CN-121499275-B - Method and system for online detection of sealing surface hardness of flat gate valve

CN121499275BCN 121499275 BCN121499275 BCN 121499275BCN-121499275-B

Abstract

The invention relates to the technical field of flat gate valve detection and discloses a method and a system for detecting the hardness of a sealing surface of a flat gate valve on line, wherein the system comprises a multi-mode data acquisition module, a signal preprocessing module, a hardness characteristic analysis module, an intelligent hardness prediction module, a diagnosis and calibration module and a cloud platform communication and monitoring module; when the hardness of the sealing surface of the flat gate valve is detected online, morphology, acoustic and thermal multidimensional data are synchronously acquired through the multi-mode sensor array, and a signal processing technology combining wavelet transformation and self-adaptive filtering is adopted, so that high-frequency noise generated by environmental vibration and electromagnetic interference can be separated and suppressed, the purity of hardness characteristic signals acquired under complex industrial working conditions is ensured, the detection error caused by noise interference is reduced, and the accuracy and the reliability of the online detection of the hardness of the sealing surface are improved.

Inventors

DING XUEFENG
CAI LEI
LI QUN
LI BO

Assignees

盐城奥凯明通阀门有限公司

Dates

Publication Date: 20260508
Application Date: 20251117

Claims (9)

1. The online detection method for the hardness of the sealing surface of the flat gate valve is characterized by comprising the following steps: S1, acquiring a historical hardness detection data set of a sealing surface of a flat gate valve, and acquiring surface morphology data and a material response signal of the sealing surface of the currently detected gate valve through a multi-mode sensor array to generate original detection data of the sealing surface; s2, performing signal preprocessing and noise filtering processing based on the original detection data of the sealing surface to generate denoising sealing surface characteristic data; S3, carrying out hardness characteristic extraction processing on the sealing surface material according to the denoising sealing surface characteristic data and an acoustic impedance analysis model, and generating a sealing surface hardness characteristic parameter set, wherein the method comprises the following steps of: s31, extracting acoustic impedance characteristics including sound wave propagation speed, attenuation coefficient and impedance ratio based on the denoising sealing surface characteristic data, and calculating based on the relation between acoustic impedance and material density; s32, combining a sealing surface material database, simulating acoustic responses under different hardness by finite element analysis, and generating a mapping relation model of hardness and acoustic impedance; s33, inputting the extracted characteristic parameters into a mapping relation model, and outputting a sealing surface hardness characteristic parameter set which comprises a hardness index, an elastic modulus estimated value and a material fatigue coefficient; s4, performing hardness value prediction calculation on the hardness characteristic parameter set of the sealing surface by adopting a machine learning regression algorithm to generate a hardness prediction value of the sealing surface; s5, performing hardness abnormality diagnosis and calibration processing based on the seal surface hardness predicted value to generate a seal surface hardness online detection result; And S6, transmitting the online detection result of the hardness of the sealing surface to a cloud supervision platform in real time through an Internet of things gateway, and realizing remote monitoring and early warning of the hardness of the sealing surface of the gate valve.
2. The method for detecting the hardness of the sealing surface of the flat gate valve on line according to claim 1, wherein the step of collecting the surface topography data and the material response signals of the sealing surface of the currently detected gate valve through the multi-mode sensor array in S1 comprises the following steps: S11, installing a multi-mode sensor array outside a valve body of the flat gate valve, wherein the multi-mode sensor array comprises a laser displacement sensor, an ultrasonic probe and an infrared thermal imager, the laser displacement sensor is used for collecting microscopic morphology point cloud data of a sealing surface, the ultrasonic probe is used for transmitting high-frequency sound waves and receiving reflection signals of the sealing surface, and the infrared thermal imager is used for capturing temperature distribution images of the sealing surface; s12, synchronously collecting sealing surface data through the multi-mode sensor array in a normal running state of the gate valve, and generating original sealing surface detection data which are multi-dimensional data matrixes including morphology, acoustic and thermal parameters.
3. The method for detecting the hardness of the sealing surface of the flat gate valve on line according to claim 1, wherein the step of performing signal preprocessing and noise filtering in the step S2 comprises the following steps of: S21, acquiring original detection data of the sealing surface, performing time-frequency domain decomposition by adopting a wavelet transformation algorithm, and separating high-frequency noise and low-frequency signal components; s22, smoothing the decomposed signals by applying an adaptive filter and a Kalman filtering model, removing environmental vibration and electromagnetic interference noise, and generating denoising sealing surface characteristic data; S23, carrying out data normalization processing on the denoised data, and mapping different sensor data to a unified numerical interval.
4. The method for detecting the hardness of the sealing surface of the flat gate valve on line according to claim 1, wherein the step of predicting the hardness value in S4 comprises the following steps: S41, constructing a prediction model by adopting a machine learning regression algorithm, and training the model based on the historical hardness detection dataset; S42, performing feature selection and dimension reduction processing on the seal surface hardness feature parameter set, and reducing redundant features by using principal component analysis; S43, inputting the processed characteristics into a trained regression model, predicting the hardness value of the sealing surface, and outputting a confidence interval and error estimation.
5. The method for detecting the hardness of the sealing surface of the flat gate valve on line according to claim 1, wherein the step of performing abnormality diagnosis and calibration in S5 comprises the steps of: S51, setting a hardness threshold range based on the hardness predicted value of the sealing surface, and triggering abnormal alarm when the predicted value exceeds a threshold value; s52, correcting the abnormal value by adopting a Bayesian inference model, and correcting the real-time working condition parameters such as pressure and temperature to generate a corrected hardness value; s53, comparing the calibrated hardness value with a standard hardness database, and outputting a seal surface hardness online detection report which comprises hardness grades, abrasion states and recommended maintenance measures.
6. The method for detecting the hardness of the sealing surface of the flat gate valve on line according to claim 1, wherein the step S6 of transmitting the line detection result to the cloud supervision platform in real time through the gateway of the Internet of things comprises the following steps: S61, packaging the online detection report of the hardness of the sealing surface into a data packet with a specific format through an Internet of things gateway, and uploading the data packet to a cloud supervision platform in real time by adopting a lightweight protocol; s62, integrating a big data analysis module by a cloud supervision platform, and performing aggregation analysis on a plurality of gate valve detection data to generate a trend prediction and health assessment report; and S63, generating a control instruction and driving a gate valve executing mechanism based on the health evaluation report so as to adjust the operation parameters and the opening and closing states of the gate valve.
7. The method for online detection of hardness of sealing surface of flat gate valve according to claim 4, wherein training the model based on the historical hardness detection data set in S41 comprises the following steps: S411, dividing the seal surface hardness characteristic parameter set into a training subset and a verification subset, and generating a plurality of training sample sets by adopting a Bootstrap sampling method; S412, constructing a plurality of heterogeneous regression models to form an integrated learner, wherein the integrated learner comprises a decision tree regressor, a neural network regressor and a support vector regressor; s413, fusing the prediction results of the regression models by a weighted average method, and dynamically adjusting weight coefficients according to the performance of the models on the verification set; And S414, performing secondary training by taking the output of the primary regressive as the input of the meta-learner by adopting a Stacking integration strategy, and generating a final seal surface hardness predicted value.
8. The method for detecting the hardness of the sealing surface of the flat gate valve on line according to claim 5, wherein the step of triggering an abnormality alarm in S51 comprises the following steps: S511, constructing a deep confidence network model, wherein the deep confidence network model comprises a plurality of limited Boltzmann machine layers and an output layer, and is used for learning a nonlinear mapping relation between hardness characteristics and abnormal modes; s512, training network parameters in a mode of combining unsupervised pre-training and supervised micro-tuning, wherein the pre-training stage uses hardness data under normal working conditions to perform feature learning; S513, calculating the reconstruction error of the real-time hardness data and the normal mode in an abnormality diagnosis stage, and triggering an abnormality alarm when the error exceeds a set threshold value; S514, carrying out weighted focusing on the abnormal characteristics by combining an attention mechanism, improving the detection sensitivity of the micro defects, and generating an abnormal diagnosis report.
9. An online detection system for the hardness of the sealing surface of a flat gate valve, which is characterized in that the online detection system is used for realizing the online detection method for the hardness of the sealing surface of the flat gate valve according to any one of claims 1-8, and is characterized in that the system comprises: The multi-mode data acquisition module is used for acquiring a historical hardness detection data set of the sealing surface of the flat gate valve, acquiring the shape point cloud data of the sealing surface of the currently detected gate valve through the laser displacement sensing unit, acquiring an acoustic reflection signal through the ultrasonic detection unit, capturing a temperature distribution image through the infrared thermal imaging unit, and generating and outputting original detection data of the sealing surface; the signal preprocessing module is used for receiving the original detection data of the sealing surface, performing time-frequency domain decomposition through the wavelet transformation unit, removing environmental noise through the self-adaptive filtering unit, and outputting the denoising sealing surface characteristic data through the data normalization unit; The hardness characteristic analysis module is used for receiving the denoising sealing surface characteristic data, extracting characteristic parameters through the acoustic impedance calculation unit, constructing a hardness-acoustic impedance relation model by utilizing the finite element mapping unit, and outputting a sealing surface hardness characteristic parameter set; The intelligent hardness prediction module is used for receiving the hardness characteristic parameter set of the sealing surface, calculating a hardness value through the machine learning regression unit, determining prediction accuracy through the confidence interval evaluation unit, and outputting a hardness predicted value of the sealing surface; the diagnosis and calibration module is used for receiving the predicted value of the hardness of the sealing surface, triggering abnormal alarm through the threshold comparison unit, correcting by utilizing the Bayesian inference calibration unit in combination with the working condition parameters, and generating and outputting an online detection result of the hardness of the sealing surface; And the cloud platform communication and monitoring module is used for receiving the online detection result of the hardness of the sealing surface, uploading data to the cloud through the internet of things protocol transmission unit, carrying out health assessment by utilizing the big data aggregation analysis unit, and realizing remote monitoring and maintenance through the early warning instruction issuing unit.

Description

Method and system for online detection of sealing surface hardness of flat gate valve Technical Field The invention relates to the technical field of flat gate valve detection, in particular to an online detection method and system for the hardness of a sealing surface of a flat gate valve. Background The gate valve is widely applied to the industrial field, and aims to control the flow of fluid in a valve body pipeline through the gate valve, the gate valve is an opening and closing piece gate plate, the moving direction of the gate plate is perpendicular to the direction of the fluid, the gate valve can only be fully opened and fully closed and can not be regulated and throttled, the gate valve is sealed through the contact of a valve seat and the gate plate, and a metal material is usually deposited on a sealing surface to increase the wear resistance. At present, as the flat gate valve is in an industrial environment with high temperature, high pressure and strong vibration for a long time, when the hardness of a sealing surface is detected on line, an original signal collected by a deployed multi-mode sensor contains a large amount of environmental noise, and cannot separate working condition vibration interference and real hardness characteristic signals in real time, and when the strong electromagnetic interference and mechanical vibration noise are mixed in a detection signal, hardness characteristic extraction distortion can be caused, so that the accuracy and reliability of hardness detection cannot be ensured. Therefore, an online detection method and an online detection system for the hardness of the sealing surface of the flat gate valve are provided to solve the problems. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a method and a system for detecting the hardness of the sealing surface of a flat gate valve on line, which solve the problems that the accuracy and the reliability of hardness detection cannot be ensured due to the extraction distortion of the hardness characteristics in the background art. In order to achieve the purpose, the invention provides the technical scheme that the method and the system for detecting the hardness of the sealing surface of the flat gate valve on line comprise the following steps: S1, acquiring a historical hardness detection data set of a sealing surface of a flat gate valve, and acquiring surface morphology data and a material response signal of the sealing surface of the currently detected gate valve through a multi-mode sensor array to generate original detection data of the sealing surface; s2, performing signal preprocessing and noise filtering processing based on the original detection data of the sealing surface to generate denoising sealing surface characteristic data; s3, carrying out hardness characteristic extraction processing on the sealing surface material according to the denoising sealing surface characteristic data and the acoustic impedance analysis model, and generating a sealing surface hardness characteristic parameter set; s4, performing hardness value prediction calculation on the hardness characteristic parameter set of the sealing surface by adopting a machine learning regression algorithm to generate a hardness predicted value of the sealing surface; S5, performing hardness abnormality diagnosis and calibration processing based on the predicted value of the hardness of the sealing surface to generate an online detection result of the hardness of the sealing surface; and S6, transmitting the online detection result of the hardness of the sealing surface to a cloud supervision platform in real time through an Internet of things gateway, and realizing remote monitoring and early warning of the hardness of the sealing surface of the gate valve. Preferably, the collecting the surface topography data and the material response signal of the sealing surface of the currently detected gate valve through the multi-mode sensor array in the step S1 comprises the following steps: s11, installing a multi-mode sensor array outside a valve body of the flat gate valve, wherein the multi-mode sensor array comprises a laser displacement sensor, an ultrasonic probe and an infrared thermal imager, the laser displacement sensor is used for collecting micro-morphology point cloud data of a sealing surface, the ultrasonic probe is used for transmitting high-frequency sound waves and receiving reflection signals of the sealing surface, and the infrared thermal imager is used for capturing temperature distribution images of the sealing surface; S12, synchronously collecting sealing surface data through the multi-mode sensor array in a normal running state of the gate valve, and generating original sealing surface detection data which are multi-dimensional data matrixes including morphology, acoustic and thermal parameters. Preferably, the signal preprocessing and noise filtering in S2 includes the following steps: