A method for predicting pressure drop loss of rotary ball valve under pulsating flow
By identifying the flow cycle and detection points of the rotary ball valve, calculating the friction pressure drop and roughness, and building a prediction model, the problem of inaccurate prediction of the pressure drop loss change of the rotary ball valve is solved, and the accurate identification and early warning of the rotary ball valve status is achieved, and the improvement of the improvement of the Pipeline maintenance efficiency.
Patent Information
- Application Number
- CN202411217295.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-02
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-09-02
AI Technical Summary
In the use of rotary ball valves, the corrosion and interference of fluid in the pipeline on the inner wall of the pipeline and the rotary ball valve gradually accumulates with the increase of time, resulting in a change in pressure drop loss. The existing technology has not conducted in-depth research on this change, resulting in The prediction of the pressure drop loss of the rotating ball valve in use is not accurate enough.
By collecting the set of pulsating flow parameters, identifying the flow cycle, determining the detection point, calculating the friction pressure drop data, building a roughness prediction model and a pressure drop loss prediction model, combining the pipeline and valve detection data, calculating the abnormal index, and realizing the pressure on the rotating ball valve Accurate identification of loss reduction and abnormal situations.
The accuracy of prediction of pressure drop loss under pulsating flow of the rotary ball valve is achieved, ensuring accurate identification and early warning of the inner wall of the pipe and the rotary ball valve status, and improving the efficiency of maintenance and maintenance.
Smart Images

Figure CN118708948B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of rotary ball valve detection, in particular to a method for predicting pressure drop loss of a rotary ball valve under pulsating flow. Background Art
[0002] Pulsating flow refers to the flow state of fluid in the pipeline showing periodic changes, usually caused by periodic changes in flow or pressure. Its characteristics are that the flow velocity, flow or pressure of the fluid shows obvious periodic fluctuations over time. Among them, the rotary ball valve is an important structure for periodic control of pulsating flow.
[0003] The rotary ball valve is a commonly used valve type. Its design feature is that a spherical valve core is installed inside the valve body. By adjusting the rotation angle of the spherical valve core, the fluid flow in the pipeline can be controlled. It is widely used in industrial and municipal fields.
[0004] Due to the resistance of the rotary ball valve in the fluid flow, a certain pressure drop loss will be caused; in the periodic fluctuation of the pulsating flow, the pressure drop loss generated by the rotary ball valve shows a certain fluctuation. The pressure drop loss of the rotary ball valve in the pulsating flow can be used to identify abnormalities, determine whether there are abnormalities, and effectively understand the working status of the rotary ball valve, which is helpful for the maintenance and inspection of the pipeline.
[0005] In the prior art, the prediction of the pressure drop loss of the rotary ball valve under pulsating flow is mainly to establish an equivalent model, calculate the pulsating pressure drop loss corresponding to each equivalent model through the pulsating fluid velocity and pulsating instantaneous flow in the equivalent model, and calculate the total pressure drop loss of the rotary ball valve under pulsating flow based on all the pulsating pressure drop losses.
[0006] The existing technology can theoretically accurately measure the pressure drop of a rotary ball valve. However, during the use of the rotary ball valve, as time goes by, the corrosion and interference of the fluid in the pipeline on the inner wall of the pipeline and the rotary ball valve gradually accumulate, causing changes in the pressure drop loss caused by the inner wall of the pipeline and the rotary ball valve. The existing technology has not conducted in-depth research on the corrosion and interference changes of the inner wall of the pipeline and the rotary ball valve, resulting in inaccurate prediction of the pressure drop loss of the rotary ball valve under use.
[0007] Therefore, a method for predicting the pressure drop loss of a rotary ball valve under pulsating flow is proposed. Summary of the invention
[0008] The purpose of the present invention is to provide a method for predicting the pressure drop loss of a rotary ball valve under pulsating flow, collect a pulsating flow parameter set, determine an identification period and a detection point according to the pulsating flow parameter set; detect the detection point within the identification period to obtain first detection data, valve detection data and second detection data; obtain friction pressure drop data according to the detection data; construct a roughness prediction model to identify the pulsating flow parameter set and the detection data to obtain the pipeline roughness, and further calculate the pipeline friction pressure drop; obtain first pressure drop loss data according to the friction pressure drop data and the pipeline friction pressure drop; construct a pressure drop loss prediction model to identify the pulsating flow parameter set and the valve detection data to obtain second pressure drop loss data; obtain an abnormality index according to the first pressure drop loss data and the second pressure drop loss data, so as to realize accurate identification and early warning of abnormalities of the rotary ball valve.
[0009] To achieve the above object, the present invention provides a method for predicting pressure drop loss of a rotary ball valve under pulsating flow, comprising:
[0010] S10. Collecting data on the pulsating flow in the pipeline to obtain a pulsating flow parameter set, identifying a flow period of the pulsating flow according to the pulsating flow parameter set; and determining an identification period according to the flow period;
[0011] S20. Determine the first detection point, the valve detection point, and the second detection point by using the pulsating flow parameter set and the distance judgment model; collect data on the detection points within the identification period to obtain first detection data, valve detection data, and second detection data;
[0012] S30. Calculate the friction pressure drop data between the first detection point and the second detection point according to the first detection data and the second detection data;
[0013] S40. Constructing a roughness prediction model to identify the pulsating flow parameter set, the first detection data and the second detection data, and obtaining the first pipeline roughness and the second pipeline roughness;
[0014] S50. Obtain a first pipeline friction pressure drop according to the first pipeline roughness, the pulsating flow parameter set and the first detection data, and obtain a second pipeline friction pressure drop according to the second pipeline roughness, the pulsating flow parameter set and the second detection data; obtain first pressure drop loss data according to the friction pressure drop data, the first pipeline friction pressure drop and the second pipeline friction pressure drop; and store the obtained data in a database;
[0015] S60. Constructing a pressure drop loss prediction model to identify the pulsating flow parameter set and the valve detection data to obtain second pressure drop loss data;
[0016] S70. Obtain an abnormality index of the rotary ball valve according to the first pressure drop loss data and the second pressure drop loss data, and issue an early warning according to the abnormality index.
[0017] The pulsating flow parameter set includes flow velocity data, pressure data, fluid density, fluid temperature, fluid viscosity and fluid corrosivity of the pipeline fluid;
[0018] The steps to identify the flow period based on the pulsating flow parameter set are:
[0019] Preprocess the flow rate data and pressure data to remove data noise and interference;
[0020] Performing Fourier transform on the preprocessed flow velocity data and pressure data to convert the time domain features in the flow velocity data and pressure data into the frequency domain;
[0021] The flow period of the pulsating flow is obtained by performing period identification on the Fourier transformed flow velocity data and pressure data.
[0022] The first detection point and the second detection point are set as follows:
[0023] In the identification cycle, taking the rotary ball valve as a starting point, identifying the fluid cross-sectional area in front of the rotary ball valve;
[0024] The fluid cross-sectional area is compared with the pipeline cross-sectional area in the order of being away from the rotary ball valve, and when the comparison result meets the threshold, the distance between the position of the fluid cross-sectional area and the position of the rotary ball valve is taken as the influence distance;
[0025] Obtaining all influence distances within the identification period, and taking the maximum value of the influence distances as the maximum influence distance;
[0026] The flow velocity data and pressure data at the rotary ball valve are collected, and the maximum impact distance is trained and fitted based on the fluid density, fluid temperature, fluid viscosity and fluid corrosivity to obtain the first distance judgment model;
[0027] The cross-sectional area of the fluid behind the rotating valve in the identification period is identified in the same manner; and a second distance judgment model is obtained through model training;
[0028] The first distance and the second distance are obtained by using the first distance judgment model and the second distance judgment model, and the first detection point and the second detection point are determined according to the first distance, the second distance and the position of the rotating ball valve.
[0029] The first detection data includes: a first initial roughness, a first distance, a first pressure, a first flow rate and a pipe diameter; the second detection data includes: a second initial roughness, a second distance, a second pressure, a second flow rate and a pipe diameter; the valve detection data includes: valve pressure, valve flow rate, valve vibration data and valve sound data.
[0030] The method for obtaining the friction pressure drop data is:
[0031]
[0032] in, is the total pressure drop between the first detection point and the second detection point, which is measured by the pressure sensor; represents the gravity pressure drop, which refers to the pressure drop caused by the height difference between the first detection point and the second detection point; Indicates velocity pressure drop, which refers to the pressure drop caused by the change in flow velocity between the first detection point and the second detection point; Indicates the friction pressure drop data, which refers to the pressure drop caused by the pipe friction and the rotating ball valve between the first test point and the second test point.
[0033] The process of obtaining the roughness prediction model is as follows:
[0034] Determine a detection position of the pipeline, and detect the detection position to obtain an initial roughness of the pipeline; the initial roughness is obtained by acquiring and identifying an image of the pipeline;
[0035] Performing a pulsating fluid experiment on the pipeline, and collecting corresponding pulsating flow parameter sets in units of flow cycles;
[0036] Then, the image of the experimental pipe is acquired and recognized to obtain the experimental roughness;
[0037] The initialization model is trained according to the initial roughness, the pulsating flow parameter set and the experimental roughness to obtain a roughness prediction model.
[0038] The method for obtaining the initial roughness through image recognition is:
[0039] Setting a detection length, and determining a region to be detected according to the detection length, a detection position of the pipeline, and a fluid flow direction, wherein the region to be detected is an inner wall of the pipeline;
[0040] Perform image recognition and splicing on the area to be detected to obtain an image of the area to be detected; the image of the area to be detected is a rectangle, the length of the rectangle is the circumference of the pipe section circle, and the width of the rectangle is the detection length;
[0041] Evenly splitting the rectangular length of the image of the area to be detected to obtain a plurality of sub-area images;
[0042] Constructing an image recognition model to perform roughness recognition on the sub-region image to obtain the sub-region roughness;
[0043] Monitoring the water level information of the detection position by a water level detection device, and determining the immersion time corresponding to the sub-region image according to the water level information within the recognition period;
[0044] weighting the sub-region roughness of the sub-region image according to the immersion time to obtain an initial roughness;
[0045] The calculation formula of the initial roughness is:
[0046]
[0047]
[0048] in, represents the initial roughness; Indicates the number of sub-areas divided; represents the immersion time of the sub-area; represents the roughness of the sub-region; represents the immersion time of the sub-area and
[0049] Inputting the first pipeline roughness, the first detection data and the pulsating flow parameter set into the Fanning formula to calculate the first pipeline friction pressure drop;
[0050]
[0051] in, represents the friction pressure drop of the first pipeline; represents Fanning's formula; represents the roughness of the first pipeline; represents the first detection data; represents the pulsating flow parameter set;
[0052] Inputting the second pipeline roughness, the second detection data and the pulsating flow parameter set into the Fanning formula to calculate the friction pressure drop of the second pipeline;
[0053]
[0054] in, represents the friction pressure drop of the second pipeline; represents Fanning's formula; Indicates the second pipeline roughness; represents the second detection data; represents the pulsating flow parameter set;
[0055] The first pressure drop loss data is calculated by the friction pressure drop data, the first pipeline friction pressure drop and the second pipeline friction pressure drop:
[0056]
[0057] in, Indicates the first pressure drop loss data.
[0058] The training process of the pressure drop loss prediction model is:
[0059] Constructing a database, inputting the pulsating flow parameter set, the first detection data, the valve detection data, the second detection data and the first pressure drop loss data within the pipeline identification period into the database for storage;
[0060] Obtaining a pulsating flow parameter set, valve detection data, and first pressure drop loss data within a historical identification period from a database;
[0061] Obtain pipeline characteristics, fluid characteristics, pressure characteristics, vibration characteristics and sound characteristics through the pulsating flow parameter set and valve detection data, fit the first pressure drop loss data, and obtain a pressure loss prediction model;
[0062] The pulsating flow parameter set and valve detection data are identified by the pressure loss prediction model to obtain the second pressure drop loss data.
[0063] The abnormality index is calculated by the difference between the first pressure drop loss data and the second pressure drop loss data;
[0064] The calculation formula of the abnormal index is:
[0065]
[0066] in, Indicates the abnormal index within the identification period; represents the variance function; represents a third data set, obtained by correspondingly subtracting the second pressure drop loss data from the first pressure drop loss data; represents the variance threshold; express and distance; Indicates the distance threshold; Indicated by natural constant An exponential function with base .
[0067] Compared with the prior art, the present invention has the following beneficial effects:
[0068] 1. The present invention takes the rotary ball valve as the starting point, and identifies the influence of the rotary ball valve on the nearby water flow within the period through the fluid cross-sectional area and the pipeline cross-sectional area to obtain the influence distance; the maximum value of the influence distance within the identification period is used as the maximum influence distance; the maximum influence distance is fitted through data collection to obtain a distance judgment model; the maximum influence range can be accurately obtained through the distance judgment model, thereby determining the accurate detection point and ensuring the accuracy of data acquisition.
[0069] 2. The present invention uses image recognition technology to identify and quantify the initial roughness inside the pipeline, then conducts a pulsating fluid experiment on the pipeline, and performs image recognition on the inside of the pipeline after the experiment to obtain experimental roughness; the initial roughness and experimental roughness can accurately reflect the impact of the fluid on the pipeline roughness during the flow cycle, thereby training a roughness prediction model to achieve accurate prediction of the roughness inside the pipeline.
[0070] 3. The present invention performs model training through pulsating flow parameter sets and valve detection data, and obtains a pressure loss prediction model by fitting pipeline characteristics, fluid characteristics, pressure characteristics, vibration characteristics and sound characteristics; the pressure loss prediction model can accurately predict the pressure drop loss data of the rotary ball valve based on the pulsating flow parameter set and valve detection data within the identification period.
[0071] 4. The present invention uses the pressure drop loss data collected and calculated through the detection point data as the first pressure drop loss data, and uses the pressure drop loss data predicted by the pressure loss prediction model as the second pressure drop loss data. The abnormality index is calculated through the comprehensive difference and individual differences between the first pressure drop loss data and the second pressure drop loss data, thereby realizing accurate identification and quantification of abnormal conditions of the rotary ball valve within the identification period. BRIEF DESCRIPTION OF THE DRAWINGS
[0072] Figure 1 It is a schematic flow chart of a method for predicting pressure drop loss of a rotary ball valve under pulsating flow of the present invention;
[0073] Figure 2 It is a schematic diagram of the process of determining the area to be detected of the present invention;
[0074] Figure 3 It is a schematic diagram of splitting the image of the area to be detected of the present invention;
[0075] Figure 4 It is a schematic diagram of the process of calculating the abnormal index of the present invention;
[0076] Figure 2 In the figure: a represents the pipeline; b represents the flow direction of the fluid; c represents the detection position; d and e represent the endpoints of the detection length line segment, de represents the detection length; f and g represent the cross-section of the pipeline; h represents the area to be detected; Figure 3 In the figure: d and e represent the endpoints of the detection length line segment; de represents the detection length; h1 represents the image of the area to be detected; h2 represents the image of the area to be detected after splitting. DETAILED DESCRIPTION
[0077] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0078] In order to effectively identify the abnormality of the pressure drop loss of a rotary ball valve in a pulsating flow, the present invention proposes a method for predicting the pressure drop loss of a rotary ball valve in a pulsating flow.
[0079] Embodiment 1
[0080] The process of the method for predicting the pressure drop loss of a rotary ball valve under pulsating flow proposed by the present invention is as follows: Figure 1 As shown, including:
[0081] S10. Collect data on the pulsating flow in the pipeline to obtain a pulsating flow parameter set, identify the flow period of the pulsating flow according to the pulsating flow parameter set; and determine an identification period according to the flow period.
[0082] The pulsating flow parameter set includes flow velocity data, pressure data, fluid density, fluid temperature, fluid viscosity and fluid corrosivity of the pipeline fluid.
[0083] Among them, flow rate data, pressure data and fluid temperature are measured by IoT sensors; fluid density, fluid viscosity and fluid corrosiveness are tested according to the type of fluid.
[0084] The steps to identify the flow period based on the pulsating flow parameter set are:
[0085] Preprocess the flow rate data and pressure data to remove data noise and interference;
[0086] Performing Fourier transform on the preprocessed flow velocity data and pressure data to convert the time domain features in the flow velocity data and pressure data into the frequency domain;
[0087] The flow period of the pulsating flow is obtained by performing period identification on the Fourier transformed flow velocity data and pressure data.
[0088] In determining the identification period, an integral multiple of the flow period may be selected as the identification period.
[0089] The present invention collects pulsating flow data in a pipeline to obtain a pulsating flow parameter set; the pulsating flow parameter set can be used to understand the fluid conditions in the pipeline in detail; at the same time, the data in the pulsating flow parameter set is subjected to Fourier transformation to convert the time domain characteristics into frequency, and the flow period of the pulsating fluid is accurately obtained; the identification period for identification detection is obtained through the flow period, which can improve the accuracy of pipeline detection.
[0090] S20. Determine a first detection point, a valve detection point, and a second detection point through a pulsating flow parameter set and a distance judgment model; collect data on the detection points within the identification period to obtain first detection data, valve detection data, and second detection data.
[0091] During the adjustment process of the rotary ball valve, when the fluid flows through the narrow cross-section of the rotary ball valve, flow separation and turbulence will occur, resulting in local flow velocity and pressure fluctuations nearby, which in turn affects the overall flow characteristics of the pipeline; if data detection is performed directly near the rotary valve, it is easy to be affected by the uneven flow field caused by flow separation and turbulence, reducing the accuracy of the measurement.
[0092] Therefore, the present invention constructs a distance judgment model based on the deep neural network DNN.
[0093] The method of setting the first detection point and the second detection point according to the distance judgment model is:
[0094] In the identification cycle, taking the rotary ball valve as a starting point, the cross-sectional area of the fluid in front of the rotary ball valve is identified; the cross-sectional area of the fluid is obtained by calculating the fluid flow rate and flow velocity;
[0095] The fluid cross-sectional area is compared with the pipe cross-sectional area in the order away from the rotary ball valve, and when the comparison result meets the threshold, the distance between the position of the fluid cross-sectional area and the position of the rotary ball valve is used as the influence distance; the threshold is obtained according to the fluid cross-sectional area when the pulsating flow is not provided with a rotary ball valve;
[0096] Obtaining all influence distances within the identification period, and taking the maximum value of the influence distances as the maximum influence distance;
[0097] The flow velocity data and pressure data at the rotary ball valve are collected, and the maximum impact distance is trained and fitted based on the fluid density, fluid temperature, fluid viscosity and fluid corrosivity to obtain the first distance judgment model;
[0098] The cross-sectional area of the fluid behind the rotating valve in the identification period is identified in the same manner; and a second distance judgment model is obtained through model training;
[0099] The first distance and the second distance are obtained by using the first distance judgment model and the second distance judgment model, and the first detection point and the second detection point are determined according to the first distance, the second distance and the position of the rotating ball valve.
[0100] The present invention takes a rotary ball valve as a starting point, and obtains the influence of the rotary ball valve on the nearby water flow within an identification period through the fluid cross-sectional area and the pipeline cross-sectional area, and obtains the influence distance; the maximum value of the influence distance within the identification period is used as the maximum influence distance; the maximum influence distance is fitted through data collection to obtain a distance judgment model, and the maximum influence range of the rotary ball valve within the identification period can be accurately obtained through the distance judgment model, thereby determining an accurate detection point.
[0101] The first detection data includes: a first initial roughness, a first distance, a first pressure, a first flow rate and a pipe diameter; the second detection data includes: a second initial roughness, a second distance, a second pressure, a second flow rate and a pipe diameter; the valve detection data includes: valve pressure, valve flow rate, valve vibration data and valve sound data.
[0102] A time interval is set within the identification cycle, and data is collected according to the time interval. For example, the first detection data is obtained in the identification cycle N. , Indicates the number of time intervals in the identification cycle, the second detection data ; Valve test data ; Indicates The valve detection data collected at each time interval includes valve pressure, valve flow rate, valve vibration data and valve sound data.
[0103] The present invention collects data according to the set first detection point, the rotary ball valve position and the second detection point to obtain fluid data, pipeline data or valve data at the location, providing a data basis for predicting the pressure drop loss of the rotary ball valve.
[0104] S30. Calculate the friction pressure drop data between the first detection point and the second detection point based on the first detection data and the second detection data.
[0105] The pressure drop in a pipeline can be broken down into several major components, including:
[0106] Gravity pressure drop: If the pipeline has a height change in the horizontal direction, the fluid will produce a change in static pressure during the rise or fall process. This change is called gravity pressure drop.
[0107] Velocity pressure drop: When a fluid flows in a pipe, the change in kinetic energy caused by the change in velocity will lead to a loss in pressure, which is called velocity pressure drop.
[0108] Friction pressure drop: The energy loss caused by the friction of the fluid on the inner wall of the pipeline is called pipeline friction pressure drop loss, which is related to factors such as the flow velocity of the fluid in the pipeline, the roughness of the inner wall of the pipeline, and the length of the pipeline.
[0109] The total pressure drop in the pipeline can be expressed as the sum of these three parts, so the friction pressure drop can be obtained by subtracting the gravity pressure drop and velocity pressure drop from the total pressure drop.
[0110] The steps for obtaining the friction pressure drop data are:
[0111]
[0112] in, is the total pressure drop between the first detection point and the second detection point, which is measured by the pressure sensor; represents the gravity pressure drop, which refers to the pressure drop caused by the height difference between the first detection point and the second detection point; Indicates velocity pressure drop, which refers to the pressure drop caused by the change in flow velocity between the first detection point and the second detection point; Indicates the friction pressure drop data, which refers to the pressure drop caused by the pipe friction and the rotating ball valve between the first test point and the second test point.
[0113] The present invention obtains the total pressure drop between detection points through a pressure sensor, obtains the gravity pressure drop through the height difference between detection points, obtains the velocity pressure drop through the velocity difference between detection points, and obtains the friction pressure drop by calculating the total pressure drop, gravity pressure drop and velocity pressure drop, thereby improving the measurement efficiency and accuracy.
[0114] S40. Construct a roughness prediction model to identify the pulsating flow parameter set, the first detection data and the second detection data to obtain the first pipeline roughness and the second pipeline roughness.
[0115] In the friction pressure drop of the pipeline, the roughness of the inner wall of the pipeline is a key indicator; the roughness is interfered by factors such as flow rate data, pressure data, fluid density, fluid temperature, fluid viscosity and fluid corrosiveness, and presents time characteristics. Therefore, the present invention constructs a roughness prediction model based on the CNN-LSTM neural network model.
[0116] The process of obtaining the roughness prediction model is as follows:
[0117] Determine a detection position of the pipeline, detect the detection position to obtain an initial roughness of the pipeline, and the initial roughness is obtained by acquiring and identifying an image of the pipeline;
[0118] Performing a pulsating fluid experiment on the pipeline, and collecting corresponding pulsating flow parameter sets in units of flow cycles;
[0119] Then, the image recognition and acquisition of the experimental pipeline were performed to obtain the experimental roughness;
[0120] The initialization model is trained according to the initial roughness, the pulsating flow parameter set and the experimental roughness to obtain a roughness prediction model; wherein the time variation is reflected by the identification period of the pulsating flow parameter set.
[0121] The present invention identifies and quantifies the initial roughness of the inside of the pipeline through image recognition technology, then conducts a pulsating fluid experiment on the pipeline, and performs image recognition again on the inside of the pipeline after the experiment to obtain experimental roughness; the initial roughness and experimental roughness can accurately reflect the impact of the fluid on the pipeline roughness during the flow cycle, thereby training a roughness prediction model to achieve accurate prediction of the roughness inside the pipeline.
[0122] The method for obtaining the initial roughness through image recognition is:
[0123] Setting a detection length, and determining a region to be detected according to the detection length, a detection position of the pipeline, and a fluid flow direction, wherein the region to be detected is an inner wall of the pipeline;
[0124] like Figure 2 As shown, in the figure, a represents the pipeline, b represents the fluid flow direction, and the determined detection position is c;
[0125] The process of determining the area to be detected according to the detection length, the detection position of the pipeline and the fluid flow direction is as follows: taking the detection point c as the center and the fluid flow direction as the direction, the detection length de is obtained, and sections f and g of the pipeline a are made at the end points d and e of the line segment respectively, and the inner wall of the pipeline between the sections f and g is obtained as the area to be detected h.
[0126] Perform image recognition and splicing on the area to be detected to obtain an image of the area to be detected; the image of the area to be detected is a rectangle, the length of which is the inner circumference of the pipe section, and the width is the detection length;
[0127] like Figure 3 As shown, the image of the area to be detected h is captured to obtain an image h1 of the area to be detected. The image h1 of the area to be detected is a rectangle, the length of which is the circumference of the pipe sections f and g, and the width is the detection length line segment de.
[0128] The rectangular length of the image of the area to be detected is evenly divided to obtain multiple sub-area images; Figure 3 As shown in h2;
[0129] Then, an image recognition model is constructed to perform roughness recognition on the sub-region image to obtain the sub-region roughness; the image recognition model is constructed based on the YOLOv9 model;
[0130] The water level information of the detection position is monitored by a water level detection device, and the immersion time corresponding to the sub-area image is determined according to the water level information within the identification period; the water level detection device includes a radar water level gauge, a capacitive water level gauge, an ultrasonic water level gauge, a float water level gauge, etc.
[0131] weighting the sub-region roughness of the sub-region image according to the immersion time to obtain an initial roughness;
[0132] The calculation formula of the initial roughness is:
[0133]
[0134]
[0135] in, represents the initial roughness; Indicates the number of sub-areas divided; represents the immersion time of the sub-area; represents the roughness of the sub-region; represents the immersion time of the sub-area and
[0136] The present invention determines the area to be detected by detecting the position, the length and the flow direction of the fluid, divides the area to be detected into multiple sub-areas, obtains the roughness of the sub-areas by image recognition, and determines the immersion time of the sub-areas by a water level detection device; the roughness of the sub-areas is weighted according to the immersion time to obtain the initial roughness, thereby realizing accurate identification of the roughness of the pipeline.
[0137] S50. Obtain a first pipeline friction pressure drop according to the first pipeline roughness, pulsating flow parameter set and first detection data, and obtain a second pipeline friction pressure drop according to the second pipeline roughness, pulsating flow parameter set and second detection data; obtain first pressure drop loss data according to the friction pressure drop data, the first pipeline friction pressure drop and the second pipeline friction pressure drop, and store the obtained data in a database.
[0138] Inputting the first pipeline roughness, the first detection data and the pulsating flow parameter set into the Fanning formula to calculate the first pipeline friction pressure drop;
[0139]
[0140] in, represents the friction pressure drop of the first pipeline; represents Fanning's formula; represents the roughness of the first pipeline; represents the first detection data; represents the pulsating flow parameter set;
[0141] The second pipeline roughness, the second detection data and the pulsating flow parameter set are input into the Fanning formula to calculate the friction pressure drop of the second pipeline;
[0142]
[0143] in, represents the friction pressure drop of the second pipeline; represents Fanning's formula; Indicates the second pipeline roughness; represents the second detection data; represents the pulsating flow parameter set;
[0144] The first pressure drop loss data is calculated by the friction pressure drop data, the first pipeline friction pressure drop and the second pipeline friction pressure drop:
[0145]
[0146] in, Indicates the first pressure drop loss data.
[0147] The obtained pipeline roughness is substituted into the Fanning formula, and the first pipeline friction pressure drop and the second pipeline friction pressure drop are calculated by combining the first detection data, the first detection data and the pulsating flow parameter set; the first pressure drop loss data is accurately calculated through the friction pressure drop data, the first pipeline friction pressure drop and the second pipeline friction pressure drop.
[0148] S60. Construct a pressure drop loss prediction model to identify the pulsating flow parameter set and the valve detection data to obtain second pressure drop loss data.
[0149] After obtaining the valve detection data and the first pressure drop loss data at the rotating ball valve within the identification period, the present invention constructs a pressure drop loss prediction model based on the deep neural network DNN.
[0150] The training process of the pressure drop loss prediction model is:
[0151] Constructing a database, inputting the pulsating flow parameter set, the first detection data, the valve detection data, the second detection data and the first pressure drop loss data within the pipeline identification period into the database for storage;
[0152] Obtaining a pulsating flow parameter set, valve detection data, and first pressure drop loss data within an identification period from a database;
[0153] Obtain pipeline characteristics, fluid characteristics, pressure characteristics, vibration characteristics and sound characteristics through the pulsating flow parameter set and valve detection data, fit the first pressure drop loss data, and obtain a pressure loss prediction model;
[0154] The pulsating flow parameter set and valve detection data are identified through the pressure loss prediction model to obtain the second pressure drop loss data.
[0155] The present invention performs model training through pulsating flow parameter sets and valve detection data, and obtains a pressure loss prediction model by fitting pipeline characteristics, fluid characteristics, pressure characteristics, vibration characteristics and sound characteristics; the pressure loss prediction model can predict the pressure drop loss data of the rotary ball valve according to the pulsating flow parameter set and valve detection data within an identification period.
[0156] S70. Obtain an abnormality index of the rotary ball valve according to the first pressure drop loss data and the second pressure drop loss data, and issue an early warning according to the abnormality index.
[0157] The process of obtaining the abnormal index is as follows: Figure 4 As shown;
[0158] The abnormality index is calculated by the difference between the first pressure drop loss data and the second pressure drop loss data;
[0159] Among them, the first pressure drop loss data within the identification period is expressed as , the second voltage drop data is expressed as , by subtracting the second pressure drop loss data from the first pressure drop loss data, a third data set is obtained, which is recorded as ;
[0160] The calculation formula of the abnormal index is:
[0161]
[0162] in, Indicates the abnormal index within the identification period; represents the variance of the third data set; represents the variance threshold; express and distance; Indicates the distance threshold; Represents an exponential function with a natural constant as its base.
[0163] The present invention uses the pressure drop loss data obtained by collecting and calculating the detection point data as the first pressure drop loss data, and uses the pressure drop loss data predicted by the pressure loss prediction model as the second pressure drop loss data. The abnormality index is calculated through the comprehensive difference and individual differences between the first pressure drop loss data and the second pressure drop loss data, thereby realizing accurate identification and quantification of the abnormal situation of the rotary ball valve within the identification period.
[0164] The process of issuing an early warning based on the abnormal index is as follows:
[0165] An abnormality threshold is set through historical data of the pipeline, a threshold judgment is performed on the abnormality index according to the abnormality threshold, and an early warning message is issued when the abnormality index meets the threshold judgment.
[0166] The present invention sets an abnormal threshold value through historical data, performs threshold judgment on the abnormal index through the abnormal threshold value, and realizes accurate early warning of abnormal conditions of the rotary ball valve.
[0167] The method for predicting the pressure drop loss of a rotary ball valve under pulsating flow described in the present invention can accurately identify and predict the pressure drop loss of the rotary ball valve and determine whether there is an abnormality.
[0168] The present invention collects a pulsating flow parameter set, determines an identification period and a detection point according to the pulsating flow parameter set; detects the detection point within the identification period to obtain first detection data, valve detection data and second detection data; obtains friction pressure drop data according to the detection data; constructs a roughness prediction model to identify the pulsating flow parameter set and the detection data to obtain pipeline roughness, and further calculates the pipeline friction pressure drop; obtains first pressure drop loss data according to the friction pressure drop data and the pipeline friction pressure drop; constructs a pressure drop loss prediction model to identify the pulsating flow parameter set and the valve detection data to obtain second pressure drop loss data; obtains an abnormality index according to the first pressure drop loss data and the second pressure drop loss data, thereby realizing accurate identification and early warning of abnormalities of the rotary ball valve.
[0169] Embodiment 2
[0170] Raw material company A provides liquid chemical raw materials to enterprises through pipelines, and the raw material supply is cyclical on a daily basis and is regulated by an electric rotary ball valve.
[0171] As the rotary ball valve ages, abnormal conditions may occur under the impact and corrosion of the fluid, such as increased friction and wear, shape changes of component parts, corrosion and deposits, aging of springs and seals, etc. As a result, its pressure drop loss usually tends to increase, affecting the transportation and regulation of the pipeline.
[0172] Therefore, regular maintenance and overhaul are very important to keep the valve in normal operation and reduce pressure drop loss.
[0173] Since the transportation pipeline spans a long distance and is transported underground, the inspection and maintenance of the rotary ball valve is relatively complicated. Therefore, anomalies are identified by using a method for predicting pressure drop loss of a rotary ball valve under pulsating flow described in the present invention.
[0174] The process of the method for predicting the pressure drop loss of a rotary ball valve under pulsating flow is as follows: Figure 1 As shown, including:
[0175] S10. Collect data on the pulsating flow in the pipeline to obtain a pulsating flow parameter set, identify the flow period of the pulsating flow according to the pulsating flow parameter set; and determine an identification period according to the flow period.
[0176] Company A's transportation pipeline exhibits periodicity in days, so the identification period is set to 1 day.
[0177] The collected pulsating flow parameter set includes flow velocity data, pressure data, fluid density, fluid temperature, fluid viscosity and fluid corrosivity of the pipeline fluid.
[0178] S20. Determine a first detection point, a valve detection point, and a second detection point through a pulsating flow parameter set and a distance judgment model; collect data on the detection points within the identification period to obtain first detection data, valve detection data, and second detection data.
[0179] The distance judgment model includes a first distance judgment model and a second distance judgment model. The pulsating flow parameter set is identified by the first distance judgment model and the second distance judgment model to obtain a first distance and a second distance respectively.
[0180] Taking the rotary ball valve as a reference point, a first detection point and a second detection point are determined according to the first distance and the second distance. According to the fluid flow direction, the first detection point, the valve detection point and the second detection point are sequentially determined.
[0181] In the identification cycle, the time interval is set to 1 hour, so there are 24 time intervals in the identification cycle; data is collected according to the time interval to obtain the first detection data , the second detection data ; Valve test data .
[0182] S30. Calculate the friction pressure drop data between the first detection point and the second detection point based on the first detection data and the second detection data.
[0183] The total pressure drop between the first detection point and the second detection point includes three parts: gravity pressure drop, velocity pressure drop and friction pressure drop. Among them, the gravity pressure drop refers to the pressure drop caused by the height difference between the first detection point and the second detection point; the velocity pressure drop refers to the pressure drop caused by the change in flow velocity between the first detection point and the second detection point; the friction pressure drop refers to the pressure drop caused by the pipeline friction and the rotating ball valve between the first detection point and the second detection point.
[0184] The friction pressure drop can be obtained by subtracting the gravity pressure drop and velocity pressure drop from the total pressure drop; the formula is:
[0185]
[0186] in, is the total pressure drop between the first detection point and the second detection point; represents the gravity pressure drop; Indicates speed reduction; represents the friction pressure drop.
[0187] S40. Construct a roughness prediction model to identify the pulsating flow parameter set, the first detection data and the second detection data to obtain the first pipeline roughness and the second pipeline roughness.
[0188] The process of obtaining the roughness prediction model is as follows:
[0189] Determine a detection position of the pipeline, detect the detection position to obtain an initial roughness of the pipeline, and the initial roughness is obtained by acquiring and identifying an image of the pipeline;
[0190] Performing a pulsating fluid experiment on the pipeline, and collecting corresponding pulsating flow parameter sets in units of flow cycles;
[0191] Then, the image recognition and acquisition of the experimental pipeline were performed to obtain the experimental roughness;
[0192] The initialization model is trained according to the initial roughness, the pulsating flow parameter set and the experimental roughness to obtain a roughness prediction model; wherein the time variation is reflected by the identification period of the pulsating flow parameter set.
[0193] S50. Obtain a first pipeline friction pressure drop according to the first pipeline roughness, pulsating flow parameter set and first detection data, and obtain a second pipeline friction pressure drop according to the second pipeline roughness, pulsating flow parameter set and second detection data; obtain first pressure drop loss data according to the friction pressure drop data, the first pipeline friction pressure drop and the second pipeline friction pressure drop.
[0194] The pipeline friction pressure drop can be calculated through pipeline roughness, pulsating flow parameter set and detection data. The pressure drop loss caused by the rotating ball valve is obtained by subtracting the pipeline friction pressure drop from the friction pressure drop data, which is used as the first pressure drop loss data.
[0195] S60. Construct a pressure drop loss prediction model to identify the pulsating flow parameter set and the valve detection data to obtain second pressure drop loss data.
[0196] After obtaining the valve detection data and the first pressure drop loss data at the rotating ball valve within the identification period, the present invention constructs a pressure drop loss prediction model based on the deep neural network DNN.
[0197] The training process of the pressure drop loss prediction model is:
[0198] Constructing a database, inputting the pulsating flow parameter set, the first detection data, the valve detection data, the second detection data and the first pressure drop loss data within the pipeline identification period into the database for storage;
[0199] Obtaining a pulsating flow parameter set, valve detection data, and first pressure drop loss data within an identification period from a database;
[0200] Obtain pipeline characteristics, fluid characteristics, pressure characteristics, vibration characteristics and sound characteristics through the pulsating flow parameter set and valve detection data, fit the first pressure drop loss data, and obtain a pressure loss prediction model;
[0201] The pulsating flow parameter set and valve detection data are identified through the pressure loss prediction model to obtain the second pressure drop loss data.
[0202] S70. Obtain an abnormality index of the rotary ball valve according to the first pressure drop loss data and the second pressure drop loss data, and issue an early warning according to the abnormality index.
[0203] The abnormality index is calculated by the difference between the first pressure drop loss data and the second pressure drop loss data;
[0204] Among them, the first pressure drop loss data within the identification period is expressed as , the second voltage drop data is expressed as , by subtracting the second pressure drop loss data from the first pressure drop loss data, a third data set is obtained, which is recorded as ;
[0205] The calculation formula of the abnormal index is:
[0206]
[0207] in, Indicates the abnormal index within the identification period; represents the variance function; represents a third data set, obtained by correspondingly subtracting the second pressure drop loss data from the first pressure drop loss data; represents the variance threshold; express and distance; Indicates the distance threshold; Indicated by natural constant An exponential function with base .
[0208] The present invention collects a pulsating flow parameter set, determines an identification period and a detection point according to the pulsating flow parameter set; detects the detection point within the identification period to obtain first detection data, valve detection data and second detection data; obtains friction pressure drop data according to the detection data; constructs a roughness prediction model to identify the pulsating flow parameter set and the detection data to obtain pipeline roughness, and further calculates the pipeline friction pressure drop; obtains first pressure drop loss data according to the friction pressure drop data and the pipeline friction pressure drop; constructs a pressure drop loss prediction model to identify the pulsating flow parameter set and the valve detection data to obtain second pressure drop loss data; obtains an abnormality index according to the first pressure drop loss data and the second pressure drop loss data, thereby realizing accurate identification and early warning of abnormalities of the rotary ball valve.
[0209] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. A method for predicting pressure drop loss of a rotary ball valve under pulsating flow, characterized in that: include: S10. Collecting data on the pulsating flow in the pipeline to obtain a pulsating flow parameter set, identifying a flow period of the pulsating flow according to the pulsating flow parameter set; and determining an identification period according to the flow period; S20. Determine the first detection point, the valve detection point, and the second detection point by using the pulsating flow parameter set and the distance judgment model; collect data on the detection points within the identification period to obtain first detection data, valve detection data, and second detection data; S30. Calculate the friction pressure drop data between the first detection point and the second detection point according to the first detection data and the second detection data; S40. Constructing a roughness prediction model to identify the pulsating flow parameter set, the first detection data and the second detection data, and obtaining the first pipeline roughness and the second pipeline roughness; S50. Obtain a first pipeline friction pressure drop according to the first pipeline roughness, the pulsating flow parameter set and the first detection data, and obtain a second pipeline friction pressure drop according to the second pipeline roughness, the pulsating flow parameter set and the second detection data; obtain first pressure drop loss data according to the friction pressure drop data, the first pipeline friction pressure drop and the second pipeline friction pressure drop; and store the obtained data in a database; S60. Constructing a pressure drop loss prediction model to identify the pulsating flow parameter set and the valve detection data to obtain second pressure drop loss data; S70. Obtain an abnormality index of the rotary ball valve according to the first pressure drop loss data and the second pressure drop loss data, and issue an early warning according to the abnormality index.
2. The method for predicting pressure drop loss of a rotary ball valve under pulsating flow according to claim 1, characterized in that: The pulsating flow parameter set includes flow velocity data, pressure data, fluid density, fluid temperature, fluid viscosity and fluid corrosivity of the pipeline fluid; The steps to identify the flow period based on the pulsating flow parameter set are: Preprocess the flow rate data and pressure data to remove data noise and interference; Performing Fourier transform on the preprocessed flow velocity data and pressure data to convert the time domain features in the flow velocity data and pressure data into the frequency domain; The flow period of the pulsating flow is obtained by performing period identification on the Fourier transformed flow velocity data and pressure data.
3. The method for predicting pressure drop loss of a rotary ball valve under pulsating flow according to claim 1, characterized in that: The first detection point and the second detection point are set as follows: In the identification cycle, taking the rotary ball valve as a starting point, identifying the fluid cross-sectional area in front of the rotary ball valve; The fluid cross-sectional area is compared with the pipeline cross-sectional area in the order of being away from the rotary ball valve, and when the comparison result meets the threshold, the distance between the position of the fluid cross-sectional area and the position of the rotary ball valve is taken as the influence distance; Obtaining all influence distances within the identification period, and taking the maximum value of the influence distances as the maximum influence distance; The flow velocity data and pressure data at the rotary ball valve are collected, and the maximum impact distance is trained and fitted based on the fluid density, fluid temperature, fluid viscosity and fluid corrosivity to obtain the first distance judgment model; The cross-sectional area of the fluid behind the rotating valve within the identification period is identified in the same manner; And obtain the second distance judgment model through model training; The first distance and the second distance are obtained by using the first distance judgment model and the second distance judgment model, and the first detection point and the second detection point are determined according to the first distance, the second distance and the position of the rotating ball valve.
4. The method for predicting pressure drop loss of a rotary ball valve under pulsating flow according to claim 1, characterized in that: The first detection data includes: a first initial roughness, a first distance, a first pressure, a first flow rate and a pipe diameter; the second detection data includes: a second initial roughness, a second distance, a second pressure, a second flow rate and a pipe diameter; the valve detection data includes: valve pressure, valve flow rate, valve vibration data and valve sound data.
5. The method for predicting pressure drop loss of a rotary ball valve under pulsating flow according to claim 1, characterized in that: The method for obtaining the friction pressure drop data is: in, is the total pressure drop between the first detection point and the second detection point, which is measured by the pressure sensor; represents the gravity pressure drop, which refers to the pressure drop caused by the height difference between the first detection point and the second detection point; Indicates velocity pressure drop, which refers to the pressure drop caused by the change in flow velocity between the first detection point and the second detection point; Indicates the friction pressure drop data, which refers to the pressure drop caused by the pipe friction and the rotating ball valve between the first test point and the second test point.
6. The method for predicting pressure drop loss of a rotary ball valve under pulsating flow according to claim 1, characterized in that: The process of obtaining the roughness prediction model is as follows: Determine a detection position of the pipeline, and detect the detection position to obtain an initial roughness of the pipeline; the initial roughness is obtained by acquiring and identifying an image of the pipeline; Performing a pulsating fluid experiment on the pipeline, and collecting corresponding pulsating flow parameter sets in units of flow cycles; Then, the image of the experimental pipe is acquired and recognized to obtain the experimental roughness; The initialization model is trained according to the initial roughness, the pulsating flow parameter set and the experimental roughness to obtain a roughness prediction model.
7. The method for predicting pressure drop loss of a rotary ball valve under pulsating flow according to claim 6, characterized in that: The method for obtaining the initial roughness through image recognition is: Setting a detection length, and determining a region to be detected according to the detection length, a detection position of the pipeline, and a fluid flow direction, wherein the region to be detected is an inner wall of the pipeline; Perform image recognition and splicing on the area to be detected to obtain an image of the area to be detected; the image of the area to be detected is a rectangle, the length of the rectangle is the circumference of the pipe section circle, and the width of the rectangle is the detection length; Evenly splitting the rectangular length of the image of the area to be detected to obtain a plurality of sub-area images; Constructing an image recognition model to perform roughness recognition on the sub-region image to obtain the sub-region roughness; Monitoring the water level information of the detection position by a water level detection device, and determining the immersion time corresponding to the sub-region image according to the water level information within the recognition period; weighting the sub-region roughness of the sub-region image according to the immersion time to obtain an initial roughness; The calculation formula of the initial roughness is: in, represents the initial roughness; Indicates the number of sub-areas divided; represents the immersion time of the sub-area; represents the roughness of the sub-region; represents the immersion time and of the sub-area.
8. The method for predicting pressure drop loss of a rotary ball valve under pulsating flow according to claim 1, characterized in that: Inputting the first pipeline roughness, the first detection data and the pulsating flow parameter set into the Fanning formula to calculate the first pipeline friction pressure drop; in, represents the friction pressure drop of the first pipeline; represents Fanning's formula; represents the roughness of the first pipeline; represents the first detection data; represents the pulsating flow parameter set; Inputting the second pipeline roughness, the second detection data and the pulsating flow parameter set into the Fanning formula to calculate the friction pressure drop of the second pipeline; in, represents the friction pressure drop of the second pipeline; represents Fanning's formula; Indicates the second pipeline roughness; represents the second detection data; represents the pulsating flow parameter set; The first pressure drop loss data is calculated by the friction pressure drop data, the first pipeline friction pressure drop and the second pipeline friction pressure drop: in, Indicates the first pressure drop loss data, Represents friction pressure drop data.
9. The method for predicting pressure drop loss of a rotary ball valve under pulsating flow according to claim 1, characterized in that: The training process of the pressure drop loss prediction model is: Construct a database, input the pulsating flow parameter set, the first detection data, the valve detection data, the second detection data and the first pressure drop loss data within the pipeline identification cycle into the database for storage; obtain the pulsating flow parameter set, the valve detection data and the first pressure drop loss data within the historical identification cycle from the database; Obtain pipeline characteristics, fluid characteristics, pressure characteristics, vibration characteristics and sound characteristics through the pulsating flow parameter set and valve detection data, fit the first pressure drop loss data, and obtain a pressure loss prediction model; The pulsating flow parameter set and valve detection data are identified by the pressure loss prediction model to obtain the second pressure drop loss data.
10. The method for predicting pressure drop loss of a rotary ball valve under pulsating flow according to claim 1, characterized in that: The abnormality index is calculated by the difference between the first pressure drop loss data and the second pressure drop loss data; The calculation formula of the abnormal index is: in, Indicates the abnormal index within the identification period; represents the variance function; represents a third data set, obtained by correspondingly subtracting the second pressure drop loss data from the first pressure drop loss data; represents the variance threshold; express and distance; represents the distance threshold; It represents the exponential function with the natural constant e as the base; Indicates the first pressure drop loss data, Indicates the second pressure drop loss data.
