exceeds the limit or shows a continuous unidirectional change (drift) within the preset time window, it is determined to be a zero-point drift or baseline jump, triggering verification encryption (increasing the acquisition frequency) or prompting maintenance; this mechanism enables the data comparability to be quickly restored after the sensor is replaced, avoiding threshold misjudgment and long-term data gap caused by replacement.

[0152] Step S3, EMD decomposition and PCA feature fusion:

[0153] 3.1 Multidimensional sequence construction: Using the calibration pressure sequence P obtained in step S1 cal (t) or P after baseline alignment in step S2 base (t) is used as the main sequence, and water temperature T(t) and flow rate Q are introduced. cal Auxiliary sequences such as W(t) and water quality W(t) are used to construct a multi-dimensional monitoring sequence.

[0154] 3.2 EMD Decomposition: Empirical Mode Decomposition (EMD) is performed on the pressure sequence to decompose it into several intrinsic mode function (IMF) components c1(t)...c K (t) and the residual term r(t) are used to separate the high-frequency disturbance and low-frequency trend term in the non-stationary signal, satisfying:

[0155] P base (t)=∑(k=1…K)c k (t)+r(t) (5)

[0156] 3.3 Feature Extraction and Fusion: Extract EMD component features (including energy, principal period, mean and variance of amplitude, kurtosis, skewness, residual term trend slope, etc. of each IMF), and fuse them with water temperature T(t) and flow rate Q. calThe synchronous characteristics of water quality W(t) (including temperature change rate, flow rate change rate, water quality index statistics, etc.) are combined to form a feature vector X(t) to characterize the linkage state of "osmotic pressure-temperature-flow-water quality" and enhance the ability to identify seasonal temperature disturbances and operating condition switching disturbances.

[0157] 3.4 PCA Dimensionality Reduction: Principal Component Analysis (PCA) is used to reduce the dimensionality and remove redundancy from the eigenvector X(t) to obtain the low-dimensional comprehensive feature Z(t). The calculation formula is as follows:

[0158] Z(t) = U T ·(X(t)-μ) (6)

[0159] Where μ is the mean of the feature vectors and U is the principal component load matrix selected in descending order of contribution rate; PCA compresses multi-source features into a small number of principal components, reducing redundancy and noise propagation, and improving the generalization ability and stability of subsequent prediction models.

[0160] Step S4, LSTM time series prediction modeling and online update:

[0161] 4.1 Predictive Modeling: Using the low-dimensional comprehensive features Z(t) obtained in step S3, an input sequence is constructed, and an LSTM time series prediction model is established to predict the permeation pressure at future times, outputting the predicted value P. pred (t+Δt), where Δt is the prediction step size (preferably consistent with the acquisition period or an integer multiple thereof).

[0162] 4.2 Online Update: A sliding time window is used for model training and online updates. Under long-term continuous monitoring conditions, the model parameters are periodically retrained or fine-tuned using the latest samples. When a sensor replacement or baseline switching event occurs in step S2, after baseline alignment is completed and verified by the external pressure gauge 6, a short-window sample is used to quickly and adaptively update the model, so that the prediction model is synchronized with the new baseline state, ensuring the continuity and comparability of predictions after the replacement.

[0163] By using dual-mode online updates, the model mismatch problem caused by sensor replacement and baseline drift is overcome, and the robustness of the prediction model in long-term continuous monitoring scenarios is improved.

[0164] 4.3 Residual Calculation: Calculate the predicted residual e(t), which serves as the basis for subsequent dynamic threshold estimation and risk assessment. The calculation formula is as follows:

[0165] e(t) = P base (t)-P pred (t) (7)

[0166] Among them, P base(t) represents the pressure data after baseline alignment; when no sensor replacement occurs and baseline alignment is not required, P base (t) can be obtained from P cal (t) substitution.

[0167] Step S5: Obtain dynamic risk thresholds and graded early warnings using the confidence interval method.

[0168] 5.1 Construction of dynamic confidence intervals: Statistical analysis of the residual sequence e(t) is performed within the sliding time window to calculate the standard deviation of the residuals. And construct the upper and lower limits of the prediction confidence interval according to the set confidence level (1-α), the calculation formula is as follows:

[0169] P upper (t)=P pred (t)+z （1-α / 2） ·σ e (8)

[0170] P lower (t)=P pred (t)-z （1-α / 2） ·σ e (9)

[0171] Among them, z （1-α / 2） These are the quantiles of the standard normal distribution.

[0172] 5.2 Over-limit detection and graded early warning: When the measured pressure P base (t) exceeds P upper (t) or below P lower When (t) is reached, it is determined to be an abnormal deviation; a tiered warning is set according to the deviation magnitude and duration: deviation exceeding 1 time. Level 1 warning, more than twice Level II warning, more than 3 times A level three warning is issued; when an abnormal deviation is accompanied by a flow rate Q cal When W(t) increases abnormally or water quality W(t) changes abruptly, the warning level is raised or the trigger duration threshold is shortened to reduce false alarms and missed alarms by using multi-parameter cross-validation.

[0173] 5.3 Suppress false alarms due to spurs: Simultaneously link the predicted trend output in step S4. When the "predicted-measured deviation" does not continuously reach the preset number of times or the predicted trend does not show a continuous increase, the abnormality is judged as a short-term disturbance and recorded, without triggering a high-level warning. This suppresses false alarms caused by single-point spurs and sensor transient jitter, and improves the ability to identify sudden leakage conditions in advance.

[0174] Step S6, Long-term monitoring and model-data collaborative maintenance:

[0175] 6.1 Dual Evaluation of Data Quality and Model Quality: A dual evaluation mechanism for data quality and model quality is established during long-term continuous monitoring. This includes assessing the completeness, accuracy, and consistency of raw and calibration data; and evaluating the rolling error of the prediction model, including the mean residual and σ. e Indicators such as stability and over-limit rate are used to determine whether the model has drifted.

[0176] 6.2 In-situ verification triggering and linkage maintenance: Obtain the reference pressure P by connecting the external pressure gauge 6 to the multi-port pipe 4. ref (t), calculate the in-situ verification deviation ΔP base (t)=P base (t)-P ref (t); When any of the following triggering conditions are met, perform in-situ verification and mark and correct the data:

[0177] (1) |ΔP base (t)|Continuously exceeds the preset allowable deviation threshold δ and the duration exceeds the preset duration τ;

[0178] (2) Standard deviation of the residual sequence e(t) within the sliding time window Exceeding the preset threshold;

[0179] (3) When valve 9 switches from closed to open for flow measurement or when the valve opening changes, the flow rate Q is affected. cal (t) exhibits a step change;

[0180] After triggering in-place verification, if an abnormal segment is identified, the abnormal segment is marked and isolated to prevent abnormal data from entering the model training set and causing model drift.

[0181] 6.3 Missing Data Imputation: Missing data is imputed using time-aligned linear interpolation. The calculation formula is as follows:

[0182] P fill =P i-1 +(P i+1 -P i-1 ) / (t i+1 -t i-1 )×(t fill -t i-1 (10)

[0183] Among them, P fill To supplement the data, t fill t represents the collection time for missing data. i-1 t i+1 This represents the acquisition time of adjacent valid data.

[0184] Step S7, Risk Assessment and Decision Support:

[0185] 7.1 Risk Quantification and Classification: Comprehensive Measured Pressure P base (t), Predicted pressure P pred (t), confidence interval exceeding information, and flow rate Q cal The linkage between W(t) and water quality W(t) is used to quantitatively assess and classify the risk of leakage in the corridor; when there are continuous exceedances, an increase in the magnitude of exceedances and an increase in flow rate / sudden change in water quality, the risk of leakage is judged to be increased.

[0186] 7.2 Event Output and Interpretation: Output includes the time of anomaly occurrence, the extent of exceeding limits, the duration, the predicted trend (whether it is accelerating), and correlative evidence with flow / water quality (e.g., Q). cal (t) increases and W(t) turbidity or ion concentration changes, etc., are used to distinguish between "seasonal temperature fluctuations / short-term disturbances" and "leakage channel development".

[0187] 7.3 Report generation and delivery: Generate a comprehensive report on the corridor seepage monitoring regularly (preferably monthly), including data quality evaluation, model prediction effect, statistics of exceeding limits, risk level assessment and maintenance recommendations, and deliver it to the management personnel terminal and archive it to support the decision-making on safe operation, maintenance and reinforcement of the dam.

[0188] The above description is only used to illustrate the technical solutions of the present invention and is not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention (such as the application of various formulas, the order of steps, etc.) without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A sealed pressure measuring port device with a replaceable sensor cavity, characterized in that, Includes a first-order sleeve (2), which is fixedly installed in the corridor (1), and the first-order sleeve (2), second-order sleeve (3), and multi-way pipe (4) are connected in sequence; The first-order sleeve (2) is a hollow tubular structure with a pressure measuring hole (203) inside. The outer wall of the front end of the first-order sleeve (2) is threaded (205), and the end of the thread (205) is wrapped with geotextile for filtration. A seepage-proof sealing structure (201) is provided behind the thread (205). The tail end of the first-order sleeve (2) is provided with a front flange (202), and a first bolt hole (204) is opened on the front flange (202). A gap for bolt fixing is left between the seepage-proof sealing structure (201) and the front flange (202). The second-order sleeve (3) is a hollow tubular structure with a rear flange (301) fixed in the middle. The rear flange (301) divides the tube body of the second-order sleeve (3) into two parts: a front guide tube (306) and a sensing tail tube (304). The outer diameter of the front guide tube (306) is smaller than the inner diameter of the pressure measuring hole (203), and it is embedded in the pressure measuring hole (203) to connect the first-order sleeve (2) and the second-order sleeve (3). A second screw hole (305) is opened on the rear flange (301), and the second screw hole (305) corresponds one-to-one with the first screw hole (204) of the front flange (202). The first-order sleeve (2) and the second-order sleeve (3) are fastened together by bolts (302) passing through the corresponding screw holes and nuts (303). The sensing tail tube (304) of the two-stage sleeve (3) is sealed and connected to the multi-port tube (4); the multi-port tube (4) has at least two interfaces in addition to the connection port with the sensing tail tube (304), at least one of which is equipped with a sealed wire threading plug (7), and the remaining interfaces are selectively connected to external monitoring instruments or equipped with ordinary sealed plugs; the sealed wire threading plug (7) has a sealed wire threading interface, and the monitoring sensor installed inside the sensing tail tube (304) has its data transmission line connected to the data cable (8); the data cable (8) is sealed and threaded through the sealed wire threading interface and led out through the sealed wire threading interface to connect to the external data processing platform; The corridor monitoring method of the sealed pressure measuring hole device includes the following steps: Step S1, Monitoring data acquisition and preprocessing: Utilizing a multi-sensor system integrated into the sealed pressure measuring device, long-term continuous synchronous data acquisition is performed, collecting osmotic pressure data P(t), water temperature data T(t), water quality data W(t), and flow rate data Q(t). Consistency verification and compensation calibration of the sensor data are then performed, including: The permeation pressure P(t) is calibrated in place using an external pressure gauge connected to a multi-port pipe, and flow correction is performed by combining valve opening and casing pipe diameter structural parameters; temperature compensation is performed on the pressure data based on water temperature data collected by a temperature sensor to calibrate the pressure P. cal (t); And for P cal Missing values ​​were removed, time alignment was performed, and normalization was applied to Q(t), W(t). Step S2, Baseline Switching and Drift Identification: When the monitoring sensor is replaced or reinstalled, baseline parameters are established for the new sensor to complete the pressure sequence baseline alignment. During operation, statistical analysis of relevant difference sequences can determine sensor zero-point drift or baseline jump, thereby increasing the acquisition frequency or prompting maintenance; enabling data availability and comparability to be quickly restored without prolonged downtime after sensor replacement. Step S3, EMD decomposition and PCA feature fusion: Using the calibrated or aligned pressure sequence as the main sequence, and introducing auxiliary sequences of water temperature, flow rate, and water quality, a multi-dimensional monitoring sequence is constructed. Empirical Mode Decomposition (EMD) is performed on the pressure sequence to decompose it into K intrinsic mode function (IMF) components and residual terms, thus separating the high-frequency disturbances and low-frequency trend terms in the non-stationary signal. Extract the EMD component features, including the energy, main period, amplitude statistics, and trend term change rate of each IMF, and combine them with the synchronous features of temperature, flow rate, and water quality to form the feature vector X(t); Principal component analysis (PCA) is used to reduce the dimensionality and remove redundancy from the eigenvector X(t) to obtain a low-dimensional comprehensive feature Z(t). This integrates the linkage information of "osmotic pressure-temperature-flow rate-water quality" at the feature level, enhancing the adaptability to seasonal temperature influences and changes in operating conditions. Step S4, LSTM time series prediction modeling and online update: Using the low-dimensional comprehensive feature Z(t) as input, an LSTM prediction model is established to predict the infiltration pressure at future times and obtain the predicted value. A sliding time window is used for model training and online updates: Under long-term continuous monitoring conditions, the model parameters are periodically retrained or fine-tuned using the latest data; when monitoring sensor replacement or baseline switching events occur, after baseline alignment is completed, short window data is used to quickly and adaptively update the model to ensure the continuity of predictions after the replacement. Calculate the predicted residual e(t) = P cal (t)-P pred (t) or P base (t)-P pred (t), and use the residual sequence as the basis for subsequent adaptive estimation of risk threshold; Step S5: Obtain dynamic risk thresholds and graded early warnings using the confidence interval method. Within the sliding time window, the residual sequence e(t) is statistically analyzed, and the residual standard deviation σ is calculated. e And construct the prediction confidence interval according to the set 1-α confidence level, the calculation formula is as follows: P upper (t)=P pred (t)+z （1-α / 2） ·σ e P lower (t)=P pred (t)-z （1-α / 2） ·σ e Among them, z （1-α / 2） P represents the quantile of the standard normal distribution. upper (t) represents the upper limit of the predicted confidence interval pressure, P lower (t) Predict the lower limit of the confidence interval pressure; When the calibrated or aligned pressure exceeds the upper or lower limit of the predicted confidence interval, it is considered an abnormal deviation; a tiered warning system is set based on the magnitude and duration of the deviation: deviation exceeding 1 times σ e Level 1 warning, exceeding 2 times σ e Level II warning, exceeding 3 times σ e This may be accompanied by an abnormal increase in flow rate Q(t) or a sudden change in water quality W(t), triggering a Level III warning. The trend prediction output by the linkage step S4 uses "deviation between predicted value and actual measurement + acceleration of prediction trend" to suppress false alarms caused by single-point spikes, reduce the false alarm rate and improve the ability to identify sudden leakage conditions in advance. Step S6, Long-term monitoring and model-data collaborative maintenance: Establish a dual evaluation mechanism for data quality and model quality: assess the completeness, accuracy, and consistency of raw and calibration data; and evaluate the rolling error of the prediction model. When the preset trigger conditions are met, perform in-situ verification, mark the data, and correct the data. Missing data is imputed using time-aligned interpolation, and abnormal segments are marked and isolated to ensure the quality of training samples and avoid model drift caused by abnormal data. Step S7, Risk Assessment and Decision Support: A mechanism for quantitatively assessing and classifying leakage risks, and generating and pushing out explanations and reports; The system integrates multi-dimensional monitoring data to quantitatively assess and classify leakage risks, and clarifies the criteria for determining risk escalation; it outputs information related to abnormal events and collaborative evidence, distinguishing between different types of disturbances and the development of leakage channels; and it regularly generates comprehensive seepage monitoring reports and pushes them for archiving, providing support for dam safety operation and maintenance and reinforcement decisions.

2. The sealed pressure measuring port device with a replaceable sensor cavity according to claim 1, characterized in that, The external monitoring instruments include an external pressure gauge (6) and a flow meter; the external pressure gauge (6) is connected to the reserved interface of the multi-port pipe (4) for monitoring the internal permeation pressure of the pipeline; the flow meter is connected in sequence through the connecting rod (5), the valve (9) and the multi-port pipe (4) for measuring the amount of water leakage in the pipeline; the monitoring sensors include an osmotic pressure gauge, a temperature sensor and a water quality sensor.

3. The sealed pressure measuring port device with a replaceable sensor cavity according to claim 2, characterized in that, The multi-port pipe (4) is a four-way pipe structure; of the four ports of the multi-port pipe (4), two ports are respectively sealed and connected to the second-order sleeve (3) and the connecting rod (5), one port is connected to the external monitoring instrument, and the other port is equipped with the sealing wire plug (7).

4. The sealed pressure measuring port device with a replaceable sensor cavity according to claim 1, characterized in that, The seepage-proof sealing structure (201) is provided with a seepage-proof sealing ring (206) on the side near the thread (205), and a sealing gasket is installed between the mating surfaces of the front flange (202) and the rear flange (301).

5. The sealed pressure measuring port device with a replaceable sensor cavity according to claim 1, characterized in that, The outer wall of the pre-conduit pipe (306) is wrapped with a filter geotextile, which forms a double filter structure with the geotextile at the threaded end of the first-order sleeve (2). The multi-pass pipe (4) is equipped with a built-in filter screen to filter impurities in the water flow, ensuring the cleanliness of the inside of the pressure measuring hole system and avoiding impurities from affecting the accuracy of the osmotic pressure measurement.

6. The sealed pressure measuring port device with a replaceable sensor cavity according to claim 1, characterized in that, In step S1, the in-situ verification involves using an external pressure gauge connected to a multi-port pipe to verify the osmotic pressure P(t) and obtain the reference pressure P. ref (t), calculate the consistency deviation ΔP(t) = P(t) - P ref (t); When |ΔP(t)| exceeds the preset allowable deviation, zero-point or proportional correction is selected or combined to correct the pressure data. If the correction still does not meet the verification requirements, the pressure data is marked as abnormal. The flow correction is completed by combining the valve opening degree and the casing diameter structural parameters. The flow data is controlled by the valve to control the on and off, and Q(t) is collected under the flow measurement condition with the valve open. The pressure data is temperature-compensated based on water temperature data collected by a temperature sensor to eliminate the influence of water temperature changes on the measurement accuracy of the osmometer. A linear regression model is used for modeling, and the calculation formula is as follows: P cal (t)=P mea (t)+k×(T(t)-T0) Among them, P cal (t) represents the osmotic pressure after calibration, in kPa; P mea (t) represents the measured osmotic pressure in kPa; k is the temperature correction factor in kPa / ℃; T(t) represents the measured water temperature in ℃; T0 is the standard reference temperature, with a value of 20℃.

7. The sealed pressure measuring port device with a replaceable sensor cavity according to claim 1, characterized in that, In step S2, the establishment of baseline parameters completes the pressure sequence baseline alignment. When the monitoring sensor in the sensing tail pipe (304) of the second-order sleeve (3) is replaced or reinstalled, a quick reset is achieved using the detachable structure of the device flange and the sealing wire plug. Within the stable operating window of the system's recovery, the reference value P of the external pressure gauge is used. ref (t) Establish baseline parameters for the new sensor and calculate the baseline bias ΔP0 = mean(P(t) - P ref (t) and baseline alignment of the pressure sequence is performed, calculated as follows: P base (t)=P(t)-ΔP0 Among them, P base (t) represents the aligned pressure sequence; based on statistics of the relevant difference sequences, the formula for calculating the difference sequence is as follows: ΔP(t)=P base (t)-P ref (t)。 8. The sealed pressure measuring port device with a replaceable sensor cavity according to claim 1, characterized in that, In step S3, the empirical mode decomposition (EMD) satisfies the following calculation formula: P base (t)=∑(k=1…K)c k (t)+r(t) Among them, c k (t) represents the k-th order intrinsic mode function (IMF) component; r(t) represents the residual term after EMD decomposition; K represents the effective number of IMFs obtained after EMD decomposition; P base (t) represents the aligned pressure sequence; Principal component analysis (PCA) was used, and the calculation formula is as follows: Z(t)=U T ·(X(t)-μ) Where X(t) is the combined eigenvector at time t, μ is the mean of the eigenvectors, U is the principal component loading matrix selected in descending order of contribution rate, and Z(t) is the principal component eigenvector obtained after PCA dimensionality reduction.

9. The sealed pressure measuring port device with a replaceable sensor cavity according to claim 1, characterized in that, In step S6, the preset triggering condition is: (1) The in-situ verification deviation |ΔP(t)| continuously exceeds the preset allowable deviation threshold δ and the duration exceeds the preset duration τ; (2) Predicted residual e(t) = P base (t)-P pred (t) Standard deviation σ within the sliding time window e Exceeding the preset threshold; (3) When valve (9) switches from closed to open for flow measurement or when the valve opening changes, the flow rate Q(t) will change stepwise. The interpolation padding is calculated using the following formula: P fill =P i-1 +(P i+1 -P i-1 ) / (t i+1 -t i-1 )×(t fill -t i-1 ) Among them, P fill To supplement the data, t fill t represents the collection time for missing data. i-1 t i+1 This represents the acquisition time of adjacent valid data.

Images