A valve life prediction method and system

Abstract

The present invention relates to the field of valve life prediction, and more specifically, the present invention relates to a valve life prediction method and system, including: collecting valve pressure drop data and flow data flowing through the valve, determining distorted data points in the pressure drop data, and selecting data points before the distorted data points to form a historical data segment; according to the severity of the change in the pressure drop data in the historical data segment of any distorted data point, calculating the initial demand degree of the distorted data point for the number of interpolation. The present invention can more accurately determine the interpolation demand degree of each distorted data point by analyzing the historical data before each distorted data point, especially the numerical characteristics of the pressure drop and flow data, and adaptively adjust the number of interpolation, so that the number of historical data referenced when calculating the distorted data point is more appropriate and effective, which helps to reduce the interference of noise data in the interpolation process, thereby improving the accuracy and stability of the final interpolation result.

Description

Technical Field

[0001] The present invention relates to the field of valve life prediction, and more specifically, to a valve life prediction method and system. Background Art

[0002] In the industrial production process, valves are key control and regulation devices, and their performance directly affects process stability and production efficiency. Therefore, accurate prediction of valve life can help companies optimize equipment management and maintenance strategies, reduce maintenance costs, and improve production efficiency. Traditional scheduled maintenance may not be efficient, and the data-driven valve life prediction system can provide accurate maintenance recommendations based on actual operating conditions and prediction models. During the operation of the valve, the pressure drop (i.e., the pressure loss when the fluid passes through the valve) will cause the valve force to change. Therefore, by monitoring and analyzing the changes in pressure drop, the wear and aging of the internal structure of the valve can be indirectly reflected, thereby inferring the life status of the valve.

[0003] In the related technology, for example, the Chinese patent document with the authorization announcement number CN110962130B discloses a valve life prediction method based on big data analysis, which obtains the optimal segmentation threshold of each decision node according to the classification difference index of each segmentation threshold, and constructs a decision tree accordingly. By analyzing the classification effect of different dimensional data in the decision tree, the node concentration index of each decision tree is calculated, and the weight of each decision tree is constructed. The prediction results of the valve life at the new moment of each decision tree are combined to obtain the valve life level at the new moment. In this way, the valve life prediction is realized, and the accuracy of the random forest prediction of the valve life is improved.

[0004] However, when collecting and transmitting valve pressure data, network fluctuations, signal interference, equipment failure and other problems may cause some data loss or transmission failure, resulting in missing pressure data points, which in turn leads to distortion of pressure drop data, resulting in inaccurate prediction of valve life. Summary of the invention

[0005] The present invention provides a valve life prediction method and system, aiming to solve the problem in the related technology that network fluctuations, signal interference, equipment failure and other problems may cause partial data loss or transmission failure, resulting in missing pressure data points, and then causing distortion of pressure drop data, thereby resulting in inaccurate prediction of valve life.

[0006] In a first aspect, the present invention provides a valve life prediction method, comprising: collecting valve pressure drop data and flow data through the valve, determining distorted data points in the pressure drop data, and selecting the distorted data points before the distorted data points. data points constitute a historical data segment; according to the severity of the change in the pressure drop data in the historical data segment of any distorted data point, the initial demand degree of the distorted data point for the number of interpolation is calculated, and the initial demand degree is optimized according to the change characteristics of the flow data to obtain the final demand degree of the distorted data point for the number of interpolation, wherein the final demand degree The calculation formula is: ; In the formula, No. The initial requirement level of distorted data points, No. The Pearson correlation coefficient between the pressure drop data and the flow data in the corresponding time period of the historical data segment of the distorted data point is and Respectively The extreme difference of the pressure drop data and flow data in the corresponding time period of the historical data segment of the distorted data point in each historical data segment, and Respectively The pressure drop data and flow rate data in the historical data segments corresponding to the time period of the historical data segments of the distorted data points are averaged in their respective historical data segments; based on the final demand degree of the distorted data points, the number of interpolation times is determined, and the final interpolation result is calculated, so as to predict the valve life using the pressure drop data after the interpolation.

[0007] In one embodiment, the valve life is predicted using the interpolated pressure drop data, including: collecting the interpolated pressure drop data, fitting the Weibull distribution by statistical methods, and establishing a Weibull distribution model; and predicting the valve life according to the Weibull distribution model.

[0008] In one embodiment, the number of interpolation times is determined based on the size of the final demand degree of the distorted data point, including: dividing the final demand degree into at least one preset range, wherein each preset range corresponds to an interpolation number; if the final demand degree falls within the corresponding preset range, the corresponding interpolation number is obtained.

[0009] In one embodiment, the final interpolation result is calculated, including: determining the voltage drop data set required for the interpolation calculation according to the interpolation times of any distorted voltage drop data, and interpolating it using a time series analysis model to obtain the final interpolation result.

[0010] In one embodiment, a time series analysis model is used to interpolate the data to obtain a final interpolation result, including: in response to the interpolation number of the distorted voltage drop data being 1, the time series analysis model is used to interpolate the distorted voltage drop data to obtain a final interpolation result; in response to the interpolation number of the distorted voltage drop data being greater than 1, the time series analysis model is used to interpolate all data in the voltage drop data set corresponding to the distorted voltage drop data, and the average of all interpolation results is used as the final interpolation result.

[0011] In one embodiment, the initial demand level The calculation formula is: ; In the formula, Indicates that before each distorted data point is selected The historical data segment consists of data points. and Respectively represent The distorted data point in the historical data segment and The pressure drop value of the pressure drop data, Indicates The information entropy value of the voltage drop value at all sampling moments in the historical data segment of the distorted data point.

[0012] In one embodiment, collecting the pressure drop data of the valve includes: using a pressure sensor to collect the pressure data on both sides of the valve, and calculating the difference between the pressure data on both sides to obtain the pressure drop data.

[0013] In one embodiment, collecting the pressure drop data of the valve and the flow data flowing through the valve further includes: performing curve fitting on the pressure drop data and the flow data.

[0014] In a second aspect of the present invention, a valve life prediction system is provided, comprising a processor and a memory, wherein the memory stores a computer program, and the processor executes the computer program to implement any of the valve life prediction methods described above.

[0015] Beneficial Effects

[0016] (1) By analyzing the historical data before each distorted data point, especially the numerical characteristics of the pressure drop and flow data, the interpolation requirement of each distorted data point can be determined more precisely, and the number of interpolation times can be adaptively adjusted, so that the number of historical data referenced when calculating the distorted data point is more appropriate and effective, which helps to reduce the interference of noise data during the interpolation process, thereby improving the accuracy and stability of the final interpolation result. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The following detailed description is read with reference to the accompanying drawings, which illustrate several embodiments of the present invention in an exemplary and non-limiting manner, and in which like or corresponding reference numerals represent like or corresponding parts, wherein:

[0018] Figure 1 is a flow chart schematically illustrating valve life prediction according to an embodiment of the present invention;

[0019] Figure 2 FIG. 1 is a schematic diagram schematically showing a system structure according to an embodiment of the present invention. DETAILED DESCRIPTION

[0020] 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 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0021] like Figure 1 As shown, a valve life prediction method is described in detail below in conjunction with the accompanying drawings, specifically including steps S101 to S105:

[0022] S101: Collecting the pressure drop data of the valve and the flow data of the valve.

[0023] In one embodiment, to obtain the pressure drop data of the valve, firstly, a pressure sensor is used to collect the pressure data on both sides of the valve, and then the pressure difference between the pressure data on both sides is calculated to obtain the pressure drop data. For the flow data flowing through the valve, the flow data can be obtained by a flow meter installed in the valve.

[0024] In one embodiment, the numerical characteristics of the pressure data points on both sides of the valve are checked to filter out all sampling moments with missing data. The pressure drop data corresponding to these sampling moments are all distorted data points, thereby determining the distorted data points in the pressure drop data, and then selecting N data points before the distorted data points to form a historical data segment.

[0025] Exemplarily, for any distorted data point, 40, 50 or 80 data points before the distorted data point may be selected to form a historical data segment.

[0026] S102: Calculate the initial requirement of the number of interpolation times for the distorted data points.

[0027] In one embodiment, the numerical variation characteristics of the historical pressure drop data within a certain time range before each distorted data point are analyzed to obtain the initial demand for the number of interpolation of each distorted data point. When analyzing this indicator, the more drastic the data change in the historical data segment of each distorted data point and the more data points with mutations therein, the greater the initial demand for the number of interpolation of the distorted data point. This is because when the historical data fluctuates greatly or the change trend is unstable, the number of interpolation times needs to be increased to better reflect the actual trend.

[0028] In one embodiment, the initial requirement for the number of interpolation times of the distorted data points is calculated. , the calculation formula is: ; In the formula, Indicates that before each distorted data point is selected The historical data segment consists of data points. and Respectively represent The distorted data point in the historical data segment and The pressure drop value of the pressure drop data, Indicates The information entropy value of the voltage drop value at all sampling moments in the historical data segment of the distorted data point.

[0029] in, Indicates The sum of the differences in the voltage drop values ​​between all adjacent sampling moments in the historical data segment with distorted data points. The larger the value, the The more dramatic the data changes in the historical data segment with distorted data points, the more potential data points with mutations there are. The greater the initial requirement for the number of interpolation times for each distorted data point. The larger the The more chaotic the distribution of the voltage drop values ​​at all sampling times in the historical data segment with distorted data points, the The more drastic the data changes in the historical data segment with distorted data points, the more potential data points with mutations, the greater the credibility, and the greater the possibility of noise data. The greater the initial requirement for the number of interpolation values, the more distorted data points there are.

[0030] In one embodiment, the preliminary demand for the number of interpolation times of each distorted data point is obtained. The analysis of this indicator is based on the pressure drop data itself, so this indicator cannot well reflect the demand for the number of interpolation times of a distorted data point. In the scenario involved in this solution, when the valve opening remains unchanged, the pressure difference before and after the valve (pressure drop data) directly affects the flow through the valve. The greater the pressure difference, the greater the flow rate will generally be. Therefore, this step can be combined with the analysis of the correlation between the change characteristics of the flow data and the change characteristics of the pressure drop data within a certain time range before the sampling moment of each distorted data point, and the preliminary demand for the number of interpolation times of each distorted data point can be optimized to obtain the demand degree.

[0031] Specifically, the weaker the connection between the flow data change characteristics and the pressure drop data change characteristics within a certain time range before the sampling time of each distorted data point, the greater the possibility of noise data in the historical data segment with the distorted data point, and the greater the demand for the number of interpolation times for the pressure drop data point with the distorted data point. This is because the existence of noise data points will weaken the correlation between the pressure drop data and the flow data in the historical data segment with the distorted data point in the corresponding time period. Therefore, the preliminary demand degree is optimized according to the above characteristics, and the optimization process is as follows.

[0032] S103: Optimize the preliminary demand level to obtain the final demand level of the distorted data point for the number of interpolation times.

[0033] In one embodiment, the calculation formula of the final demand degree is: ; In the formula Indicates the final degree of interpolation required for the distorted data points, Indicates the initial degree of interpolation requirements for distorted data points. Indicates The Pearson correlation coefficient between the pressure drop data and the flow data in the corresponding time period of the historical data segment of the distorted data point is and Respectively represent The extreme difference of the pressure drop data and flow data in the corresponding time period of the historical data segment of the distorted data point in each historical data segment, and Respectively represent The historical data segment of the distorted data point is the mean value of the pressure drop data and flow data in each historical data segment in the corresponding time period.

[0034] Among them, the formula The larger the The stronger the correlation between the pressure drop data and the flow data in the corresponding time period of the historical data segment with distorted data points, the The smaller the probability that there is noise data in the historical data segment with distorted data points, the smaller the probability that there is noise data in the historical data segment with distorted data points. The more distorted data points there are, the less interpolation times are needed. Indicates The ratio of the range to the mean in the historical data segment with distorted data points can reflect the The degree of data dispersion in a historical data segment with distorted data points. Indicates The ratio of the range to the mean in the traffic data segment in the corresponding time period of the historical data segment with distorted data points can reflect the The degree of data dispersion in the traffic data segment in the corresponding time period of the historical data segment with distorted data points. Indicates the difference between the two. The smaller the value, the The stronger the correlation between the pressure drop data and the flow data in the corresponding time period of the historical data segment with distorted data points, the stronger the correlation between the pressure drop data and the flow data in the corresponding time period. The greater the Pearson correlation coefficient between the pressure drop data and the flow data in the corresponding time period of the historical data segment with distorted data points, the greater the credibility. Then, the less likely it is that there is noise data in the xth historical data segment with distorted data points. The more distorted data points there are, the greater the need for interpolation times.

[0035] S104: Determine the number of interpolation times and calculate the final interpolation result.

[0036] In one embodiment, the above steps analyze and obtain the degree of requirement for the number of interpolation times at each distorted data point, and this step specifies the number of interpolation times required for each distorted data point based on the indicator.

[0037] For example, the final requirement for the number of interpolation times for the distorted voltage drop data is . And according to the final demand degree of distorted data points The size of determines the number of interpolation times, including:

[0038] like When , the interpolation times of the distorted voltage drop data is 1;

[0039] like When , the interpolation times of the distorted voltage drop data is 2;

[0040] like When , the interpolation times of the distorted voltage drop data is 3;

[0041] like When , the interpolation times of the distorted voltage drop data is 4.

[0042] In one embodiment, after obtaining the difference times of each distorted data point, the voltage drop data set required for the interpolation calculation is determined according to the interpolation times of any distorted voltage drop data, and the ARIMA model (time series analysis model) is used to interpolate it to obtain the final interpolation result.

[0043] When the interpolation times are specified to be 1, 2, 3 or 4, for the distorted data point, 50, 100, 150 or 200 voltage drop data points before the sampling time are selected to calculate the interpolation result.

[0044] Specifically, if the interpolation times of the distorted voltage drop data is 1, the ARIMA model is used to calculate the above 50 voltage drop data points to obtain the final interpolation result; if the interpolation times of the distorted voltage drop data is greater than 1, the ARIMA model is used to interpolate all the data in the voltage drop data set corresponding to the distorted voltage drop data, and the average of all interpolation results is used as the final interpolation result. The numerical interpolation of distorted data points using the ARIMA model is an existing technology and will not be elaborated here.

[0045] Exemplarily, when the interpolation times of the distorted voltage drop data is 1, the ARIMA model is used to calculate 50 voltage drop data points before the sampling time of the distorted voltage drop data to obtain the final interpolation result of the distorted voltage drop data.

[0046] When the interpolation times of the distorted voltage drop data is 2, the ARIMA model is used to calculate the 50 voltage drop data points before the sampling time of the distorted voltage drop data to obtain an interpolation result, and then the 100 voltage drop data points before the sampling time of the distorted voltage drop data are calculated to obtain an interpolation result, and the average value of the two interpolation results is calculated, and the average value is used as the final interpolation result of the distorted voltage drop data;

[0047] When the interpolation times of the distorted voltage drop data is 2, the ARIMA model is used to calculate 50 voltage drop data points before the sampling time of the distorted voltage drop data to obtain an interpolation result, and then the 100 voltage drop data points before the sampling time of the distorted voltage drop data are calculated, and then the 150 voltage drop data points before the sampling time of the distorted voltage drop data are calculated to obtain an interpolation result, and the average value of these three interpolation results is calculated, and the average value is used as the final interpolation result of the distorted voltage drop data;

[0048] When the interpolation times of the distorted voltage drop data is 4, the ARIMA model is used to calculate 50 voltage drop data points before the sampling time of the distorted voltage drop data to obtain an interpolation result, and then the 100 voltage drop data points before the sampling time of the distorted voltage drop data are calculated, and then the 150 voltage drop data points before the sampling time of the distorted voltage drop data are calculated, and then the 200 voltage drop data points before the sampling time of the distorted voltage drop data are calculated to obtain an interpolation result, and the average value of these four interpolation results is calculated, and the average value is used as the final interpolation result of the distorted voltage drop data.

[0049] S105: Predicting the valve life using the pressure drop data after interpolation.

[0050] In one embodiment, the valve life is predicted using the Weibull distribution model based on the pressure drop data after interpolation, and the specific method is as follows:

[0051] Collect the real-time pressure drop data after denoising, and collect the historical pressure drop data of valve failure or maintenance, and record the pressure drop data at each failure. The pressure drop data at the time of failure can help determine the specific mode or cause of valve failure. Different pressure drop modes may be associated with different failure mechanisms, such as wear, leakage, blockage, etc. By analyzing the pressure drop data at the time of failure, the pressure changes and operating conditions to which the valve is subjected during use can be revealed, and its life characteristics can be understood;

[0052] In one embodiment, the collected data is used to fit the Weibull distribution through a statistical method, wherein the preset shape parameter is 1 and the scale parameter is 3. Finally, the valve life is predicted based on the established Weibull distribution model.

[0053] According to the above steps, by analyzing the historical data before each distorted data point, especially the numerical characteristics of the pressure drop and flow data, the degree of interpolation requirement for each distorted data point can be determined more precisely. For cases where the historical data changes relatively smoothly or the periodic changes are obvious, the number of interpolation times can be appropriately reduced; for cases where the historical data fluctuates greatly or the trend of change is unstable, the number of interpolation times can be increased to better reflect the actual trend. The number of interpolation times is adaptively adjusted so that the number of historical data referenced when calculating the distorted data points is more appropriate and effective, which helps to reduce the interference of noise data during the interpolation process, thereby improving the accuracy and stability of the final interpolation result.

[0054] The present invention also provides a valve life prediction system. Figure 2As shown, the system includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a valve life prediction method according to the first aspect of the present invention is implemented.

[0055] The system also includes other components familiar to those skilled in the art, such as a communication bus and a communication interface, whose configuration and functions are known in the art and thus will not be described in detail here.

[0056] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in the present invention may be implemented using computer-readable / executable instructions that may be stored or otherwise maintained by such a computer-readable medium.

[0057] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0058] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0059] The above-mentioned embodiments only express several implementation methods of the present invention, and the description is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent application. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A valve life prediction method, characterized in that: include: Collect the pressure drop data of the valve and the flow data through the valve, determine the distorted data points in the pressure drop data, and select the data points before the distorted data points. Data points constitute a historical data segment; According to the severity of the change in the voltage drop data in the historical data segment of any distorted data point, the initial demand for the number of interpolation times of the distorted data point is calculated. , the calculation formula is: ; In the formula, Indicates that before each distorted data point is selected The historical data segment consists of data points. and Respectively represent The distorted data point in the historical data segment and The pressure drop value of the pressure drop data, Indicates The information entropy value of the pressure drop value at all sampling moments in the historical data segment of the distorted data point; and according to the change characteristics of the flow data, the preliminary demand degree is optimized to obtain the final demand degree of the distorted data point for the number of interpolation times, where the final demand degree The calculation formula is: ； In the formula, No. The initial requirement level of distorted data points, No. The Pearson correlation coefficient between the pressure drop data and the flow data in the corresponding time period of the historical data segment of the distorted data point is and Respectively The extreme difference of the pressure drop data and flow data in the corresponding time period of the historical data segment of the distorted data point in each historical data segment, and Respectively The mean of the pressure drop data and flow data in the historical data segment corresponding to the time period of the historical data segment of the distorted data point, is the normalization function; Based on the final demand level of the distorted data points, the number of interpolation times is determined, and the final interpolation result is calculated, so as to predict the valve life using the pressure drop data after the interpolation.

2. The valve life prediction method according to claim 1, characterized in that: The interpolated pressure drop data is used to predict valve life, including: Collect the interpolated pressure drop data, fit the Weibull distribution by statistical method, and establish the Weibull distribution model; The valve life is predicted according to the Weibull distribution model.

3. The valve life prediction method according to claim 1, characterized in that: Based on the final required degree of the distorted data point, the number of interpolation times is determined, including: Dividing the final demand degree into at least one preset range, wherein each preset range corresponds to an interpolation number; If the final demand level falls within the corresponding preset range, the corresponding interpolation times are obtained.

4. The valve life prediction method according to claim 1, characterized in that: Determine the number of interpolation times and calculate the final interpolation result, including: According to the interpolation times of any distorted voltage drop data, the voltage drop data set required for the interpolation calculation is determined, and the data is interpolated using the time series analysis model to obtain the final interpolation result.

5. The valve life prediction method according to claim 4, characterized in that: Use the time series analysis model to interpolate it and get the final interpolation results, including: In response to the interpolation number of the distorted voltage drop data being 1, the distorted voltage drop data is interpolated using a time series analysis model to obtain a final interpolation result; In response to the interpolation times of the distorted voltage drop data being greater than 1, all data in the voltage drop data set corresponding to the distorted voltage drop data are interpolated using the time series analysis model, and the average of all interpolation results is used as the final interpolation result.

6. The valve life prediction method according to claim 1, characterized in that: Collect valve pressure drop data, including: The pressure sensor is used to collect the pressure data on both sides of the valve, and the difference between the pressure data on both sides is calculated to obtain the pressure drop data.

7. The valve life prediction method according to claim 1, characterized in that: Collects data on the pressure drop across the valve and the flow rate through the valve, including: Curve fitting is performed on the pressure drop data and flow rate data.

8. A valve life prediction system, comprising a processor and a memory, characterized in that: The memory stores a computer program, and the processor executes the computer program to implement the valve life prediction method according to any one of claims 1 to 7.

Images