A multi-dimensional feature fusion determination method and system for detecting the sealing performance of a water meter

Abstract

The application discloses a kind of multi-dimensional feature fusion determination method and system for water meter sealing detection, comprising: obtaining the nameplate information of the measured water meter, at least including water meter caliber and / or model information;According to the nameplate information, query the pre-stored detection resource configuration library, match the detection control parameter set corresponding to the measured water meter, at least including clamping pressure, test pressure, pressure holding time and dynamic determination threshold, clamp the measured water meter sealing;According to test pressure and pressure holding time, execute automatic exhaust, pressurize to target pressure, enter pressure holding detection stage;Collect the pressure-time variation data inside the measured water meter, extract multi-dimensional pressure decay characteristic parameters for representing the leakage state of water meter;Multi-dimensional pressure decay characteristic parameters are input into weighted fusion determination model, compare pressure decay characteristic parameters with corresponding dynamic determination threshold, output the sealing determination result of the measured water meter;Based on qualified sample data, update dynamic determination threshold, so that determination standard is matched with actual detection working condition.

Description

Technical Field

[0001] This invention relates to the field of water meter testing technology, and in particular to a multi-dimensional feature fusion judgment method and system for water meter sealing performance testing. Background Technology

[0002] As a key metering device in water resource billing, the sealing performance of water meters directly affects metering accuracy and fair trade. Leaks, spills, or damage to water meters not only lead to water waste and economic disputes but can also damage pipelines and meter structures due to long-term leakage, impacting water resource management efficiency. According to the national metrological verification regulation JJG162-2019 "Drinking Cold Water Meters," water meters must undergo sealing verification before being put into use. For large-diameter water meters (DN50~DN300), there is currently no dedicated sealing test equipment for large-diameter liquid flow metering instruments, both domestically and internationally. In particular, there is a lack of professional devices that meet the sealing test requirements for large-diameter water meters. Existing devices cannot perform sealing checks, necessitating the fabrication of simple sealing mechanisms, which are cumbersome and inefficient. Most of the publicly available water meter sealing testing devices are designed for small-diameter water meters and do not explicitly specify devices applicable to large-diameter ranges. Some devices use gas as the test medium, which is not suitable for the statutory verification of actual water meter sealing. In terms of the testing process, key parameters such as clamping pressure and test pressure must be manually set according to the water meter diameter, and the result judgment also depends on manual observation of the pressure gauge or visual inspection for bubbles and leaks. The degree of automation is low, the operation is cumbersome and subjective, and it is difficult to meet the needs of large-scale and high-efficiency testing.

[0003] Regarding water meter sealing testing methods, systems, and devices, the following comparative patents and documents exist: "An Ultrasonic Water Meter Sealing Test Method," patent publication number CN120232589A. This method uses ultrasound as a signal source to test the sealing performance of water meters. It utilizes the strong penetrating power of ultrasound to determine leaks without disassembling the water meter, and data analysis and recording can be performed using a computer system. However, the differences between this prior art and the present invention are: it uses an ultrasonic testing principle instead of a hydrostatic test, which does not meet the hydrostatic pressure test requirements specified in JJG162-2019; it does not involve a scheme for automatically matching detection control parameters based on the water meter diameter or model; its applicability is mainly for water meter pipe sections and does not specifically address the large diameter range of DN50 to DN300; its judgment method is based on ultrasonic signal analysis rather than the multi-dimensional pressure attenuation characteristic parameter fusion comparison used in the present invention, resulting in a relatively singular judgment dimension.

[0004] "A Water Meter Sealing Detection Device," patent number CN118168729A. This solution uses a pneumatic pressure detection method. When the pneumatic pressure shows no significant change, high-temperature gas is introduced into the water meter to evaporate the permeated water. A humidity sensor is used to determine if there is an increase in the humidity of the discharged gas, thus identifying a leak. The leak point can be located by gradually exposing the water meter to the water surface. However, the difference between this prior art and the present invention lies in the following: it uses an indirect detection method of "pneumatic pressure detection + humidity sensing" instead of a direct water pressure detection method. The detection process involves multiple steps, including pneumatic pressure detection, water injection, hot air drying, and humidity detection, resulting in a long detection cycle and low efficiency. Its automation level is low, and the detection process involves multiple manual or semi-automatic steps. Regarding parameter configuration, it does not involve steps for automatically matching clamping pressure, test pressure, and other control parameters according to different diameters; the parameters required for each step still need to be manually set. In terms of the judgment method, it uses a humidity sensor threshold for binary judgment, resulting in a single judgment dimension. In contrast, the present invention continuously collects pressure-time data during the pressure holding stage, extracts multi-dimensional pressure decay characteristics, and performs fusion comparison, achieving full-process automation and highly efficient judgment results.

[0005] "An Adaptive Clamping Device for Water Meter Testing," authorized publication number CN220398685U. This solution relates to an adaptive clamping device for water meter testing, which achieves clamping and fixing of water meters of different sizes through the cooperation of a clamping mechanism, a height adjustment mechanism, a moving mechanism, and a connecting mechanism. However, the difference between this prior art and the present invention is that: it is only a single clamping device in the water meter processing stage rather than a dedicated device for sealing testing, and does not involve complete testing process control and result judgment; it adopts a mechanical clamping structure rather than an electro-hydraulic method, and does not have the ability to automatically monitor and control pressure; it does not involve a parameter adaptive matching algorithm based on caliper identification and a multi-dimensional pressure attenuation feature fusion judgment algorithm, resulting in a low degree of automation.

[0006] In summary, the main technical problems in the current field of water meter sealing performance testing are as follows: First, existing testing methods mostly use ultrasonic testing or air pressure testing, rather than the hydrostatic pressure test method specified in JJG162-2019, resulting in deviations in the testing principle; second, existing solutions are not specifically designed for the characteristics of large-diameter water meters (DN50~DN300), such as large weight, the need for mechanical hoisting, and high clamping force requirements, and lack a complete large-diameter water meter sealing performance testing solution from clamping and sealing to testing; third, the configuration of testing parameters and result judgment still largely rely on manual operation, and parameters such as clamping pressure and test pressure required for different diameters or models of water meters cannot be automatically adapted. In particular, there is a lack of an automated means that can perform fusion judgment based on multi-dimensional pressure attenuation characteristics, resulting in judgment methods that are mostly based on manual observation or single-dimensional threshold judgment, and a fully automated judgment and automatic data recording testing process based on the fusion of multi-dimensional pressure attenuation characteristics has not yet been formed. Summary of the Invention

[0007] To address the aforementioned technical problems, the purpose of this invention is to provide a multi-dimensional feature fusion judgment method and system for water meter sealing performance detection.

[0008] The objective of this invention is achieved through the following technical solution: A multi-dimensional feature fusion determination method for water meter sealing performance testing includes: Step S1: Obtain the nameplate information of the water meter being tested. The nameplate information includes at least the water meter diameter and / or model information. Step S2 queries the pre-stored detection resource configuration library based on the nameplate information and matches the detection control parameter set corresponding to the water meter under test. The detection control parameter set includes at least clamping pressure, test pressure, pressure holding time, and dynamic judgment threshold. Step S3 controls the clamping device to clamp and seal the water meter under test according to the clamping pressure in the set of detection control parameters; and according to the test pressure and pressure holding time, automatically venting and pressurizing to the target pressure are executed in sequence, and then the pressure holding test stage is entered. Step S4 involves continuously collecting pressure-time change data inside the water meter under test during the pressure holding test stage, and extracting multi-dimensional pressure decay characteristic parameters to characterize the leakage state of the water meter based on the pressure-time change data. Step S5 inputs the multi-dimensional pressure attenuation characteristic parameters into the pre-constructed weighted fusion judgment model, and the weighted fusion judgment model compares the pressure attenuation characteristic parameters with the corresponding dynamic judgment threshold, and outputs the sealing judgment result of the water meter under test based on the comparison result. Step S6: Based on the qualified sample data accumulated from this and historical tests, update the dynamic judgment threshold to match the judgment criteria with the actual testing conditions.

[0009] Furthermore, the set of detection and control parameters also includes the upper limit of clamping pressure, the rate of increase of test pressure, and the sampling frequency during the pressure holding phase.

[0010] Further, in step S2, matching the set of detection control parameters corresponding to the water meter being tested from the pre-stored detection resource configuration library includes: When a record matching the structural characteristic parameters of the water meter being measured exists in the detection resource configuration library, the detection control parameter set in the corresponding record is called. When no matching record is found in the detection resource configuration library, it is determined whether the structural characteristic parameters of the water meter being tested are outside the preset detection range: if they are outside the preset detection range, an over-range alarm signal is generated and the detection is terminated; if they are within the preset detection range but no corresponding parameter record is found, a parameter missing prompt signal is generated to prompt the supplementation of the corresponding detection parameters.

[0011] Furthermore, the multi-dimensional pressure attenuation characteristic parameters in step S4 include at least two of the following: (a) The pressure difference between the initial pressure and the final pressure during the pressure holding phase, i.e., the total pressure drop; (b) The pressure decay rate per unit time during the pressure holding phase, i.e., the average decay rate; (c) The maximum value of the pressure change rate between adjacent sampling times during the pressure holding phase, i.e., the maximum instantaneous decay rate; (d) The fitting residual between the pressure-time change curve during the pressure holding stage and the preset fitting model, i.e., the nonlinear fitting residual.

[0012] Further, in step S5, comparing the pressure attenuation characteristic parameters with the corresponding dynamic judgment thresholds by the weighted fusion judgment model includes: comparing each pressure attenuation characteristic parameter with its corresponding dynamic judgment threshold, calculating the normalized overscaling of each characteristic parameter, constructing a weighted fusion leakage risk index LRI, and comparing the calculated fusion leakage risk index LRI with the global risk judgment threshold. When comparing, When the water meter being tested is deemed to have an unsatisfactory seal, it is determined that the water meter is not up to standard. At that time, the water meter being tested was deemed to have passed the test for sealing.

[0013] Furthermore, in step S6, the dynamic judgment threshold update includes: after completing a preset number of tests, automatically extracting all samples from historical test data that were judged as qualified and verified as truly qualified by manual inspection; calculating the statistical distribution characteristics of pressure attenuation characteristic parameters of each dimension in qualified samples, and recalculating the judgment threshold based on the statistical distribution characteristics, so that the judgment threshold dynamically converges to a better value as test data accumulates.

[0014] Furthermore, the method also includes: forming test traceability data from the pressure-time data, test control parameters, and sealing judgment results during the test process, and storing, querying, and exporting such data.

[0015] Furthermore, the method for obtaining the nonlinear fitting residual includes: performing a second-order polynomial fitting on the pressure-time variation data, with the fitting model being: ; Estimating coefficients using the least squares method , , To minimize the sum of squared residuals: ; The nonlinear fitting residual R is defined as: ; in, N The number of sampling points. Let be the measured pressure value at the i-th sampling point. This represents the corresponding fitted pressure value.

[0016] Furthermore, the formula for calculating the weighted fusion leakage risk index (LRI) is as follows: ; in, , , , These are the normalized overscaling values ​​for total voltage drop, average decay rate, maximum instantaneous decay rate, and nonlinear fitting residuals, respectively. , , , They are respectively , , , The corresponding weight coefficients, and satisfying: ; The The value range is 0.20~0.30. The value range is 0.25~0.35. The value range is 0.15~0.25. The value range is 0.30~0.40; The global risk assessment threshold The calibration method is as follows: while calibrating the threshold values ​​for pressure attenuation characteristic parameters in each dimension, the LRI value of each qualified sample is calculated. As a threshold for global risk assessment, among which The mean of the LRI values. This represents the standard deviation of the LRI value.

[0017] Furthermore, methods for recalculating the judgment threshold based on statistical distribution characteristics include: Calculate the mean values ​​of pressure attenuation characteristic parameters in each dimension of the qualified sample. and standard deviation ,by This serves as the new judgment threshold.

[0018] A multi-dimensional feature fusion judgment system for water meter sealing performance detection includes: The information acquisition module is used to acquire the nameplate information of the water meter being tested; The parameter active adaptation module is used to match the pre-stored detection resource configuration library according to the water meter structural parameter characteristics in the nameplate information, and output the corresponding detection control parameter set. The detection execution module is used to control the clamping device to clamp and seal the water meter under test according to the detection control parameter set, and to perform automatic air venting, pressurization and pressure holding detection. The pressure acquisition module is used to continuously acquire the pressure-time change data inside the water meter under test during the pressure holding test phase. The feature extraction module is used to extract multi-dimensional pressure decay feature parameters based on the pressure-time change data; The fusion judgment module is used to construct a fusion judgment model, input the multi-dimensional pressure attenuation characteristic parameters into the fusion judgment model for comprehensive comparison, and output the sealing judgment result. The dynamic threshold update module is used to adaptively update the judgment threshold based on qualified samples in historical detection data.

[0019] Furthermore, the clamping device is an electro-hydraulic clamping device; The booster device is an injection booster driven by a variable frequency water pump; The clamping device and the pressurizing device are respectively equipped with pressure detection elements for real-time detection of clamping pressure and test pressure; The sampling frequency of the pressure acquisition module is not less than 1Hz.

[0020] Compared with the prior art, one or more embodiments of the present invention may have the following advantages: This invention automates the entire process of water meter sealing performance testing, from proactive parameter adaptation to multi-dimensional feature fusion judgment and dynamic threshold updating, and is applicable to large-diameter water meters ranging from DN50 to DN300. Its core innovations are: First, it overcomes the limitations of single-threshold judgment by extracting total pressure drop, average attenuation rate, maximum instantaneous attenuation rate, and nonlinear fitting residuals to construct a multi-dimensional feature system, comprehensively capturing the pressure attenuation characteristics caused by leakage; second, it constructs a multi-dimensional feature fusion judgment model, effectively overcoming the shortcomings of single-feature judgment which is susceptible to environmental noise and sensor fluctuations, significantly improving judgment accuracy; finally, it establishes a dynamic self-updating mechanism for the judgment threshold, enabling the judgment criteria to be continuously optimized with the accumulation of test data, further enhancing the robustness and adaptability of the judgment. This invention can be widely applied to the sealing performance testing of various metrology institutions and manufacturers of large-diameter water meters and flow meters, effectively improving testing efficiency and reducing manual intervention, providing an efficient and reliable sealing performance testing solution for large-diameter water meters used in trade settlement and important flow metering instruments such as electromagnetic flow meters and ultrasonic flow meters. Attached Figure Description

[0021] Figure 1 This is a flowchart of a multi-dimensional feature fusion judgment method for water meter sealing performance testing; Figure 2 This is a schematic diagram of the water meter sealing test device. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be described in further detail below with reference to the embodiments and accompanying drawings.

[0023] The following example, using the sealing performance test of any water meter within the DN50 to DN300 range, illustrates in detail the complete implementation process of the multi-dimensional feature fusion judgment method for water meter sealing performance provided by this invention (e.g., Figure 1 (as shown), including: Step S1: Obtain the nameplate information of the water meter being tested by means of barcode scanning or manual input, wherein the nameplate information includes at least the water meter diameter and / or model information; Step S2, based on the water meter structural parameter characteristics in the nameplate information, actively adapts the detection control parameter set corresponding to the water meter under test from the pre-stored detection resource configuration library. The detection control parameter set includes at least clamping pressure, test pressure, pressure holding time, and dynamic judgment thresholds corresponding to multi-dimensional pressure attenuation characteristic parameters. Step S3 controls the clamping device to clamp and seal the water meter under test according to the clamping pressure in the set of detection control parameters; and according to the test pressure and pressure holding time, automatically venting and pressurizing to the target pressure are executed in sequence, and then the pressure holding test stage is entered. Step S4 involves continuously collecting pressure-time change data inside the water meter under test during the pressure holding test stage, and extracting multi-dimensional pressure decay characteristic parameters to characterize the leakage state of the water meter based on the pressure-time change data. Step S5 inputs the multi-dimensional pressure attenuation characteristic parameters into the pre-constructed weighted fusion judgment model, and the weighted fusion judgment model compares the pressure attenuation characteristic parameters with the corresponding dynamic judgment threshold, and outputs the sealing judgment result of the water meter under test based on the comparison result. Step S6: Based on the qualified sample data accumulated from this and historical tests, update the dynamic judgment threshold to match the judgment criteria with the actual testing conditions.

[0024] Step S1 further includes: the water meter under test is hoisted onto the flange-type test bench by the hoisting equipment and positioned, and then automatically pulled into the clamping position by the meter pulling device; the operator uses a barcode scanner to scan the barcode on the water meter nameplate, and the host computer virtual instrument control software automatically reads the water meter information stored in the barcode; if the barcode is damaged or missing, the operator can also manually input the above nameplate information through the human-computer interaction interface of the host computer software, and the system shall take the manually input information as the standard.

[0025] Step S2 further includes: the host computer virtual instrument control software matches the water meter structural parameter characteristics (especially the diameter) in the identified nameplate information with a pre-stored detection resource configuration library. The detection resource configuration library stores the correspondence between different water meter structural parameter characteristics and detection control parameter sets. This correspondence is pre-established based on the provisions of JJG162-2019 "Verification Procedure for Drinking Cold Water Meters" and GB / T 778.3-2018 "Drinking Cold Water Meters and Hot Water Meters Part 3: Test Methods" and combined with a large amount of experimental calibration data.

[0026] When a record matching the diameter and / or model of the water meter being tested exists in the pre-stored testing resource configuration library, the testing control parameter set in the corresponding record is called. If no matching record exists, it is further determined whether the water meter diameter is outside the range of DN50 to DN300. If so, an over-range alarm signal is generated and the testing is terminated. In this embodiment, this is specifically manifested as follows: the host computer human-machine interface displays the message "The diameter of the water meter being tested exceeds the applicable range of this device (DN50~DN300), testing terminated," and the three-color LED indicator flashes a red alarm. The clamping device and the pressurizing device do not start, and the testing process is terminated. When the diameter is within the range of DN50 to DN300 and there is no corresponding matching record, the operator is prompted to enter the testing control parameter set corresponding to the model. In this embodiment, this is specifically manifested as follows: the host computer human-machine interface displays the message "No matching testing control parameters are available for the diameter of the water meter being tested. Please enter the corresponding parameters," and the three-color LED indicator flashes a red alarm. The sealing test device enters standby mode.

[0027] The set of detection control parameters with caliber as a structural characteristic parameter of the water meter, stored in the pre-stored detection resource configuration library, is shown in Tables 1 and 2 (the values ​​in the tables are only examples, and the actual values ​​are determined according to the specific water meter model and experimental calibration):

[0028]

[0029] These represent the clamping pressure, clamping pressure upper limit, total pressure drop threshold, average attenuation rate threshold, maximum instantaneous attenuation rate threshold, and nonlinear fitting residual threshold for a corresponding diameter of DN x, respectively. x can be 50, 80, 100, 150, 200, 250, or 300, corresponding to diameters of DN50, DN80, DN100, DN150, DN200, DN250, and DN300.

[0030] Wherein, the target test pressure value = test pressure coefficient × maximum allowable working pressure (MAP), and the clamping pressure target value increases with the increase of the orifice diameter; the relevant judgment threshold ( The calibration method is as follows: Select several water meter samples (no fewer than 30) of the same diameter that have been confirmed as having qualified sealing performance by traditional methods. Under standard test conditions (water temperature 20℃±5℃, ambient temperature 20℃±5℃), conduct tests according to the testing procedure of this invention. Collect pressure-time data for each sample during the pressure holding stage, calculate the statistical distribution of each attenuation characteristic parameter, and use the mean of each characteristic parameter as the basis for the calibration. Adding three times the standard deviation σ as the judgment threshold for this caliber, i.e. To ensure that the false positive rate of qualified samples does not exceed 0.3%, the judgment thresholds for different calibers are independently calibrated and regularly (at least once a year) verified and adjusted using standard leaking components.

[0031] Step S3 further includes: the host computer sends the matched clamping pressure target value to the electro-hydraulic system of the clamping device, the clamping device starts, and completes the maximum extension stroke within 30 seconds. The clamping pressure is monitored in real time by a 0.2-level high-precision pressure transmitter. When the clamping pressure reaches the target value, clamping stops to maintain axial clamping and sealing of the water meter flange. During the clamping process, the system monitors in real time whether the clamping pressure exceeds the upper limit value. If so, clamping stops immediately and an alarm is triggered to prevent damage to the water meter housing due to excessive clamping force.

[0032] After clamping, the host computer sends the matched target test pressure value to the booster device. The booster device uses a regular water pump with a frequency converter, and uses computer algorithms to control and drive the injection booster to smoothly increase the pressure. During the boosting process, the system first performs an automatic venting operation, that is, opens the vent valve to completely expel the air in the water meter and pipeline, ensuring that the test pressure medium is water (static pressure). After venting is completed, the vent valve is closed, and the booster device continues to pressurize to the target pressure. The entire pressurization process uses a 0.2-level high-precision pressure transmitter to monitor the pressure value in real time, and uses a PID control algorithm to adjust the output frequency of the frequency converter to control the pressure rise rate to not exceed 0.05MPa / s, ensuring a smooth pressure rise and avoiding pressure shock.

[0033] Once the pressure reaches the target test pressure value, the system enters the pressure holding state, and the synchronous timing device starts timing. The pressure holding time is the matched pressure holding time (60s as required by the procedure in this embodiment). During the pressure holding period, the system controls the pressurization device to maintain a constant pressure and zero flow rate, which meets the hydrostatic pressure test requirements specified in JJG162-2019. The working status is indicated by a three-color LED indicator during the test: the red LED lights up during pressurization, and the green LED lights up after entering the pressure holding state.

[0034] Step S4 further includes: during the pressure holding stage, the pressure acquisition module continuously acquires the pressure data inside the water meter being tested at a matched sampling frequency (10Hz in this embodiment), generating a pressure-time series. P ( t ), t =1, 2, …, N, where N = sampling frequency × holding time.

[0035] Start time of pressure holding phase The corresponding pressure is Termination time The corresponding pressure is The host computer software extracts the following multi-dimensional pressure decay characteristic parameters from the pressure-time series: (a) Total pressure drop The difference between the initial pressure and the final pressure during the pressure holding phase.

[0036] (1) (b) Average pressure decay rate The ratio of the total pressure drop during the holding phase to the holding time.

[0037] (2) in This refers to the pressure holding time.

[0038] (c) Maximum instantaneous pressure decay rate The maximum rate of pressure change between any two adjacent sampling points during the pressure holding phase.

[0039] (3) Where Δ t Sampling interval (Δ) t = 1 / sampling frequency).

[0040] (d) Nonlinear fitting residuals of the pressure-time curve R Perform a second-order polynomial fitting on the pressure-time series and calculate the fitting residuals. Let the fitting model be: (4) Estimating coefficients using the least squares method To minimize the sum of squared residuals: (5) Nonlinear fitting residual R Defined as: (6) R The value reflects the degree to which the pressure decay curve deviates from linear decay. If the water meter has a leak, the pressure decay curve typically exhibits a non-linear, accelerated decay characteristic. R The value increased significantly.

[0041] The reason for constructing the above four characteristic parameters in this invention is based on an in-depth analysis of the physical process of leakage: the total pressure drop reflects the overall energy loss caused by leakage and is the most direct macroscopic characterization; the average decay rate characterizes the steady-state persistence of leakage; the maximum instantaneous decay rate is sensitive to sudden leakage or instantaneous instability of the sealing surface; and the nonlinear fitting residual can effectively distinguish between normal thermodynamic fluctuations and nonlinear energy dissipation processes caused by leakage. A single feature is insufficient to fully depict the entire picture of leakage, while the fusion of multi-dimensional features can comprehensively capture leakage information from different time scales and different physical dimensions, significantly improving the accuracy of judgment and anti-interference ability.

[0042] Step S5 further includes: constructing a weighted fusion judgment model to comprehensively judge the four pressure attenuation feature parameters extracted in step S4. The specific judgment process is as follows: Calculate the normalized overscaling of each feature parameter. The four pressure attenuation feature parameters extracted in step S4 are compared with the corresponding dynamic judgment thresholds obtained in step S2, and the normalized excess ratio (NER) of each feature parameter is calculated: (7)

[0043] (8)

[0044] (9)

[0045] (10) Normalized outscaling maps four feature parameters with different dimensions and numerical ranges to a dimensionless ratio space, eliminating dimensional differences between features and making them comparable. When the normalized outscaling of a feature is greater than 1, it indicates that the single feature has exceeded its independent threshold, suggesting a possible leak; however, a normalized outscaling of less than 1 does not necessarily mean that the water meter is qualified, because there may be cumulative coupling effects between the features.

[0046] Construct a weighted fusion leakage risk index (LRI): This invention further constructs a weighted fusion leakage risk index (LRI), which combines four normalized overscaling parameters with weights: (11) Wherein, the weight coefficient vector Satisfy the normalization constraint: (12) The principles for determining each weighting coefficient are as follows: Total pressure drop (weight) The total energy loss caused by leakage is the most intuitive macroscopic representation, but it is easily affected by ambient temperature fluctuations and initial sensor drift. Therefore, its weight should not be too high, and the preferred range is 0.20~0.30. Total pressure drop (weight) ): Characterizes the steady-state persistence of leakage, reflects the average leakage level throughout the pressure holding period, has strong anti-interference ability, and is a stable indicator for judging leakage. The preferred range is 0.25~0.35. Maximum instantaneous decay rate (weight) ): It is highly sensitive to sudden leakage or instantaneous instability of the sealing surface and can capture instantaneous pressure jumps. However, it is easily affected by external vibration or electromagnetic interference, which can generate spike noise. Therefore, the weight needs to be controlled appropriately, and the preferred range is 0.15~0.25. Nonlinear fitting residual (weight) ): This is the most distinctive feature parameter of the present invention. It can effectively distinguish between normal linear thermodynamic fluctuations and nonlinear energy dissipation processes caused by leakage. It has the strongest ability to identify minute leaks and is the core sensitive feature for leak determination. Therefore, it is given the highest weight, with a preferred range of 0.30~0.40.

[0047] A typical optimal weight configuration is as follows: In this configuration, the nonlinear fitting residual has the highest weight, followed by the average decay rate, while the other two have roughly equal weights.

[0048] Global risk assessment: The calculated fusion leakage risk index LRI is compared with the global risk assessment threshold. Comparison: when If the water meter being tested is found to be unqualified in terms of sealing, then the test result is considered unqualified. when At that time, the water meter being tested was deemed to have passed the test for sealing.

[0049] Global risk assessment threshold The calibration method is as follows: In step S2, a threshold is determined ( , , , Simultaneously with calibration, the LRI value for each sample is calculated, and the mean LRI value is used. Add 3 times the standard deviation This serves as the global judgment threshold corresponding to this caliber, i.e. .

[0050] Step S6 further includes: After the system completes a certain batch (e.g., after testing 100 water meters) or reaches a preset time period, it automatically triggers a threshold update process. First, it extracts the pressure-time data of all water meter samples that were judged as "qualified" and verified as truly qualified by manual inspection from the historical testing database; then, it recalculates the multi-dimensional pressure attenuation characteristic parameters of each sample according to the method in step S4; next, it statistically analyzes the distribution of each characteristic parameter and calculates its mean. and standard deviation Finally, according to The criteria are updated to determine the threshold for multi-dimensional pressure attenuation characteristic parameters; simultaneously, the LRI value for each sample is calculated, and the mean of the LRI values ​​is used. Add 3 times the standard deviation This serves as the global judgment threshold corresponding to this caliber, i.e. Through this mechanism, the judgment threshold is not static, but can dynamically converge to a better value as qualified sample data accumulates. This enables the judgment criteria to adaptively match the actual production and testing conditions, effectively solving the problems of deviations in the preset threshold due to insufficient experimental data in the early stage, and the increased misjudgment rate of the fixed threshold due to sensor drift or changes in operating conditions.

[0051] In addition to the standard steps S1-S6 described above, this application embodiment can also perform the following after completing step S6: After the judgment is completed, the host computer automatically saves the pressure-time data, detection control parameters (clamping pressure, test pressure, holding time, and various judgment thresholds) and judgment results as structured data and stores them in the system database; at the same time, the system automatically generates an Excel-formatted test record, including basic water meter information, detection parameters, pressure-time curves, calculated values ​​of various attenuation characteristic parameters, and judgment conclusions, supporting retrieval and export; after the test, the system automatically resets—the clamping pressure is released, and the meter-pulling device automatically sends the water meter out of the clamping position. If the judgment result is unqualified, the three-color LED indicator lights up a yellow LED to provide a warning.

[0052] This embodiment also provides a multi-dimensional feature fusion judgment system for performing the above steps S1 to S6 water tightness testing. The system includes an information acquisition module for acquiring the nameplate information of the water meter under test through barcode scanning or manual input; a parameter active adaptation module for matching the corresponding test control parameter set from a pre-stored test resource configuration library using the water meter structural parameter features reflected in the nameplate information as an index; a test execution module for controlling the clamping device to clamp the seal and automatically venting, pressurizing, and maintaining pressure according to the matched test pressure and holding time; a pressure acquisition module for continuously acquiring internal pressure-time change data of the water meter under test during the holding pressure test stage; a feature extraction module for extracting multi-dimensional pressure attenuation feature parameters based on the pressure-time change data; a fusion judgment module for fusing and comparing the multi-dimensional pressure attenuation feature parameters with a judgment threshold and outputting the tightness judgment result; and a dynamic threshold update module for periodically updating the mean and standard deviation of each attenuation feature parameter according to the number of tests performed by the tightness testing device, thus updating the judgment threshold.

[0053] like Figure 2 The diagram illustrates the structural principle of the water meter sealing test device. The device includes a vertical flange-type test stand, an automatic meter-pulling device with an extendable clamping position, an electro-hydraulic clamping device, a variable frequency water pump-driven injection booster, a synchronous timing device, a three-color LED indicator, and virtual instrument control software installed on a host computer. The water meter under test is hoisted onto the flange-type test stand using a crane or electric hoist and positioned. The automatic meter-pulling device then automatically pulls it into the clamping position, scans and records the meter information, and the electro-hydraulic clamping device performs clamping and sealing. The water pump-driven injection booster performs pressure testing. The entire process is uniformly coordinated and controlled by the host computer's virtual instrument control software.

[0054] While the embodiments disclosed in this invention are as described above, the content is merely for the purpose of facilitating understanding of the invention and is not intended to limit the invention. Any person skilled in the art to which this invention pertains may make any modifications and variations in form and detail of the implementation without departing from the spirit and scope disclosed herein; however, the scope of patent protection for this invention shall still be determined by the scope defined in the appended claims.

Claims

1. A multi-dimensional feature fusion determination method for water meter sealing performance detection, characterized in that, Includes the following steps: Step S1: Obtain the nameplate information of the water meter being tested. The nameplate information includes at least the water meter diameter and / or model information. Step S2 queries the pre-stored detection resource configuration library based on the nameplate information and matches the detection control parameter set corresponding to the water meter under test. The detection control parameter set includes at least clamping pressure, test pressure, pressure holding time, and dynamic judgment threshold. Step S3 controls the clamping device to clamp and seal the water meter under test according to the clamping pressure in the set of detection control parameters; and according to the test pressure and pressure holding time, automatically venting and pressurizing to the target pressure are executed in sequence, and then the pressure holding test stage is entered. Step S4 involves continuously collecting pressure-time change data inside the water meter under test during the pressure holding test stage, and extracting multi-dimensional pressure decay characteristic parameters to characterize the leakage state of the water meter based on the pressure-time change data. Step S5 inputs the multi-dimensional pressure attenuation characteristic parameters into the pre-constructed weighted fusion judgment model, and the weighted fusion judgment model compares the pressure attenuation characteristic parameters with the corresponding dynamic judgment threshold, and outputs the sealing judgment result of the water meter under test based on the comparison result. Step S6 updates the dynamic judgment threshold based on the qualified sample data accumulated from the current and historical tests, so that the judgment criteria match the actual testing conditions.

2. The multi-dimensional feature fusion determination method for water meter sealing performance detection according to claim 1, characterized in that, The set of detection and control parameters also includes the upper limit of clamping pressure, the rate of increase of test pressure, and the sampling frequency during the pressure holding phase.

3. The multi-dimensional feature fusion determination method for water meter sealing performance detection according to claim 1, characterized in that, In step S2, matching the set of detection control parameters corresponding to the water meter being tested from the pre-stored detection resource configuration library includes: When a record matching the structural characteristic parameters of the water meter being measured exists in the detection resource configuration library, the detection control parameter set in the corresponding record is called. When no matching record is found in the detection resource configuration library, it is determined whether the structural characteristic parameters of the water meter being tested are outside the preset detection range: if they are outside the preset detection range, an over-range alarm signal is generated and the detection is terminated; if they are within the preset detection range but no corresponding parameter record is found, a parameter missing prompt signal is generated to prompt the supplementation of the corresponding detection parameters.

4. The multi-dimensional feature fusion determination method for water meter sealing performance detection according to claim 1, characterized in that, The multi-dimensional pressure attenuation characteristic parameters in step S4 include at least two of the following: (a) The pressure difference between the initial pressure and the final pressure during the pressure holding phase, i.e., the total pressure drop; (b) The pressure decay rate per unit time during the pressure holding phase, i.e., the average decay rate; (c) The maximum value of the pressure change rate between adjacent sampling times during the pressure holding phase, i.e., the maximum instantaneous decay rate; (d) The fitting residual between the pressure-time change curve during the pressure holding stage and the preset fitting model, i.e., the nonlinear fitting residual.

5. The multi-dimensional feature fusion determination method for water meter sealing performance detection according to claim 1, characterized in that, In step S5, comparing the pressure attenuation characteristic parameters with the corresponding dynamic judgment thresholds by the weighted fusion judgment model includes: comparing each pressure attenuation characteristic parameter with its corresponding dynamic judgment threshold, calculating the normalized overscaling of each characteristic parameter, constructing the weighted fusion leakage risk index LRI, and comparing the calculated fusion leakage risk index LRI with the global risk judgment threshold. When comparing, When the water meter being tested is deemed to have an unsatisfactory seal, it is determined that the water meter is not up to standard. At that time, the water meter being tested was deemed to have passed the test for sealing.

6. The multi-dimensional feature fusion determination method for water meter sealing performance detection according to claim 1, characterized in that, In step S6, the dynamic judgment threshold update includes: after completing a preset number of tests, automatically extracting all samples from historical test data that are judged as qualified and verified as truly qualified by manual inspection; calculating the statistical distribution characteristics of pressure attenuation characteristic parameters of each dimension in qualified samples, and recalculating the judgment threshold based on the statistical distribution characteristics, so that the judgment threshold dynamically converges to a better value as the test data accumulates.

7. The multi-dimensional feature fusion determination method for water meter sealing performance detection according to claim 1, characterized in that, The method further includes: forming test traceability data from the pressure-time data, test control parameters, and sealing judgment results during the test process, and storing, querying, and exporting it.

8. The multi-dimensional feature fusion determination method for water meter sealing performance detection according to claim 4, characterized in that, The method for obtaining the nonlinear fitting residuals includes: performing a second-order polynomial fitting on the pressure-time variation data, with the fitting model being: ； Estimating coefficients using the least squares method , , To minimize the sum of squared residuals: ； The nonlinear fitting residual R is defined as: ； in, N The number of sampling points. Let be the measured pressure value at the i-th sampling point. This represents the corresponding fitted pressure value.

9. The multi-dimensional feature fusion determination method for water meter sealing performance detection according to claim 5, characterized in that, The formula for calculating the weighted fusion leakage risk index (LRI) is as follows: ； in, , , , These are the normalized overscaling values ​​for total voltage drop, average decay rate, maximum instantaneous decay rate, and nonlinear fitting residuals, respectively. , , , They are respectively , , , The corresponding weight coefficients, and satisfying: ； The The value range is 0.20~0.

30. The value range is 0.25~0.

35. The value range is 0.15~0.

25. The value range is 0.30~0.40; The global risk assessment threshold The calibration method is as follows: while calibrating the threshold values ​​for pressure attenuation characteristic parameters in each dimension, the LRI value of each qualified sample is calculated. As a threshold for global risk assessment, among which The mean of the LRI values. This represents the standard deviation of the LRI value.

10. The multi-dimensional feature fusion determination method for water meter sealing performance detection according to claim 6, characterized in that, Methods for recalculating the judgment threshold based on statistical distribution characteristics include: Calculate the mean values ​​of pressure attenuation characteristic parameters in each dimension of the qualified sample. and standard deviation ,by This serves as the new judgment threshold.

11. A multi-dimensional feature fusion judgment system for water meter sealing performance detection, characterized in that, include: The information acquisition module is used to acquire the nameplate information of the water meter being tested; The parameter active adaptation module is used to match the pre-stored detection resource configuration library according to the water meter structural parameter characteristics in the nameplate information, and output the corresponding detection control parameter set. The detection execution module is used to control the clamping device to clamp and seal the water meter under test according to the detection control parameter set, and to perform automatic air venting, pressurization and pressure holding detection. The pressure acquisition module is used to continuously acquire the pressure-time change data inside the water meter under test during the pressure holding test phase. The feature extraction module is used to extract multi-dimensional pressure decay feature parameters based on the pressure-time change data; The fusion judgment module is used to construct a fusion judgment model, input the multi-dimensional pressure attenuation characteristic parameters into the fusion judgment model for comprehensive comparison, and output the sealing judgment result. The dynamic threshold update module is used to adaptively update the judgment threshold based on qualified samples in historical detection data.

12. The multi-dimensional feature fusion judgment system for water meter sealing performance detection according to claim 11, characterized in that, The clamping device is an electro-hydraulic clamping device; The booster device is an injection booster driven by a variable frequency water pump; The clamping device and the pressurizing device are respectively equipped with pressure detection elements for real-time detection of clamping pressure and test pressure; The sampling frequency of the pressure acquisition module is not less than 1Hz.

Images