Method for testing fatigue strength of valve clack of copper valve based on intelligent sensor

Abstract

The invention relates to the technical field of mechanical structure fatigue testing, and discloses a copper valve clack fatigue strength testing method based on an intelligent sensor. Pressure sensors are arranged at an inlet and an outlet of the valve, pressure in the whole opening and closing process is continuously collected to form an original time sequence, and pulsating pressure is obtained by subtracting the average pressure in the whole time period; an overlapping analysis window is constructed based on pulsating pressure, frequency energy of each window is solved by using a preset frequency mapping function and weighted to obtain a frequency spectrum gravity center and statistical characteristics of the frequency spectrum gravity center changing along with the window, and accordingly, a single-side fatigue damage index of each sensor is constructed and a most unfavorable measuring point is selected as an overall damage index. And combining the pulsating pressure root-mean-square value to obtain a fatigue strength evaluation value. Through the process, pressure fluctuation of the valve under all working conditions is reflected, static bias and noise influences are restrained, fatigue related frequency energy is highlighted, and online quantitative evaluation and early warning of the fatigue evolution degree of the valve clack are achieved.

Description

Technical Field

[0001] This invention relates to the field of mechanical structure fatigue testing technology, specifically to a method for testing the fatigue strength of a copper valve disc based on an intelligent sensor. Background Technology

[0002] In the field of mechanical engineering, especially during the long-term use of valves and their accessories, the fatigue strength of the valve disc is a key parameter for evaluating valve reliability and service life. Valve discs undergo cyclic loading under high pressure, particularly in industries such as petrochemicals, chemicals, and energy, where valves experience significant pressure pulsations during operation, leading to the accumulation of fatigue damage. Traditional methods for testing valve disc fatigue strength typically rely on physical experiments, often depending on coarse static analysis or finite element analysis. These methods cannot monitor changes in valve condition in real time and often neglect the impact of pulsating loads in the working environment on the valve disc's fatigue strength.

[0003] A common fatigue testing method in existing technologies uses vibration sensors or strain gauges to estimate the fatigue life of a valve disc by directly measuring its strain changes. These methods are generally limited to measuring a single physical quantity, often providing only limited information about the valve disc's stress state. Furthermore, the testing process needs to be conducted in a laboratory environment, failing to reflect the valve disc's fatigue process under actual operating conditions in real time. In addition, due to the various complex stress states and non-uniform loads experienced by the valve disc during operation, traditional methods often cannot comprehensively reflect the actual fatigue damage of the valve disc under complex operating conditions. Moreover, most current detection methods rely on externally applied loading or offline testing, failing to provide timely and continuous monitoring of field equipment and making it difficult to dynamically assess the valve's fatigue damage state. This results in the inability to provide early warning or prediction when valve disc fatigue failure occurs, severely impacting the reliability and safety of the equipment. With the rapid development of intelligent sensor technology, especially the advancements in pressure sensors, vibration sensors, and their supporting analysis algorithms, more and more real-time field monitoring technologies have emerged. Currently, pressure sensors are widely used to monitor pressure fluctuations in liquid fluid pipelines; however, these sensors are generally only used for basic pressure measurement, lacking in-depth mining and application of pressure pulsation information. Therefore, valve fatigue testing methods based on intelligent sensors, especially the extraction of fatigue strength information from pressure pulsation signals, are gradually becoming a promising research direction. Although existing technologies have made progress in pressure detection and vibration monitoring using sensors, real-time analysis of pressure pulsation signals, extraction of spectral features, and assessment of valve disc fatigue status based on these features are still in their infancy. Existing technologies have failed to fully utilize spectral analysis techniques to conduct in-depth analysis of the fatigue characteristics of valve discs under complex operating conditions, resulting in insufficient information for equipment maintenance and fault prediction.

[0004] Therefore, this study aims to propose a method for testing the fatigue strength of copper valve discs based on intelligent sensors. First, high-precision pressure sensors are deployed at the valve inlet and outlet to collect pressure signals throughout the opening and closing process in real time, which are then converted into raw pressure time series. Next, the static component is stripped to obtain a pulsating pressure time series. Based on this series, overlapping sliding windows are used for segmented analysis, ensuring temporal continuity in frequency domain feature extraction. Then, a preset mapping function is used to perform spectral energy mapping on the pressure samples within each window to obtain the frequency energy distribution. Normalization and weighted calculations are then used to obtain the spectral centroid of each window. Based on this, the average value and dispersion of the centroid's evolution over time are statistically analyzed to construct a unilateral fatigue damage index. Finally, the valve disc fatigue strength is comprehensively evaluated by combining this index with effective pressure amplitude. Summary of the Invention

[0005] This invention provides a method for testing the fatigue strength of copper valve discs based on intelligent sensors, which helps to solve the problems mentioned in the background art.

[0006] This invention provides the following technical solution: a method for testing the fatigue strength of a copper valve disc based on a smart sensor, comprising:

[0007] Pressure sensors were installed at the inlet and outlet pipe sections of the copper valve under test. Uniform sampling parameters were set, and pressure measurements were continuously collected throughout the opening and closing process to form the original pressure time series.

[0008] The average pressure was obtained across the entire sampling interval. At each sampling point, the average pressure was subtracted from the original pressure to obtain the pulsating pressure time series.

[0009] Based on the pulsating pressure time series, the number of sampling points and the sliding step size are set in the window to construct a sequence of analysis windows that partially overlap, and a continuous sample sequence is formed within each window;

[0010] Based on the preset discrete frequency mapping function sequence, the sample sequences of each window are linearly transformed and accumulated to obtain the linear projection value and spectral energy quantization value distribution of each discrete frequency;

[0011] The total spectral energy is obtained by accumulating the quantized values ​​of spectral energy in each window, normalizing it, and weighting it according to the discrete frequency index to calculate the spectral centroid.

[0012] Calculate the change in the centroid of the spectrum between adjacent analysis windows, form a sequence of changes according to the window order, and obtain its arithmetic mean and the average of squared deviations;

[0013] The single-sided fatigue damage index of each pressure sensor is constructed based on the arithmetic mean and the average of squared deviations. When there are multiple pressure sensors, the index with the larger value is selected as the overall fatigue damage index.

[0014] The root mean square value of the pulsating pressure is calculated from the time series of the pulsating pressure of each sensor. The larger value is taken as the equivalent pulsating pressure amplitude. This value is then combined with the overall fatigue damage index according to a preset relationship to obtain the fatigue strength assessment value.

[0015] Optionally, the step of installing pressure sensors on the inlet and outlet pipe sections of the copper valve under test, setting uniform sampling parameters, and continuously collecting pressure measurements throughout the opening and closing process to form an original pressure time series specifically includes:

[0016] A first pressure sensor is installed in the inlet pipe section of the copper valve under test, and a second pressure sensor is installed in the outlet pipe section of the copper valve under test. They are respectively numbered as the first pressure sensor and the second pressure sensor. The first pressure sensor collects the pressure measurement value at the inlet of the copper valve under test, and the second pressure sensor collects the pressure measurement value at the outlet of the copper valve under test.

[0017] Set a uniform sampling frequency, and set the total number of sampling points for each pressure sensor to be no less than four. Calculate the total continuous sampling time based on the total number of sampling points and the uniform sampling frequency, so that the continuous sampling process covers the entire working condition change process of the copper valve under test from fully closed to fully open and then from fully open to fully closed.

[0018] Based on the uniform sampling frequency and the total continuous sampling duration, the entire sampling period is evenly divided into several discrete sampling moments, and the discrete sampling moments are numbered according to the time sequence to obtain the sampling point sequence from the first sampling point to the last sampling point.

[0019] At each discrete sampling moment, the pressure measurement values ​​of the first pressure sensor and the second pressure sensor are recorded respectively. The pressure measurement values ​​of the two pressure sensors are arranged in the order of the sampling point sequence to form the original pressure time series of the first pressure sensor and the second pressure sensor. The number of sampling points in each original pressure time series is equal to the total number of sampling points.

[0020] During continuous sampling, the copper valve under test is kept fully closed during the sampling period corresponding to the first quarter of the sampling point sequence. Before the sampling time at the first quarter position in the sampling point sequence, the copper valve under test is switched from fully closed to fully open. During the sampling period corresponding to the first quarter to the half position in the sampling point sequence, the copper valve under test is kept fully open. Before the sampling time at the half position in the sampling point sequence, the copper valve under test is switched from fully open to fully closed. During the sampling period corresponding to the last half of the sampling point sequence, the copper valve under test is kept fully closed.

[0021] Optionally, the step of obtaining the average pressure across the entire sampling interval, and subtracting the average pressure from the original pressure at each sampling point to obtain the pulsating pressure time series, specifically includes:

[0022] For each pressure sensor, iterate through all the pressure values ​​of the original pressure time series of each pressure sensor throughout the entire sampling time interval, add up all the pressure values ​​of the sampling points and divide by the total number of sampling points to obtain the average pressure of each pressure sensor during the entire sampling period.

[0023] For each pressure sensor, at each sampling point, the average pressure of the corresponding pressure sensor is subtracted from the sampling point pressure value in the corresponding original pressure time series. The difference is used as the pulsating pressure value of the corresponding sampling point. The pulsating pressure time series of the corresponding pressure sensor is constructed by arranging the sampling points in the order of sampling points, so that each pulsating pressure time series is consistent with the corresponding original pressure time series in terms of the number and order of sampling points.

[0024] Optionally, the step of setting the number of sampling points and the sliding step size based on the pulsating pressure time series to construct a sequence of partially overlapping analysis windows, and forming a continuous sample sequence within each window, specifically includes:

[0025] The number of sampling points in each analysis window is set according to the total number of sampling points in the pulsating pressure time series, so that the number of sampling points in the analysis window is not less than the preset minimum number of sampling points in the window and not less than one-third of the total number of sampling points, and the set number of sampling points is used as the number of sampling points in the window.

[0026] Based on the number of sampling points in the window and the uniform sampling frequency, the time length corresponding to a single analysis window is calculated to obtain the time length parameter of the analysis window.

[0027] Set the sliding step size between the starting sampling points of adjacent analysis windows to half the number of sampling points in the window, and ensure that the sliding step size is not less than one sampling point interval and not greater than the number of sampling points in the window.

[0028] Under the condition that the total number of sampling points is not less than the sum of the number of window sampling points and twice the sliding step size, and the total number of analysis windows constructed by the sliding step size is not less than three, starting from the first sampling point of the pulsating pressure time series, the starting sampling point of the analysis window is moved sequentially according to the sliding step size, and each time a continuous pulsating pressure sample containing the number of window sampling points is extracted, and the number of analysis windows obtained is used as the total number of windows that can be constructed.

[0029] For each pressure sensor and each analysis window, continuous pulsating pressure samples are extracted in the sampling order within each analysis window. A sampling position number is assigned to each sampling position within each analysis window. The pulsating pressure samples corresponding to each sampling position number are arranged in the order of the sampling position number to form the pulsating pressure sample sequence of the corresponding pressure sensor in each analysis window.

[0030] Optionally, the step of performing a linear transformation and accumulating the sample sequences of each window based on a preset discrete frequency mapping function sequence to obtain the linear projection value and spectral energy quantization value distribution of each discrete frequency specifically includes:

[0031] The number of discrete frequencies is set according to the number of window sampling points, and a discrete frequency index is assigned to each discrete frequency. A preset discrete frequency mapping function sequence is used to define the function value of each mapping function at each sampling position number by weighted superposition of cosine and sine terms, so that the length of each mapping function is equal to the number of window sampling points.

[0032] For each pressure sensor, each analysis window, and each discrete frequency index, the pulsating pressure sample value corresponding to each sampling position number in each analysis window is multiplied one by one with the function value of the corresponding discrete frequency mapping function at the same sampling position number, and then accumulated sequentially along the sampling position number to obtain the linear projection value of the corresponding discrete frequency index of the corresponding pressure sensor under each analysis window.

[0033] For each pressure sensor, each analysis window, and each discrete frequency index, the corresponding linear projection value is multiplied by the corresponding linear projection value, and the resulting product is used as the spectral energy quantization value of the corresponding pressure sensor at the corresponding discrete frequency index under each analysis window.

[0034] For each pressure sensor, the quantized values ​​of each spectrum energy are filled into the corresponding positions one by one according to the analysis window number as the row index and the discrete frequency index as the column index, thus forming the spectrum energy matrix of the corresponding pressure sensor.

[0035] Optionally, the total spectral energy obtained by accumulating the quantized spectral energy within each window, normalizing it, and weighting it according to the discrete frequency index to calculate the spectral centroid quantity specifically includes:

[0036] For each pressure sensor and each analysis window, the quantized values ​​of the spectral energy corresponding to all discrete frequencies are accumulated within each analysis window to obtain the total spectral energy of the corresponding pressure sensor in each analysis window.

[0037] For each pressure sensor, each analysis window, and each discrete frequency, when the total spectral energy is greater than the preset energy lower limit, the quantized value of the spectral energy of the corresponding discrete frequency is divided by the total spectral energy to obtain the normalized spectral energy ratio of the corresponding discrete frequency. When the total spectral energy is not greater than the preset energy lower limit, the same normalized spectral energy ratio is assigned to all discrete frequencies within each analysis window.

[0038] Based on the ratio between discrete frequency and the number of sampling points in the window, the discrete frequency index of each discrete frequency is obtained. The discrete frequency index is multiplied by the corresponding normalized spectral energy ratio and accumulated over all discrete frequencies to obtain the spectral centroid of each pressure sensor in each analysis window.

[0039] Optionally, the step of calculating the change in spectral centroid between adjacent analysis windows, forming a sequence of changes according to the window order, and obtaining its arithmetic mean and average squared deviation specifically includes:

[0040] For each pressure sensor, starting from the second analysis window, calculate the difference between the spectral centroid value of the current analysis window and the spectral centroid value of the previous analysis window, and use the difference as the spectral centroid change of the current analysis window.

[0041] For each pressure sensor, the changes in the centroid of the spectrum corresponding to each analysis window are arranged in order of the analysis window number to form a sequence of changes in the centroid of the spectrum of the corresponding pressure sensor.

[0042] For each pressure sensor, sum all the changes in the centroid of the spectrum in the sequence of changes in the centroid of the spectrum and divide by the number of changes to obtain the arithmetic mean of the changes in the centroid of the spectrum of the corresponding pressure sensor.

[0043] For each pressure sensor, the difference is obtained by subtracting the arithmetic mean of the changes in the centroid of the spectrum from each change in the centroid of the spectrum in the sequence of changes in the centroid of the spectrum. The differences are then squared and summed, and divided by the number of changes to obtain the average squared deviation of the centroid of the spectrum of the corresponding pressure sensor. This average is used as a statistical feature to describe the degree of dispersion of the centroid of the spectrum of the corresponding pressure sensor.

[0044] Optionally, the step of constructing a one-sided fatigue damage index for each pressure sensor based on the arithmetic mean and the average of squared deviations, and selecting the index with the larger value as the overall fatigue damage index when there are multiple pressure sensors, specifically includes:

[0045] For each pressure sensor, the arithmetic mean of the change in the centroid of the spectrum is squared to obtain the squared result of the arithmetic mean of the change in the centroid of the spectrum.

[0046] For each pressure sensor, the average of the squared deviation of the change in the centroid of the spectrum and the square of the arithmetic mean of the change in the centroid of the spectrum, along with a preset minimum constant, are summed to obtain the combined value of the statistical quantity of the change in the centroid of the spectrum.

[0047] For each pressure sensor, the square of the arithmetic mean of the change in the centroid of the spectrum is used as the numerator, and the combined value of the statistical values ​​of the change in the centroid of the spectrum is used as the denominator. The ratio between the two is constructed to obtain the unilateral fatigue damage index of the corresponding pressure sensor.

[0048] When a first pressure sensor and a second pressure sensor are installed on the inlet and outlet pipe sections of the copper valve under test, respectively, the larger value of the single-sided fatigue damage index from the first and second pressure sensors is selected as the overall fatigue damage index.

[0049] Optionally, the step of calculating the root mean square value of the pulsating pressure from the time series of each sensor's pulsating pressure, taking the larger value as the equivalent pulsating pressure amplitude, and combining it with the overall fatigue damage index according to a preset relationship to obtain the fatigue strength assessment value, specifically includes:

[0050] For the pulsating pressure time series of each pressure sensor, the pulsating pressure values ​​of all sampling points are squared over the entire sampling time interval. All squared results are added together and divided by the total number of sampling points. The square root of the quotient is then taken to obtain the root mean square value of the pulsating pressure of the corresponding pressure sensor.

[0051] When a first pressure sensor and a second pressure sensor are installed on the inlet and outlet pipe sections of the copper valve being tested, respectively, the larger value of the root mean square value of the pulsating pressure from the first pressure sensor and the second pressure sensor is selected as the equivalent pulsating pressure amplitude.

[0052] The equivalent pulsating pressure amplitude is squared, and the squared result is used as the quantified value of the pressure amplitude. The overall fatigue damage index is added to the preset benchmark constant to form the fatigue damage normalization factor. The quantified value of the pressure amplitude is then divided by the fatigue damage normalization factor to obtain the fatigue strength assessment value of the copper valve disc.

[0053] The present invention has the following beneficial effects:

[0054] 1. Dual-channel pressure sensors are simultaneously deployed at both the inlet and outlet pipe sections, ensuring consistent sampling parameters that cover the entire opening and closing cycle, forming a complete raw pressure time series. Compared to traditional on-site sampling or pressure measurement at only one location, simultaneous dual-channel acquisition can fully reflect the pressure distribution differences before and after the valve disc, providing a more comprehensive data foundation for subsequent vibration and fatigue analysis. Furthermore, continuous high-frequency sampling ensures accurate reproduction of transient pressure fluctuations, avoiding the loss of critical signals caused by low-frequency sampling. Based on this, the solution employs a structured time series construction process, arranging the raw on-site readings in chronological order. This generates a data format directly usable for algorithm analysis without manual preprocessing, improving data processing efficiency and reducing the risk of errors caused by data format conversion.

[0055] 2. In the acquired raw sequence, the scheme uses the average pressure across the entire range as the static component. This component is then removed using a point-by-point difference method to obtain a pulsating pressure sequence that reflects only dynamic fluctuations. This processing not only removes long-term steady-state bias but also amplifies the minute pressure fluctuations during valve opening and closing, giving subsequent spectral analysis more prominent dynamic characteristics. Unlike traditional direct spectral analysis, which is easily overwhelmed by DC components, this method ensures that fatigue-related oscillation energy can be considered during the spectral mapping stage. It eliminates the need for complex filter design and avoids introducing phase distortion, maintaining the integrity and reversibility of the time-domain signal. Its practical significance lies in its ability to sensitively capture weak oscillations such as wear and sealing changes during long-term valve operation, allowing for early detection of fatigue accumulation trends.

[0056] 3. The pulsation sequence is segmented into overlapping segments according to a preset window length and sliding step size, effectively balancing time and frequency resolution. Traditional single non-overlapping segmentation leads to boundary effects and information loss, while overlapping segmentation improves the smoothness of feature extraction while ensuring the continuity of local signals. The scheme automatically sets the window length to the larger of the minimum value and one-third of the total number of points, avoiding the difficulty of manual parameter tuning and enhancing the algorithm's adaptability. In addition, the half-overlapping sliding step size strategy ensures that each time step is covered by at least two windows, reducing the interference of occasional noise on single features and making fatigue characteristic extraction more robust. It provides a flexible and efficient time-frequency decomposition method, laying a reliable foundation for subsequent spectrum mapping and centroid calculation.

[0057] 4. The proposed method employs a pre-defined sequence of discrete frequency mapping functions to map pulsating samples within a window to several discrete frequency energy values, which are then summed to obtain the corresponding spectral energy quantization distribution. Unlike traditional Fourier transform-based spectral analysis, this method constructs the mapping function using a weighted superposition approach. It does not rely on a fast algorithm for converting the entire time domain to the frequency domain, but rather customizes the frequency mapping sequence based on the specific fatigue physics mechanism, flexibly highlighting fatigue-induced characteristic frequency bands. This mapping function is defined by the superposition of cosine and sine terms, possessing adjustable frequency band resolution and energy focusing capabilities, suppressing scene noise interference and focusing on fatigue-related frequencies. Compared to standard spectral analysis, this method is more targeted and scalable, allowing the mapping function to be adjusted according to valve size and fluid characteristics, thus improving diagnostic accuracy.

[0058] 5. After obtaining the spectral energy distribution, this scheme calculates the weighted sum of each discrete frequency and its normalized energy ratio to obtain the spectral centroid, which characterizes the energy center position of each window. Compared to traditional methods that only focus on peak frequencies, the centroid reflects the energy tilt trend of the entire spectrum, providing a more intuitive understanding of energy accumulation and migration during fatigue. This statistic is unaffected by occasional peak fluctuations, exhibiting higher stability. In the extreme scenario where the total spectral energy is zero, the scheme sets a uniform default value to avoid computational failure. This effectively solves the problem of fuzzy feature extraction under multi-peak spectra, providing a global and continuous spectral feature index for fatigue accumulation assessment.

[0059] 6. This scheme constructs statistical features describing the amplitude and dispersion of spectral energy evolution by calculating the change in the centroid of the spectrum between adjacent windows and obtaining the arithmetic mean and deviation mean of their sequences. Traditional fatigue diagnosis often directly evaluates peak or root mean square values, neglecting the predictive role of frequency energy migration speed and its stability in fatigue damage. This scheme uses dual statistics to characterize the fatigue evolution process from a dynamic perspective: the average value reflects the overall migration trend, and the deviation mean reflects the range of oscillation intensity fluctuations. It can identify the differences between the fatigue instability stage and the stable stage, providing more physically meaningful parameters for subsequent damage index calculations.

[0060] 7. Based on the statistical characteristics of the center of gravity changes of each pressure sensor, this scheme constructs a unilateral fatigue damage index and selects the maximum value as the overall damage index in a multi-sensor scenario. Traditional methods that rely solely on single-point measurements or direct averaging may overlook the most severely damaged local locations, while the maximum value strategy can promptly capture the maximum cumulative fatigue risk at any critical location. This fusion method is simple and easy to implement, requiring no complex fusion weight learning, ensuring that critical damage signals are not diluted by averaging. Its practical significance lies in meeting the health monitoring needs of multi-distributed measurement points in actual valves, accurately locating and warning of the most dangerous areas, and improving overall monitoring reliability.

[0061] 8. Finally, the equivalent pulsating pressure amplitude and the overall fatigue damage index are combined according to a preset relationship to obtain the fatigue strength assessment value of the copper valve disc. This comprehensive assessment method differs from traditional offline laboratory fatigue curve comparison, directly converting online dynamic characteristics into residual fatigue strength. Combining real-time data with statistical damage indicators, no manual interpretation is required, providing an automated decision-making basis for operation and maintenance. Achieving online real-time fatigue strength estimation provides scientific quantitative indicators for preventive maintenance and life management of critical valve equipment, and has significant application and promotion value. Attached Figure Description

[0062] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0063] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] Example, refer to Figure 1 A method for testing the fatigue strength of a copper valve disc based on a smart sensor, comprising:

[0065] Pressure sensors were installed at the inlet and outlet pipe sections of the copper valve under test. Uniform sampling parameters were set, and pressure measurements were continuously collected throughout the opening and closing process to form the original pressure time series.

[0066] The average pressure was obtained across the entire sampling interval. At each sampling point, the average pressure was subtracted from the original pressure to obtain the pulsating pressure time series.

[0067] Based on the pulsating pressure time series, the number of sampling points and the sliding step size are set in the window to construct a sequence of analysis windows that partially overlap, and a continuous sample sequence is formed within each window;

[0068] Based on the preset discrete frequency mapping function sequence, the sample sequences of each window are linearly transformed and accumulated to obtain the linear projection value and spectral energy quantization value distribution of each discrete frequency;

[0069] The total spectral energy is obtained by accumulating the quantized values ​​of spectral energy in each window, normalizing it, and weighting it according to the discrete frequency index to calculate the spectral centroid.

[0070] Calculate the change in the centroid of the spectrum between adjacent analysis windows, form a sequence of changes according to the window order, and obtain its arithmetic mean and the average of squared deviations;

[0071] The single-sided fatigue damage index of each pressure sensor is constructed based on the arithmetic mean and the average of squared deviations. When there are multiple pressure sensors, the index with the larger value is selected as the overall fatigue damage index.

[0072] The root mean square value of the pulsating pressure is calculated from the time series of the pulsating pressure of each sensor. The larger value is taken as the equivalent pulsating pressure amplitude. This value is then combined with the overall fatigue damage index according to a preset relationship to obtain the fatigue strength assessment value.

[0073] By deploying high-precision pressure sensors at the valve inlet and outlet and continuously collecting pressure data throughout the opening and closing process with unified parameters, this method eliminates the shortcomings of traditional sampling or single-point monitoring in reflecting full-condition fluctuations. Next, the original pressure is differentiated from the average value across the entire range to intuitively filter out steady-state components, ensuring that subsequent analysis focuses on true pulsating changes and solving the problem of static bias affecting feature extraction accuracy in spectrum analysis. Furthermore, an adaptive window and overlapping sliding are used to construct a time-series sequence, balancing time and frequency resolution and overcoming the drawbacks of easily missing transient features when using only fixed segment lengths or non-overlapping segments. Subsequently, a preset mapping function is applied to map the time-domain samples to key frequency bands, replacing standard Fourier full-spectrum analysis, allowing for the prominent extraction of fatigue key frequency energy and avoiding the accuracy degradation caused by broadband noise interference in traditional spectrum analysis. Finally, by combining the spectral centroid and its change statistics with unilateral damage indicators and pulsating pressure amplitude fusion assessment, the valve fatigue strength value is given in real time, providing a quantifiable and traceable strength assessment indicator for online health monitoring during operation, significantly improving the timeliness and accuracy of early warning.

[0074] The process involves installing pressure sensors on the inlet and outlet pipe sections of the copper valve under test, setting uniform sampling parameters, and continuously collecting pressure measurements throughout the opening and closing process to form an original pressure time series. Specifically, this includes:

[0075] A first pressure sensor is installed in the inlet pipe section of the copper valve under test, and a second pressure sensor is installed in the outlet pipe section of the copper valve under test. They are respectively numbered as the first pressure sensor and the second pressure sensor. The first pressure sensor collects the pressure measurement value at the inlet of the copper valve under test, and the second pressure sensor collects the pressure measurement value at the outlet of the copper valve under test.

[0076] Set a uniform sampling frequency, and set the total number of sampling points for each pressure sensor to be no less than four. Calculate the total continuous sampling time based on the total number of sampling points and the uniform sampling frequency, so that the continuous sampling process covers the entire working condition change process of the copper valve under test from fully closed to fully open and then from fully open to fully closed.

[0077] Based on the uniform sampling frequency and the total continuous sampling duration, the entire sampling period is evenly divided into several discrete sampling moments, and the discrete sampling moments are numbered according to the time sequence to obtain the sampling point sequence from the first sampling point to the last sampling point.

[0078] At each discrete sampling moment, the pressure measurement values ​​of the first pressure sensor and the second pressure sensor are recorded respectively. The pressure measurement values ​​of the two pressure sensors are arranged in the order of the sampling point sequence to form the original pressure time series of the first pressure sensor and the second pressure sensor. The number of sampling points in each original pressure time series is equal to the total number of sampling points.

[0079] During continuous sampling, the copper valve under test is kept fully closed during the sampling period corresponding to the first quarter of the sampling point sequence. Before the sampling time at the first quarter position in the sampling point sequence, the copper valve under test is switched from fully closed to fully open. During the sampling period corresponding to the first quarter to the half position in the sampling point sequence, the copper valve under test is kept fully open. Before the sampling time at the half position in the sampling point sequence, the copper valve under test is switched from fully open to fully closed. During the sampling period corresponding to the last half of the sampling point sequence, the copper valve under test is kept fully closed.

[0080] A first pressure sensor, numbered as follows, is installed on the inlet pipe section of the copper valve being tested. A second pressure sensor, numbered [number], is installed on the outlet pipe section of the copper valve being tested. ;in, Number the pressure sensor;

[0081] Set the pressure signal sampling frequency to [value]. Set the total number of sampling points for each sensor to be [number]. ,and Calculate the total duration of continuous sampling as ;

[0082] The discrete sampling time is constructed as follows: , ;in, The discrete sampling point number; For the first The sampling time corresponding to each sampling point;

[0083] At each sampling time The output values ​​of the two pressure sensors were recorded to form the original pressure time series: , , ;in, For the first Each pressure sensor at the sampling time The original pressure measurement value;

[0084] Sampling point number The valve remains fully closed; at the sampling time Previously, the valve was switched from fully closed to fully open, which resulted in the sampling point sequence number. The valve remains fully open; at the sampling time Previously, the valve was switched from fully open to fully closed, which resulted in the sampling point sequence number. Keep the valve fully closed.

[0085] The process of obtaining the average pressure across the entire sampling interval, and subtracting the average pressure from the original pressure at each sampling point to obtain the pulsating pressure time series, specifically includes:

[0086] For each pressure sensor, iterate through all the pressure values ​​of the original pressure time series of each pressure sensor throughout the entire sampling time interval, add up all the pressure values ​​of the sampling points and divide by the total number of sampling points to obtain the average pressure of each pressure sensor during the entire sampling period.

[0087] For each pressure sensor, at each sampling point, the average pressure of the corresponding pressure sensor is subtracted from the sampling point pressure value in the corresponding original pressure time series. The difference is used as the pulsating pressure value of the corresponding sampling point. The pulsating pressure time series of the corresponding pressure sensor is constructed by arranging the sampling points in the order of sampling points, so that each pulsating pressure time series is consistent with the corresponding original pressure time series in terms of the number and order of sampling points.

[0088] For each sensor The average pressure within the sampling time interval is calculated as follows: ;in, For the first The average pressure of each sensor during the entire sampling period;

[0089] The time series of pulsating pressure is calculated for each sampling point as follows: , , ;in, For the first Each sensor at the sampling time The pulsating pressure value.

[0090] The method of setting the number of sampling points and the sliding step size based on the pulsating pressure time series to construct a sequence of partially overlapping analysis windows, and forming a continuous sample sequence within each window, specifically includes:

[0091] The number of sampling points in each analysis window is set according to the total number of sampling points in the pulsating pressure time series, so that the number of sampling points in the analysis window is not less than the preset minimum number of sampling points in the window and not less than one-third of the total number of sampling points, and the set number of sampling points is used as the number of sampling points in the window.

[0092] Based on the number of sampling points in the window and the uniform sampling frequency, the time length corresponding to a single analysis window is calculated to obtain the time length parameter of the analysis window.

[0093] Set the sliding step size between the starting sampling points of adjacent analysis windows to half the number of sampling points in the window, and ensure that the sliding step size is not less than one sampling point interval and not greater than the number of sampling points in the window.

[0094] Under the condition that the total number of sampling points is not less than the sum of the number of window sampling points and twice the sliding step size, and the total number of analysis windows constructed by the sliding step size is not less than three, starting from the first sampling point of the pulsating pressure time series, the starting sampling point of the analysis window is moved sequentially according to the sliding step size, and each time a continuous pulsating pressure sample containing the number of window sampling points is extracted, and the number of analysis windows obtained is used as the total number of windows that can be constructed.

[0095] For each pressure sensor and each analysis window, continuous pulsating pressure samples are extracted in the sampling order within each analysis window. A sampling position number is assigned to each sampling position within each analysis window. The pulsating pressure samples corresponding to each sampling position number are arranged in the order of the sampling position number to form the pulsating pressure sample sequence of the corresponding pressure sensor in each analysis window.

[0096] Set the number of sampling points included in each analysis window to ;

[0097] Calculate the time length corresponding to a single window. ;

[0098] Set the slide step size between the starting positions of adjacent windows. ,satisfy: ;

[0099] Set the total number of windows that can be constructed to , , ;

[0100] For each sensor and the number of each window The pulsating pressure sequence within the constructed window is as follows: , ;in, Number the windows; The position number inside the window; For the first The sensor at the first The first window The pulsating pressure sample values ​​corresponding to each location.

[0101] The step of performing a linear transformation and accumulating the sample sequences of each window based on a preset discrete frequency mapping function sequence to obtain the linear projection value and spectral energy quantization value distribution of each discrete frequency specifically includes:

[0102] The number of discrete frequencies is set according to the number of window sampling points, and a discrete frequency index is assigned to each discrete frequency. A preset discrete frequency mapping function sequence is used to define the function value of each mapping function at each sampling position number by weighted superposition of cosine and sine terms, so that the length of each mapping function is equal to the number of window sampling points.

[0103] For each pressure sensor, each analysis window, and each discrete frequency index, the pulsating pressure sample value corresponding to each sampling position number in each analysis window is multiplied one by one with the function value of the corresponding discrete frequency mapping function at the same sampling position number, and then accumulated sequentially along the sampling position number to obtain the linear projection value of the corresponding discrete frequency index of the corresponding pressure sensor under each analysis window.

[0104] For each pressure sensor, each analysis window, and each discrete frequency index, the corresponding linear projection value is multiplied by the corresponding linear projection value, and the resulting product is used as the spectral energy quantization value of the corresponding pressure sensor at the corresponding discrete frequency index under each analysis window.

[0105] For each pressure sensor, the quantized values ​​of each spectrum energy are filled into the corresponding positions one by one according to the analysis window number as the row index and the discrete frequency index as the column index, thus forming the spectrum energy matrix of the corresponding pressure sensor.

[0106] The discrete mapping function is constructed as follows: , , ;in, For the first A discrete mapping function; The mapping sequence number;

[0107] For each sensor, each window number, and each mapping sequence number, calculate the projection value as follows: ;in, For the first The sensor at the first Within the window, for the first The cumulative result of the projection of each mapping function;

[0108] The calculation window and frequency energy are as follows: ;in, For the first The sensor at the first Each window, mapping number is The energy quantization value at that time;

[0109] For each sensor The energy matrix is ​​constructed as follows: ;in, For the first The energy matrix corresponding to each sensor.

[0110] The total spectral energy is obtained by accumulating the quantized spectral energy values ​​within each window, normalizing it, and weighting it according to the discrete frequency index to calculate the spectral centroid. Specifically, this includes:

[0111] For each pressure sensor and each analysis window, the quantized values ​​of the spectral energy corresponding to all discrete frequencies are accumulated within each analysis window to obtain the total spectral energy of the corresponding pressure sensor in each analysis window.

[0112] For each pressure sensor, each analysis window, and each discrete frequency, when the total spectral energy is greater than the preset energy lower limit, the quantized value of the spectral energy of the corresponding discrete frequency is divided by the total spectral energy to obtain the normalized spectral energy ratio of the corresponding discrete frequency. When the total spectral energy is not greater than the preset energy lower limit, the same normalized spectral energy ratio is assigned to all discrete frequencies within each analysis window.

[0113] Based on the ratio between discrete frequency and the number of sampling points in the window, the discrete frequency index of each discrete frequency is obtained. The discrete frequency index is multiplied by the corresponding normalized spectral energy ratio and accumulated over all discrete frequencies to obtain the spectral centroid of each pressure sensor in each analysis window.

[0114] For each sensor and each window number, calculate the total spectral energy of that window as follows: ;in, For the first The sensor at the first Within each window, all mapping numbers The total energy;

[0115] For each window number and each mapping sequence number, the normalized spectral energy is constructed as follows: ;in, This is the normalized energy ratio value;

[0116] The discrete frequency index is constructed as follows: , ;in, Mapping sequence number The corresponding discrete frequency index;

[0117] For each sensor and each window, the frequency centroid is calculated as follows: ;in, For the first The sensor at the first The spectral centroid of each window.

[0118] The step of calculating the change in spectral centroid between adjacent analysis windows, forming a sequence of changes according to the window order, and obtaining its arithmetic mean and average squared deviation specifically includes:

[0119] For each pressure sensor, starting from the second analysis window, calculate the difference between the spectral centroid value of the current analysis window and the spectral centroid value of the previous analysis window, and use the difference as the spectral centroid change of the current analysis window.

[0120] For each pressure sensor, the changes in the centroid of the spectrum corresponding to each analysis window are arranged in order of the analysis window number to form a sequence of changes in the centroid of the spectrum of the corresponding pressure sensor.

[0121] For each pressure sensor, sum all the changes in the centroid of the spectrum in the sequence of changes in the centroid of the spectrum and divide by the number of changes to obtain the arithmetic mean of the changes in the centroid of the spectrum of the corresponding pressure sensor.

[0122] For each pressure sensor, the difference is obtained by subtracting the arithmetic mean of the changes in the centroid of the spectrum from each change in the centroid of the spectrum in the sequence of changes in the centroid of the spectrum. The differences are then squared and summed, and divided by the number of changes to obtain the average squared deviation of the centroid of the spectrum of the corresponding pressure sensor. This average is used as a statistical feature to describe the degree of dispersion of the centroid of the spectrum of the corresponding pressure sensor.

[0123] For each sensor and window number The change in the spectral centroid of adjacent windows is calculated as follows: ;in, For the first The sensor at the first The change in the center of gravity of each window relative to the previous window;

[0124] For each sensor The sequence of changes in the spectral centroid is constructed as follows: ;in, For the first A sequence of changes in the center of gravity of each sensor;

[0125] The average value of the change in the centroid of the spectrum is calculated as follows: ;in, For the first The average value of the sequence of changes in the center of gravity of each sensor;

[0126] The mean square dispersion of the change in the centroid of the spectrum is calculated as follows: ;in, For the first The mean square dispersion of the change in the center of gravity of a sensor relative to its average value.

[0127] The method involves constructing a one-sided fatigue damage index for each pressure sensor based on the arithmetic mean and the average of squared deviations. When multiple pressure sensors exist, the index with the larger value is selected as the overall fatigue damage index. Specifically, this includes:

[0128] For each pressure sensor, the arithmetic mean of the change in the centroid of the spectrum is squared to obtain the squared result of the arithmetic mean of the change in the centroid of the spectrum.

[0129] For each pressure sensor, the average of the squared deviation of the change in the centroid of the spectrum and the square of the arithmetic mean of the change in the centroid of the spectrum, along with a preset minimum constant, are summed to obtain the combined value of the statistical quantity of the change in the centroid of the spectrum.

[0130] For each pressure sensor, the square of the arithmetic mean of the change in the centroid of the spectrum is used as the numerator, and the combined value of the statistical values ​​of the change in the centroid of the spectrum is used as the denominator. The ratio between the two is constructed to obtain the unilateral fatigue damage index of the corresponding pressure sensor.

[0131] When a first pressure sensor and a second pressure sensor are installed on the inlet and outlet pipe sections of the copper valve under test, respectively, the larger value of the single-sided fatigue damage index from the first and second pressure sensors is selected as the overall fatigue damage index.

[0132] For each sensor, the fatigue damage index is constructed as follows: ;in, For the first Fatigue damage indicators corresponding to each sensor. To prevent extremely small constants with a denominator of zero;

[0133] Based on the damage indices of the pressure sensors on both sides, the overall fatigue damage index is calculated as follows: ;in, It serves as an indicator of overall fatigue damage.

[0134] The process involves calculating the root mean square value of the pulsating pressure from the time series of each sensor's pulsating pressure, taking the larger value as the equivalent pulsating pressure amplitude, and combining it with the overall fatigue damage index according to a preset relationship to obtain the fatigue strength assessment value. Specifically, this includes:

[0135] For the pulsating pressure time series of each pressure sensor, the pulsating pressure values ​​of all sampling points are squared over the entire sampling time interval. All squared results are added together and divided by the total number of sampling points. The square root of the quotient is then taken to obtain the root mean square value of the pulsating pressure of the corresponding pressure sensor.

[0136] When a first pressure sensor and a second pressure sensor are installed on the inlet and outlet pipe sections of the copper valve being tested, respectively, the larger value of the root mean square value of the pulsating pressure from the first pressure sensor and the second pressure sensor is selected as the equivalent pulsating pressure amplitude.

[0137] The equivalent pulsating pressure amplitude is squared, and the squared result is used as the quantified value of the pressure amplitude. The overall fatigue damage index is added to the preset benchmark constant to form the fatigue damage normalization factor. The quantified value of the pressure amplitude is then divided by the fatigue damage normalization factor to obtain the fatigue strength assessment value of the copper valve disc.

[0138] The root mean square value of the pulsating pressure is calculated for each sensor as follows: ;in, For the first The root mean square value of the pulsating pressure of each sensor;

[0139] The equivalent pulsating pressure amplitude is calculated based on the root mean square value of the pulsating pressure on both sides: ;in, This is the equivalent pulsating pressure amplitude;

[0140] Based on the overall fatigue damage index and the equivalent pulsating pressure amplitude, the fatigue strength assessment value of the copper valve disc is calculated as follows: ;in, This is the fatigue strength assessment value for the valve disc of a copper valve.

[0141] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0142] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for testing the fatigue strength of a copper valve disc based on a smart sensor, characterized in that, include: Pressure sensors were installed at the inlet and outlet pipe sections of the copper valve under test. Uniform sampling parameters were set, and pressure measurements were continuously collected throughout the opening and closing process to form the original pressure time series. The average pressure was obtained across the entire sampling interval. At each sampling point, the average pressure was subtracted from the original pressure to obtain the pulsating pressure time series. Based on the pulsating pressure time series, the number of sampling points and the sliding step size are set in the window to construct a sequence of analysis windows that partially overlap, and a continuous sample sequence is formed within each window; Based on the preset discrete frequency mapping function sequence, the sample sequences of each window are linearly transformed and accumulated to obtain the linear projection value and spectral energy quantization value distribution of each discrete frequency; The total spectral energy is obtained by accumulating the quantized values ​​of spectral energy in each window, normalizing it, and weighting it according to the discrete frequency index to calculate the spectral centroid. Calculate the change in the centroid of the spectrum between adjacent analysis windows, form a sequence of changes according to the window order, and obtain its arithmetic mean and the average of squared deviations; The single-sided fatigue damage index of each pressure sensor is constructed based on the arithmetic mean and the average of squared deviations. When there are multiple pressure sensors, the index with the larger value is selected as the overall fatigue damage index. The root mean square value of the pulsating pressure is calculated from the time series of the pulsating pressure of each sensor. The larger value is taken as the equivalent pulsating pressure amplitude. This value is then combined with the overall fatigue damage index according to a preset relationship to obtain the fatigue strength assessment value.

2. The method for testing the fatigue strength of a copper valve disc based on a smart sensor according to claim 1, characterized in that, The process involves installing pressure sensors on the inlet and outlet pipe sections of the copper valve under test, setting uniform sampling parameters, and continuously collecting pressure measurements throughout the opening and closing process to form an original pressure time series. Specifically, this includes: A first pressure sensor is installed in the inlet pipe section of the copper valve under test, and a second pressure sensor is installed in the outlet pipe section of the copper valve under test. They are respectively numbered as the first pressure sensor and the second pressure sensor. The first pressure sensor collects the pressure measurement value at the inlet of the copper valve under test, and the second pressure sensor collects the pressure measurement value at the outlet of the copper valve under test. Set a uniform sampling frequency, and set the total number of sampling points for each pressure sensor to be no less than four. Calculate the total continuous sampling time based on the total number of sampling points and the uniform sampling frequency, so that the continuous sampling process covers the entire working condition change process of the copper valve under test from fully closed to fully open and then from fully open to fully closed. Based on the uniform sampling frequency and the total continuous sampling duration, the entire sampling period is evenly divided into several discrete sampling moments, and the discrete sampling moments are numbered according to the time sequence to obtain the sampling point sequence from the first sampling point to the last sampling point. At each discrete sampling moment, the pressure measurement values ​​of the first pressure sensor and the second pressure sensor are recorded respectively. The pressure measurement values ​​of the two pressure sensors are arranged in the order of the sampling point sequence to form the original pressure time series of the first pressure sensor and the second pressure sensor. The number of sampling points in each original pressure time series is equal to the total number of sampling points. During continuous sampling, the copper valve under test is kept fully closed during the sampling period corresponding to the first quarter of the sampling point sequence. Before the sampling time at the first quarter position in the sampling point sequence, the copper valve under test is switched from fully closed to fully open. During the sampling period corresponding to the first quarter to the half position in the sampling point sequence, the copper valve under test is kept fully open. Before the sampling time at the half position in the sampling point sequence, the copper valve under test is switched from fully open to fully closed. During the sampling period corresponding to the last half of the sampling point sequence, the copper valve under test is kept fully closed.

3. The method for testing the fatigue strength of a copper valve disc based on a smart sensor according to claim 2, characterized in that, The process of obtaining the average pressure across the entire sampling interval, and subtracting the average pressure from the original pressure at each sampling point to obtain the pulsating pressure time series, specifically includes: For each pressure sensor, iterate through all the pressure values ​​of the original pressure time series of each pressure sensor throughout the entire sampling time interval, add up all the pressure values ​​of the sampling points and divide by the total number of sampling points to obtain the average pressure of each pressure sensor during the entire sampling period. For each pressure sensor, at each sampling point, the average pressure of the corresponding pressure sensor is subtracted from the sampling point pressure value in the corresponding original pressure time series. The difference is used as the pulsating pressure value of the corresponding sampling point. The pulsating pressure time series of the corresponding pressure sensor is constructed by arranging the sampling points in the order of sampling points, so that each pulsating pressure time series is consistent with the corresponding original pressure time series in terms of the number and order of sampling points.

4. The method for testing the fatigue strength of a copper valve disc based on a smart sensor according to claim 3, characterized in that, The method of setting the number of sampling points and the sliding step size based on the pulsating pressure time series to construct a sequence of partially overlapping analysis windows, and forming a continuous sample sequence within each window, specifically includes: The number of sampling points in each analysis window is set according to the total number of sampling points in the pulsating pressure time series, so that the number of sampling points in the analysis window is not less than the preset minimum number of sampling points in the window and not less than one-third of the total number of sampling points, and the set number of sampling points is used as the number of sampling points in the window. Based on the number of sampling points in the window and the uniform sampling frequency, the time length corresponding to a single analysis window is calculated to obtain the time length parameter of the analysis window. Set the sliding step size between the starting sampling points of adjacent analysis windows to half the number of sampling points in the window, and ensure that the sliding step size is not less than one sampling point interval and not greater than the number of sampling points in the window. Under the condition that the total number of sampling points is not less than the sum of the number of window sampling points and twice the sliding step size, and the total number of analysis windows constructed by the sliding step size is not less than three, starting from the first sampling point of the pulsating pressure time series, the starting sampling point of the analysis window is moved sequentially according to the sliding step size, and each time a continuous pulsating pressure sample containing the number of window sampling points is extracted, and the number of analysis windows obtained is used as the total number of windows that can be constructed. For each pressure sensor and each analysis window, continuous pulsating pressure samples are extracted in the sampling order within each analysis window. A sampling position number is assigned to each sampling position within each analysis window. The pulsating pressure samples corresponding to each sampling position number are arranged in the order of the sampling position number to form the pulsating pressure sample sequence of the corresponding pressure sensor in each analysis window.

5. The method for testing the fatigue strength of a copper valve disc based on a smart sensor according to claim 4, characterized in that, The step of performing a linear transformation and accumulating the sample sequences of each window based on a preset discrete frequency mapping function sequence to obtain the linear projection value and spectral energy quantization value distribution of each discrete frequency specifically includes: The number of discrete frequencies is set according to the number of window sampling points, and a discrete frequency index is assigned to each discrete frequency. A preset discrete frequency mapping function sequence is used to define the function value of each mapping function at each sampling position number by weighted superposition of cosine and sine terms, so that the length of each mapping function is equal to the number of window sampling points. For each pressure sensor, each analysis window, and each discrete frequency index, the pulsating pressure sample value corresponding to each sampling position number in each analysis window is multiplied one by one with the function value of the corresponding discrete frequency mapping function at the same sampling position number, and then accumulated sequentially along the sampling position number to obtain the linear projection value of the corresponding discrete frequency index of the corresponding pressure sensor under each analysis window. For each pressure sensor, each analysis window, and each discrete frequency index, the corresponding linear projection value is multiplied by the corresponding linear projection value, and the resulting product is used as the spectral energy quantization value of the corresponding pressure sensor at the corresponding discrete frequency index under each analysis window. For each pressure sensor, the quantized values ​​of each spectrum energy are filled into the corresponding positions one by one according to the analysis window number as the row index and the discrete frequency index as the column index, thus forming the spectrum energy matrix of the corresponding pressure sensor.

6. The method for testing the fatigue strength of a copper valve disc based on a smart sensor according to claim 5, characterized in that, The total spectral energy is obtained by accumulating the quantized spectral energy values ​​within each window, normalizing it, and weighting it according to the discrete frequency index to calculate the spectral centroid. Specifically, this includes: For each pressure sensor and each analysis window, the quantized values ​​of the spectral energy corresponding to all discrete frequencies are accumulated within each analysis window to obtain the total spectral energy of the corresponding pressure sensor in each analysis window. For each pressure sensor, each analysis window, and each discrete frequency, when the total spectral energy is greater than the preset energy lower limit, the quantized value of the spectral energy of the corresponding discrete frequency is divided by the total spectral energy to obtain the normalized spectral energy ratio of the corresponding discrete frequency. When the total spectral energy is not greater than the preset energy lower limit, the same normalized spectral energy ratio is assigned to all discrete frequencies within each analysis window. Based on the ratio between discrete frequency and the number of sampling points in the window, the discrete frequency index of each discrete frequency is obtained. The discrete frequency index is multiplied by the corresponding normalized spectral energy ratio and accumulated over all discrete frequencies to obtain the spectral centroid of each pressure sensor in each analysis window.

7. The method for testing the fatigue strength of a copper valve disc based on a smart sensor according to claim 6, characterized in that, The step of calculating the change in spectral centroid between adjacent analysis windows, forming a sequence of changes according to the window order, and obtaining its arithmetic mean and average squared deviation specifically includes: For each pressure sensor, starting from the second analysis window, calculate the difference between the spectral centroid value of the current analysis window and the spectral centroid value of the previous analysis window, and use the difference as the spectral centroid change of the current analysis window. For each pressure sensor, the changes in the centroid of the spectrum corresponding to each analysis window are arranged in order of the analysis window number to form a sequence of changes in the centroid of the spectrum of the corresponding pressure sensor. For each pressure sensor, sum all the changes in the centroid of the spectrum in the sequence of changes in the centroid of the spectrum and divide by the number of changes to obtain the arithmetic mean of the changes in the centroid of the spectrum of the corresponding pressure sensor. For each pressure sensor, the difference is obtained by subtracting the arithmetic mean of the changes in the centroid of the spectrum from each change in the centroid of the spectrum in the sequence of changes in the centroid of the spectrum. The differences are then squared and summed, and divided by the number of changes to obtain the average squared deviation of the centroid of the spectrum of the corresponding pressure sensor. This average is used as a statistical feature to describe the degree of dispersion of the centroid of the spectrum of the corresponding pressure sensor.

8. The method for testing the fatigue strength of a copper valve disc based on a smart sensor according to claim 7, characterized in that, The method involves constructing a one-sided fatigue damage index for each pressure sensor based on the arithmetic mean and the average of squared deviations. When multiple pressure sensors exist, the index with the larger value is selected as the overall fatigue damage index. Specifically, this includes: For each pressure sensor, the arithmetic mean of the change in the centroid of the spectrum is squared to obtain the squared result of the arithmetic mean of the change in the centroid of the spectrum. For each pressure sensor, the average of the squared deviation of the change in the centroid of the spectrum and the square of the arithmetic mean of the change in the centroid of the spectrum, along with a preset minimum constant, are summed to obtain the combined value of the statistical quantity of the change in the centroid of the spectrum. For each pressure sensor, the square of the arithmetic mean of the change in the centroid of the spectrum is used as the numerator, and the combined value of the statistical values ​​of the change in the centroid of the spectrum is used as the denominator. The ratio between the two is constructed to obtain the unilateral fatigue damage index of the corresponding pressure sensor. When a first pressure sensor and a second pressure sensor are installed on the inlet and outlet pipe sections of the copper valve under test, respectively, the larger value of the single-sided fatigue damage index from the first and second pressure sensors is selected as the overall fatigue damage index.

9. A method for testing the fatigue strength of a copper valve disc based on a smart sensor according to claim 8, characterized in that, The process involves calculating the root mean square value of the pulsating pressure from the time series of each sensor's pulsating pressure, taking the larger value as the equivalent pulsating pressure amplitude, and combining it with the overall fatigue damage index according to a preset relationship to obtain the fatigue strength assessment value. Specifically, this includes: For the pulsating pressure time series of each pressure sensor, the pulsating pressure values ​​of all sampling points are squared over the entire sampling time interval. All squared results are added together and divided by the total number of sampling points. The square root of the quotient is then taken to obtain the root mean square value of the pulsating pressure of the corresponding pressure sensor. When a first pressure sensor and a second pressure sensor are installed on the inlet and outlet pipe sections of the copper valve being tested, respectively, the larger value of the root mean square value of the pulsating pressure from the first pressure sensor and the second pressure sensor is selected as the equivalent pulsating pressure amplitude. The equivalent pulsating pressure amplitude is squared, and the squared result is used as the quantified value of the pressure amplitude. The overall fatigue damage index is added to the preset benchmark constant to form the fatigue damage normalization factor. The quantified value of the pressure amplitude is then divided by the fatigue damage normalization factor to obtain the fatigue strength assessment value of the copper valve disc.

Images