Performance detection method and system for industrial valve

Abstract

The invention discloses a performance detection method and system for an industrial valve, and relates to the technical field of valve detection.The method comprises the steps that an operation action event of the valve is monitored and recorded, and an action time sequence is obtained; for each operation action event, multi-modal response signals of the valve are synchronously collected, and a multi-modal response sequence is obtained; the multi-modal response signal at least comprises a vibration signal, an acoustic emission signal and a pressure fluctuation signal; performing time sequence causal inference on the response signal mode corresponding to each operation action event to obtain causal effect intensity of the response signal, and forming a causal effect intensity matrix by a plurality of causal effect intensity values; according to the invention, valve vibration, acoustic emission and pressure fluctuation multi-mode response signals can be synchronously collected, and the causal effect intensity between the mining signals can be deduced by combining time sequence causality, so that weak signal correlation characteristics caused by sealing surface microcosmic damage under a dynamic working condition can be accurately captured.

Description

Technical Field

[0001] This invention relates to the field of valve testing technology, and in particular to a method and system for testing the performance of industrial valves. Background Technology

[0002] Most existing valve performance testing methods employ periodic offline testing or internal leakage rate testing under steady-state conditions. While these methods can detect the sealing performance of valves under stable conditions, they cannot effectively identify intermittent, slight internal leakage at the valve sealing surface under varying operating conditions, media containing particulate matter, or temperature and pressure fluctuations.

[0003] In actual industrial processes, valves are often in a dynamic working state. The medium may contain solid particles, and the sealing surface is prone to micro-cracks or local damage due to erosion, wear, thermal deformation, etc. These damages will manifest as intermittent internal leakage during valve opening or closing, or under certain pressure and temperature fluctuations. They are difficult to reproduce and detect in steady-state testing. Existing technologies lack specific analysis methods for the correlation between micro-leakage of the sealing surface and vibration characteristics, resulting in a high rate of missed early fault detection and inaccurate predictive maintenance. Summary of the Invention

[0004] This invention provides a method and system for performance testing of industrial valves to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a performance testing method for industrial valves, comprising:

[0006] S1. Monitor and record the valve's operation events to obtain the action timing sequence;

[0007] S2. For each operational event, synchronously acquire the valve's multimodal response signal to obtain a multimodal response sequence; the multimodal response signal includes at least vibration signal, acoustic emission signal, and pressure fluctuation signal;

[0008] S3. Perform time series causal inference on the response signal mode corresponding to each operation action event to obtain the causal effect intensity of the response signal, and form a causal effect intensity matrix composed of multiple causal effect intensity values;

[0009] S4. Compare the causal effect intensity matrix with the baseline causal pattern of a normal valve, analyze the causal deviation, and obtain the causal anomaly index.

[0010] S5. Based on causal anomaly indicators, determine whether the valve has internal leakage, and generate an internal leakage detection report based on the judgment results.

[0011] Preferably, the monitoring and recording of valve operation events to obtain an action timing sequence includes:

[0012] The valve stem rotation is monitored using a high-precision encoder to obtain a valve stem rotation angle-time series.

[0013] Based on the valve stem angle-time sequence, continuous angle changes are divided into action segments to obtain independent operation action events in units of complete opening and closing cycles;

[0014] For each independent operation event, the current signal of the drive device is synchronously acquired, and the time-domain waveform of the current signal is time-aligned and marked with the valve stem rotation angle-time sequence to obtain the action time sequence carrying the drive state characteristics.

[0015] Preferably, the step of synchronously acquiring the multimodal response signal of the valve for each operational event to obtain a multimodal response sequence includes:

[0016] Based on the start and end times of each operation event in the action sequence, multiple sets of sensors arranged in key parts of the valve body and flow channel are triggered to collect data synchronously, and multi-channel raw signals corresponding to the event are obtained.

[0017] Based on the structural characteristics and signal transmission path of the valve, the vibration, acoustic emission and pressure fluctuation signals in the multi-channel raw signals are spatially assigned to obtain a response signal with position attributes.

[0018] Aligning positional response signals belonging to the same operational event along the time axis yields a multimodal response sequence with a spatiotemporal correlation structure.

[0019] Preferably, the step of performing time-series causal inference on the response signal mode corresponding to each operational action event to obtain the causal effect strength of the response signal includes:

[0020] Cross-correlation calculation is performed on the vibration signal and acoustic emission signal in the multimodal response sequence to obtain the time shift corresponding to the maximum cross-correlation value. This time shift is determined as the first dominant time lag in which the acoustic emission signal leads the vibration signal.

[0021] Align the acoustic emission signal and the vibration signal with the first dominant time lag, and calculate the proportion of the energy variance of the vibration signal that can be linearly explained by the acoustic emission signal under this alignment state to obtain the first causal effect intensity value.

[0022] Preferably, the causal effect intensity matrix composed of multiple causal effect intensity values ​​includes:

[0023] Under the same operational event, the causal effect intensity values ​​of vibration signal on pressure fluctuation signal and pressure fluctuation signal on vibration signal are normalized with the causal effect intensity value of vibration originating from acoustic emission to obtain a set of comparable standardized intensity values.

[0024] Based on the correspondence between the inherent order of signal modes and spatial location labels, the standardized intensity values ​​are arranged into a two-dimensional square matrix. The rows and columns of this two-dimensional square matrix represent the signals as causes and the signals as results, respectively, thus obtaining the causal effect intensity matrix.

[0025] Preferably, comparing the causal effect intensity matrix with the baseline causal pattern of a normal valve includes:

[0026] From the pre-established normal valve sample library, the causal effect intensity matrix of each historical sample is extracted, and the intensity values ​​at the same row and column positions in all matrices are statistically analyzed to obtain the benchmark interval mapping relationship that characterizes the allowable fluctuation range of the intensity of each causal edge under normal conditions.

[0027] Each element in the causal effect intensity matrix of the valve to be tested is compared one by one with the corresponding interval in the baseline interval mapping relationship to determine whether it falls within the allowable fluctuation range, thus obtaining a consistency comparison matrix composed of Boolean values.

[0028] Preferably, the analysis of causal deviation to obtain causal anomaly indicators includes:

[0029] Based on the consistency comparison matrix, identify all abnormal intensity values ​​that fall outside the allowable fluctuation range, and trace back to the corresponding specific signal mode causal edge according to their row and column positions in the matrix;

[0030] For each abnormal intensity value, calculate its standard deviation multiple relative to the mean intensity of the same side in historical normal data, and use the standard deviation multiple as a contribution factor.

[0031] By weighting and aggregating all anomaly intensity values ​​with their corresponding contribution factors, a quantitative causal anomaly index is obtained.

[0032] Preferably, determining whether the valve has internal leakage based on causal anomaly indicators includes:

[0033] Based on the action time sequence, extract the continuous change curve of the causal anomaly index on the entire operation action event time axis;

[0034] Identify whether the change curve shows a monotonous and rapid upward trend during the valve opening phase, and whether it remains at a high level and the fluctuation amplitude is lower than the preset threshold during the subsequent valve stabilization phase;

[0035] If both a monotonous and rapid upward trend and a high-level maintenance state are met simultaneously, it is determined that there are signs of internal leakage.

[0036] Preferably, generating an internal leakage detection report based on the judgment result includes:

[0037] When signs of internal leakage are identified, the top N causal edges with the largest contribution to their intensity value during the stable phase are selected from the causal effect intensity matrix and marked as key abnormal causal paths.

[0038] The report integrates the conclusions of the internal leakage indications, the mean and variance of the change curve in the stable phase, and the specific signal modes involved in the key abnormal causal path into an internal leakage detection report.

[0039] To address the above problems, the present invention also provides a performance testing system for industrial valves, the system comprising:

[0040] The action monitoring module is used to monitor and record the valve's operation action events to obtain the action timing sequence;

[0041] The multi-signal synchronous acquisition module is used to synchronously acquire the multimodal response signals of the valve for each operation event, and obtain the multimodal response sequence; the multimodal response signals include at least vibration signals, acoustic emission signals, and pressure fluctuation signals;

[0042] The causal inference module is used to perform time-series causal inference on the response signal mode corresponding to each operation action event, obtain the causal effect intensity of the response signal, and form a causal effect intensity matrix composed of multiple causal effect intensity values;

[0043] The pattern comparison module is used to compare the causal effect intensity matrix with the baseline causal pattern of a normal valve, analyze the causal deviation, and obtain causal anomaly indicators.

[0044] The report generation module is used to determine whether a valve has internal leakage based on causal anomaly indicators, and to generate an internal leakage detection report based on the judgment results.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] 1. By synchronously acquiring multimodal response signals of valve vibration, acoustic emission, and pressure fluctuation, and combining time series causal inference to explore the causal effect strength between signals, it can accurately capture the weak signal correlation characteristics caused by microscopic damage to the sealing surface under dynamic operating conditions. It can effectively identify intermittent and early weak internal leakage under varying operating conditions, media containing particulate matter, or temperature and pressure fluctuations, significantly improving the detection efficiency of early valve failures and ensuring the comprehensiveness and timeliness of valve sealing performance monitoring in industrial processes.

[0047] 2. By leveraging the action time sequence carrying driving state characteristics to achieve precise spatiotemporal alignment of multimodal signals, and then using the causal effect intensity matrix to structurally characterize the multidimensional signal correlation, the synergistic effect of the two not only achieves quantitative assessment of the degree of anomaly, but also accurately locks the causal path of key anomalies, making the detection results both quantitatively accurate and fault-oriented, providing a clear technical basis for valve predictive maintenance, and greatly improving the scientificity and efficiency of maintenance decisions. Attached Figure Description

[0048] Figure 1 This is a schematic flowchart of a performance testing method for industrial valves provided in an embodiment of the present invention;

[0049] Figure 2 This is a functional block diagram of a performance testing system for industrial valves provided in an embodiment of the present invention;

[0050] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0051] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0052] This application provides a performance testing method for industrial valves. The execution entity of this performance testing method for industrial valves includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, a performance testing method for industrial valves can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be a standalone server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0053] Example 1, referring to Figure 1 The diagram shown is a flowchart illustrating a performance testing method for industrial valves according to an embodiment of the present invention. In this embodiment, the performance testing method for industrial valves includes:

[0054] S1. Monitor and record the valve's operation events to obtain the action timing sequence;

[0055] S2. For each operational event, synchronously acquire the valve's multimodal response signal to obtain a multimodal response sequence; the multimodal response signal includes at least vibration signal, acoustic emission signal, and pressure fluctuation signal;

[0056] S3. Perform time series causal inference on the response signal mode corresponding to each operation action event to obtain the causal effect intensity of the response signal, and form a causal effect intensity matrix composed of multiple causal effect intensity values;

[0057] S4. Compare the causal effect intensity matrix with the baseline causal pattern of a normal valve, analyze the causal deviation, and obtain the causal anomaly index.

[0058] S5. Based on causal anomaly indicators, determine whether the valve has internal leakage, and generate an internal leakage detection report based on the judgment results.

[0059] In this embodiment of the invention, the operation events of the valve are monitored and recorded to obtain an action timing sequence, including:

[0060] The valve stem rotation is monitored using a high-precision encoder to obtain a valve stem rotation angle-time series.

[0061] Based on the valve stem angle-time sequence, continuous angle changes are divided into action segments to obtain independent operation action events in units of complete opening and closing cycles;

[0062] For each independent operation event, the current signal of the drive device is synchronously acquired, and the time-domain waveform of the current signal is time-aligned and marked with the valve stem rotation angle-time sequence to obtain the action time sequence carrying the drive state characteristics.

[0063] In practice, a high-precision encoder is mechanically coupled to the valve stem of an industrial valve, enabling the encoder to synchronously collect angle data as the valve stem rotates. During the collection process, the angle values ​​are continuously recorded at fixed time intervals, and a one-to-one correspondence is established between each time point and the corresponding angle value, forming a valve stem angle-time sequence arranged in chronological order.

[0064] Specifically, the time interval for data acquisition is dynamically determined based on the valve operation rate. The faster the valve stem rotates, the shorter the time interval is set, thus fully capturing all the details of the angle changes during the valve stem rotation process and ensuring that the valve stem angle-time series can accurately reflect the actual rotation state of the valve stem.

[0065] Furthermore, a segmented analysis was conducted on the continuous angle changes in the valve stem angle-time series to identify the opening action as the stage in the sequence where the angle changes continuously from the initial closed value to the maximum opening value and then remains stable.

[0066] Then, the phase in which the angle continuously changes from the maximum opening value to the initial closing value and remains stable at the close is identified as the closing action. One consecutive opening action and one closing action are integrated into a complete opening-closing cycle. The valve stem angle-time series is divided into action segments with each complete opening-closing cycle as the boundary. Each segment corresponds to an independent operation action event based on the complete opening and closing cycle, ensuring that each independent operation action event completely contains all the angle change information of the corresponding opening and closing cycle.

[0067] Finally, while monitoring the valve stem rotation, the current signal of the valve drive device is continuously acquired through the current acquisition element. The acquisition time interval of the current signal is kept completely consistent with the time interval of the encoder acquiring the rotation angle data. The current value of each time node in the time domain waveform of the current signal is precisely matched with the rotation angle value of the same time node in the valve stem rotation angle-time sequence. The current value of the corresponding time node is directly marked at the corresponding time position of the valve stem rotation angle-time sequence, so that the sequence is integrated with the current state information of the drive device, and finally a timing sequence of actions carrying drive state characteristics is formed.

[0068] In summary, this solution uses a high-precision encoder to capture the valve stem rotation angle-time sequence in real time, not just recording the start and end points. It can completely reconstruct the details of the action process and simultaneously collect the drive current and align it with the time, deeply integrating mechanical action with drive state. It can predict potential problems at the drive end through current fluctuations and avoid missing hidden faults by relying on a single mechanical signal. By dividing the complete opening and closing cycle into independent events, it accurately eliminates interference from segmented actions and provides a millisecond-level time reference for subsequent multi-modal signal synchronous acquisition.

[0069] Overall, this solution achieves high-precision restoration of action timing, provides status warnings, and lays a precise spatiotemporal foundation for subsequent causal inference of signals. This dual gain effect cannot be achieved by a single mechanical or drive signal monitoring method.

[0070] In this embodiment of the invention, for each operational event, the multimodal response signal of the valve is synchronously acquired to obtain a multimodal response sequence, including:

[0071] Based on the start and end times of each operation event in the action sequence, multiple sets of sensors arranged in key parts of the valve body and flow channel are triggered to collect data synchronously, and multi-channel raw signals corresponding to the event are obtained.

[0072] Based on the structural characteristics and signal transmission path of the valve, the vibration, acoustic emission and pressure fluctuation signals in the multi-channel raw signals are spatially assigned to obtain a response signal with position attributes.

[0073] Aligning positional response signals belonging to the same operational event along the time axis yields a multimodal response sequence with a spatiotemporal correlation structure.

[0074] In practice, the start time of each independent operation event in the action sequence is used as the trigger signal, and multiple sets of sensors pre-arranged in key parts of the valve body and flow channel are activated simultaneously.

[0075] Specifically, vibration sensors are attached to the valve body sidewall and valve seat connection, acoustic emission sensors are fixed to the corresponding housing position of the valve core, and pressure sensors are embedded in the inlet and outlet of the flow channel and both sides of the valve core. All sensors synchronously acquire signals at time intervals consistent with the valve stem rotation angle acquisition. The acquisition process continues until the end of the operation event, and finally, multi-channel raw signals covering three types of signals—vibration, acoustic emission, and pressure fluctuation—corresponding to this event are obtained, ensuring that signal acquisition and operation event are completely synchronized.

[0076] Furthermore, by combining the structural characteristics of the valve, the signal transmission path of each part is clarified. When the vibration signal is transmitted along the metal structure of the valve body, it will be accompanied by energy attenuation. When the acoustic emission signal is transmitted between the fluid and the shell, the frequency characteristics are different. The pressure fluctuation signal only propagates along the medium in the flow channel.

[0077] Based on these characteristics, the original signals from multiple channels are identified one by one, the source of each signal is determined, and the corresponding acquisition location information is marked for vibration, acoustic emission and pressure fluctuation signals respectively. The spatial assignment of the signals is completed, and a response signal with position attributes is obtained to ensure that each signal corresponds to a clear valve body or flow channel position.

[0078] Furthermore, using the start time of the operation action event as a unified time reference, all response signals with position attributes belonging to the event are time-calibrated, and the time axes of each signal are adjusted so that their starting points completely coincide.

[0079] Each subsequent time point corresponds to the same action stage, while retaining the positional attribute labels of each signal, so that signals of different modes and different locations form a precise correspondence in the time dimension, ultimately obtaining a multimodal response sequence with a spatiotemporal correlation structure, providing a signal foundation with both time synchronization and spatial orientation for subsequent analysis.

[0080] In summary, this solution uses precise start and end times of action sequence to trigger data acquisition, avoiding redundant signals caused by fixed-frequency acquisition and reducing subsequent processing load. By combining valve structure and signal transmission path for spatial attribution, the specific source location of each signal is clearly defined, solving the problem of ambiguous correspondence between signals and valve body parts, and providing support for subsequent fault location.

[0081] Overall, the spatiotemporal correlation structure multimodal sequence that is finally constructed achieves both time synchronization and preservation of spatial attributes, allowing signals from different modalities and locations to form strong correlations. This not only improves signal effectiveness but also enables precise identification of signal correlation paths for subsequent causal inferences through spatiotemporal linkage. This level of precision and directionality cannot be achieved by solutions that only perform time alignment or do not allocate space.

[0082] In this embodiment of the invention, time-series causal inference is performed on the response signal mode corresponding to each operation action event to obtain the causal effect strength of the response signal, including:

[0083] Cross-correlation calculation is performed on the vibration signal and acoustic emission signal in the multimodal response sequence to obtain the time shift corresponding to the maximum cross-correlation value. This time shift is determined as the first dominant time lag in which the acoustic emission signal leads the vibration signal.

[0084] Align the acoustic emission signal and the vibration signal with the first dominant time lag, and calculate the proportion of the energy variance of the vibration signal that can be linearly explained by the acoustic emission signal under this alignment state to obtain the first causal effect intensity value.

[0085] The formula for calculating the intensity value of the first causal effect is as follows:

[0086] ;

[0087] ;

[0088] In the formula, This represents the strength value of the first causal effect. This indicates the time-aligned acoustic emission signal at time [time value missing]. amplitude, This indicates the vibration signal after time alignment at time... amplitude, This indicates a time lag in the first dominant factor.

[0089] In the formula, Indicated by The vibration signal fitted value obtained through linear regression is used as the independent variable. This represents the intercept of the linear regression model. The slope coefficient of the linear regression model is represented by the equation of the equation. and The result was obtained by least-squares fitting. Represents the aligned vibration signal In terms of time length The average value within, Indicates the duration of the selected action event.

[0090] In practice, vibration signals and acoustic emission signals with the same location attribute in the multimodal response sequence are selected. Using the vibration signal as a reference, the acoustic emission signals are time-shifted for different durations one by one. After each time shift is completed, the sum of the products of the amplitudes of the two sets of signals at the corresponding time is calculated to obtain the cross-correlation value corresponding to the time shift.

[0091] Furthermore, after traversing all possible time shifts, the time shift with the largest cross-correlation value is selected and determined as the first dominant time lag that the acoustic emission signal leads the vibration signal, ensuring that the time shift can accurately reflect the temporal correlation between the two sets of signals.

[0092] Furthermore, the time axis of the acoustic emission signal is adjusted according to the first dominant time lag to align the acoustic emission signal with the vibration signal in time, thus obtaining the aligned acoustic emission signal and vibration signal.

[0093] Furthermore, least squares fitting is performed on the two aligned signals. By repeatedly adjusting the intercept and slope coefficient, the sum of squares of the difference between the vibration signal fitting value calculated with the acoustic emission signal amplitude as the independent variable and the actual vibration signal amplitude is minimized, thus determining the final intercept and slope coefficient.

[0094] Finally, calculate the average amplitude of the vibration signal at all times after alignment, and then calculate the sum of squares of the deviations between the vibration signal amplitude and the average value at each time, as well as the sum of squares of the deviations between the vibration signal amplitude and the corresponding fitted value at each time. Subtract the latter from the former and divide the result by the former to obtain the value of the first causal effect intensity.

[0095] The causal effect strength matrix, which comprises multiple causal effect strength values, includes:

[0096] Under the same operational event, the causal effect intensity values ​​of vibration signal on pressure fluctuation signal and pressure fluctuation signal on vibration signal are normalized with the causal effect intensity value of vibration originating from acoustic emission to obtain a set of comparable standardized intensity values.

[0097] Based on the correspondence between the inherent order of signal modes and spatial location labels, the standardized intensity values ​​are arranged into a two-dimensional square matrix. The rows and columns of this two-dimensional square matrix represent the signals as causes and the signals as results, respectively, thus obtaining the causal effect intensity matrix.

[0098] The formula for calculating the standardized strength value is as follows:

[0099] ;

[0100] In the formula, Represents any pair of signal modes arrive The original causal effect strength value, , This represents the mean and standard deviation of the reference dataset used for normalization. This represents the standardized causal effect strength value.

[0101] In practice, the same method as the first causal effect intensity value is used to calculate the original causal effect intensity values ​​of the vibration signal on the pressure fluctuation signal and the pressure fluctuation signal on the vibration signal under the same operation action event.

[0102] Furthermore, multiple sets of normal valve operation events under the same working conditions are selected, and the three types of original causal effect intensity values ​​corresponding to each event are extracted to form a reference dataset. The mean and standard deviation of all values ​​in the reference dataset are calculated. The mean is obtained by summing all values ​​and dividing by the total number of values, and the standard deviation is obtained by taking the square root of the sum of squares of the deviations of each value from the mean and dividing by the total number of values.

[0103] Then, the mean is subtracted from the original causal effect strength value for each category, and then divided by the standard deviation to obtain a set of standardized strength values ​​that can be compared with each other.

[0104] The inherent order of signal modes is pre-defined as acoustic emission signal, vibration signal, and pressure fluctuation signal. At the same time, a mapping relationship is established with the spatial position labels of each signal to ensure that each signal mode and position label form a unique correspondence.

[0105] Finally, the standardized intensity values ​​are arranged in a two-dimensional matrix. The rows of the matrix represent the signals as causes and their corresponding positions, and the columns represent the signals as results and their corresponding positions. Each standardized intensity value is filled into the intersection of the corresponding cause signal row and result signal column to form a causal effect intensity matrix, which can clearly show the causal correlation intensity between different signal modes.

[0106] It should be noted that those skilled in the art can also use other methods, such as Granger causality testing, to preprocess multimodal signals for stationarity, construct an autoregressive model, determine the causal relationship and calculate the test statistic by judging whether the lag term of one signal can significantly improve the prediction accuracy of another signal, use the statistic as the strength of the causal effect, and then arrange them into a matrix according to the signal mode category; or calculate the amplitude change rate of different mode signals within the same time window, use the lead-lag relationship of the change rate to determine the causal direction, determine the strength according to the ratio of the change rate, and divide the matrix into rows and columns only according to the signal type.

[0107] In summary, the normalization process of this scheme uses the mean and standard deviation of the reference dataset under normal operating conditions to eliminate differences in signal magnitudes, making the intensity values ​​comparable across operating conditions. The matrix construction combines the inherent order of signal modes with spatial location labels, which not only reflects the causal relationship between modes but also associates the location of the signal source, forming a spatiotemporally coupled causal matrix.

[0108] Overall, this solution avoids the dependence of traditional causal inference on stationary signals and allows the matrix to have both modal correlation and spatial orientation. Subsequently, the signal path corresponding to the fault can be accurately located through the matrix. This is a precise and traceable causal analysis effect that cannot be achieved by solutions that only focus on modality or ignore spatial correlation, and it is suitable for the complex operating conditions of industrial valves.

[0109] In this embodiment of the invention, the causal effect intensity matrix is ​​compared with the baseline causal pattern of a normal valve, including:

[0110] From the pre-established normal valve sample library, the causal effect intensity matrix of each historical sample is extracted, and the intensity values ​​at the same row and column positions in all matrices are statistically analyzed to obtain the benchmark interval mapping relationship that characterizes the allowable fluctuation range of the intensity of each causal edge under normal conditions.

[0111] Each element in the causal effect intensity matrix of the valve to be tested is compared one by one with the corresponding interval in the baseline interval mapping relationship to determine whether it falls within the allowable fluctuation range, thus obtaining a consistency comparison matrix composed of Boolean values.

[0112] In practice, the causal effect intensity matrix of all historical samples is extracted from a pre-established normal valve sample library. The sample library consists of causal effect intensity matrices generated after the same type of valve completes operation events under different normal operating conditions, covering a variety of common working load and media parameter scenarios.

[0113] Furthermore, all intensity values ​​are categorized according to the same row and column positions in the matrix. For the categorized values ​​at each row and column position, the mean and standard deviation are calculated. The mean is obtained by summing all values ​​at that position and dividing by the total number of values. The standard deviation is obtained by first calculating the sum of squared deviations of each value from the mean, then dividing by the total number of values ​​and taking the square root.

[0114] Then, with the mean as the center, the allowable fluctuation range of the intensity value at that location is determined by adding or subtracting three times the standard deviation from the mean. A correspondence is established between each row and column position and the corresponding allowable fluctuation range to obtain the benchmark interval mapping relationship that represents the normal state.

[0115] Furthermore, the causal effect intensity matrix of the valve to be tested is extracted, and each element in the matrix is ​​extracted one by one. Based on the row and column position of the element, the corresponding allowable fluctuation range is retrieved from the benchmark interval mapping relationship.

[0116] Furthermore, it is determined whether the value of the element is within the allowed fluctuation range. If it is within the range, it is recorded as true; if it is outside the range, it is recorded as false. According to the row and column order of the causal effect intensity matrix of the valve to be tested, all the Boolean values ​​corresponding to the judgment results are arranged in sequence to obtain a consistency comparison matrix composed of Boolean values, ensuring that the matrix structure is completely consistent with the causal effect intensity matrix of the valve to be tested.

[0117] In the implementation of this invention, the causal deviation is analyzed to obtain causal anomaly indicators, including:

[0118] Based on the consistency comparison matrix, identify all abnormal intensity values ​​that fall outside the allowable fluctuation range, and trace back to the corresponding specific signal mode causal edge according to their row and column positions in the matrix;

[0119] For each abnormal intensity value, calculate its standard deviation multiple relative to the mean intensity of the same side in historical normal data, and use the standard deviation multiple as a contribution factor.

[0120] By weighting and aggregating all anomaly intensity values ​​with their corresponding contribution factors, a quantitative causal anomaly index is obtained.

[0121] The contribution factor is calculated using the following formula:

[0122] ;

[0123] In the formula, This indicates the position of the valve in the causal effect intensity matrix under test. line, number The strength value of the column, This indicates that all samples are in the same position in the matrix. The average intensity value, Indicates standard deviation, This represents the contribution factor, i.e., the standard deviation multiple.

[0124] In practice, the consistency comparison matrix is ​​traversed to filter out all positions where the records are false. Based on the row and column identifiers corresponding to these positions, the abnormal intensity value corresponding to the causal effect intensity matrix of the valve to be tested is located.

[0125] Then, by combining the correspondence between the rows and columns of the causal effect intensity matrix and the signal mode and spatial location, we trace back the cause signal mode, result signal mode and specific spatial acquisition location corresponding to each abnormal intensity value, clarify the causal edge of the specific signal mode to which each abnormal intensity value belongs, and establish a unique association between the abnormal intensity value and the causal edge.

[0126] Furthermore, for each abnormal intensity value, the mean and standard deviation of its corresponding row and column position in the historical data of normal samples are retrieved. These values ​​are derived from the previous statistical analysis results of the same row and column positions in the normal sample matrix.

[0127] Then, subtract the corresponding mean from the abnormal intensity value, and divide the difference by the corresponding standard deviation. The result is the contribution factor. The larger the value of this factor, the more significantly the abnormal intensity value deviates from the normal level.

[0128] Finally, weights are assigned based on the degree of influence of each signal mode causal edge on the valve internal leakage fault. Causal edges with high internal leakage correlation have higher weights than those with low correlation. Each anomaly intensity value is multiplied by the corresponding contribution factor and weight, and all product results are summed to obtain a quantitative causal anomaly index.

[0129] In summary, this solution constructs a dynamic benchmark interval by statistically analyzing the mean and standard deviation of the same row and column positions of normal samples, rather than a fixed threshold. This can adapt to signal fluctuations under normal operating conditions, avoid misjudgments, and solve the problem that fixed thresholds cannot cope with differences in operating conditions. Furthermore, by comparing each element to generate a Boolean matrix, it can accurately locate local abnormal elements. Compared with overall similarity analysis, it can avoid the defect that serious local anomalies are masked by overall compliance.

[0130] Meanwhile, by quantifying the degree of abnormal deviation through contribution factors and combining it with the weighted aggregation of causal edge fault correlation, the abnormal indicators not only reflect the number of abnormalities, but also the severity of the abnormalities and their impact weights. Furthermore, it is possible to trace back the causal edges of the signal modes corresponding to the abnormalities, providing direction for subsequent fault localization.

[0131] Overall, this solution balances accuracy, adaptability, and traceability, avoiding both misjudgments and omissions, while also quantifying anomaly levels and pinpointing problem paths, providing more comprehensive support for valve fault diagnosis.

[0132] In this invention, determining whether a valve has internal leakage based on causal anomaly indicators includes:

[0133] Based on the action time sequence, extract the continuous change curve of the causal anomaly index on the entire operation action event time axis;

[0134] Identify whether the change curve shows a monotonous and rapid upward trend during the valve opening phase, and whether it remains at a high level and the fluctuation amplitude is lower than the preset threshold during the subsequent valve stabilization phase;

[0135] If both a monotonous and rapid upward trend and a high-level maintenance state are met simultaneously, it is determined that there are signs of internal leakage.

[0136] In practice, the causal anomaly index values ​​are extracted at each moment by combining the time nodes of the action sequence. The values ​​are then connected in sequence according to time to draw a continuous change curve of the causal anomaly index on the time axis of the entire operation action event. This ensures that each data point on the curve corresponds precisely to a specific moment in the action sequence, fully presenting the change pattern of the index with the operation action, and that the curve trend is synchronized with the switching of valve operation stages.

[0137] Furthermore, based on the characteristics of the valve stem angle change in the action sequence, the time stages are divided. The stage in which the valve stem angle changes continuously but does not reach the maximum opening is the valve opening stage, and the stage in which the angle remains constant is the valve stabilization stage.

[0138] Furthermore, observe the change curve during the initial stage. If the curve value continues to increase over time and the growth rate does not decrease significantly, it indicates a monotonous and rapid upward trend.

[0139] Specifically, the preset fluctuation threshold is set by statistically analyzing the fluctuation data of the indicators during the stable phase of the same type of normal valve. The values ​​of causal abnormal indicators during the stable phase of the normal valve are collected multiple times, the deviation of each value from the mean is calculated, and the maximum deviation is taken as the preset threshold.

[0140] Finally, observe the curve during the stable phase. If the value remains at a high level above the final value of the opening phase, and the deviation between the value at all times and the average value of the stable phase is lower than the preset threshold, then the high-level maintenance state is satisfied. If both of these conditions are met, it is determined that there is an internal leakage.

[0141] In this embodiment of the invention, generating an internal leakage detection report based on the judgment result includes:

[0142] When signs of internal leakage are identified, the top N causal edges with the largest contribution to their intensity value during the stable phase are selected from the causal effect intensity matrix and marked as key abnormal causal paths.

[0143] The report integrates the conclusions of the internal leakage indications, the mean and variance of the change curve in the stable phase, and the specific signal modes involved in the key abnormal causal path into an internal leakage detection report.

[0144] In practice, when signs of internal leakage are detected, all causal edge strength values ​​corresponding to the valve's stable phase are extracted from the causal effect strength matrix, and the causal edges are sorted in descending order of strength value.

[0145] It should be noted that the preset N value is determined by combining the number of common abnormal causal edges of internal leakage faults of valves of the same model, and is adjusted to balance the accuracy of fault location and the efficiency of troubleshooting.

[0146] Then, select the top N causal edges in the sorting and mark them as key anomaly causal paths. Simultaneously record the cause signal mode, result signal mode, and specific spatial acquisition location corresponding to each path to clarify the signal correlation of the anomaly source.

[0147] Furthermore, the mean and variance of all indicator values ​​in the stable phase of the change curve are calculated. The mean is obtained by summing all indicator values ​​in the stable phase and dividing by the total number of values. The variance is obtained by calculating the sum of squared deviations of each indicator value from the mean and then dividing by the total number of values.

[0148] Finally, the conclusions on the identification of internal leakage signs, the mean and variance data of the stable phase curve, the causal paths of key anomalies and the corresponding signal mode combinations are systematically organized, and the source and core meaning of each piece of information are clearly marked to avoid omissions or ambiguities. In the end, a complete and detailed internal leakage detection report is formed, which provides an accurate basis for subsequent fault location and investigation.

[0149] In summary, this solution combines the continuous change curve of action timing analysis indicators, focuses on the trend characteristics specific to internal leakage, and can accurately distinguish between internal leakage and indicator anomalies caused by operating condition fluctuations and instantaneous interference, avoiding misjudgments and omissions. Furthermore, through two-stage trend verification, it locks in the core characteristics of internal leakage, rather than simply relying on numerical magnitude, and adapts to the indicator fluctuation patterns under different valve operating conditions.

[0150] Overall, this solution report combines accuracy in judgment with guidance in troubleshooting, enabling precise leak identification and source tracing. It addresses the pain point of traditional methods that can only qualitatively determine the problem but cannot pinpoint the problem, thus significantly improving the efficiency of subsequent maintenance.

[0151] Implementation 2, such as Figure 2 The diagram shown is a functional block diagram of a performance testing system for industrial valves provided in an embodiment of the present invention.

[0152] This invention discloses a performance testing system for industrial valves that can be installed in an electronic device. Depending on the functions implemented, the performance testing system for industrial valves may include an action monitoring module 101, a multi-signal synchronous acquisition module 102, a causal inference module 103, a pattern comparison module 104, and a report generation module 105. The modules of this invention can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and perform a fixed function, and are stored in the memory of the electronic device.

[0153] In this embodiment, the functions of each module / unit are as follows:

[0154] The action monitoring module 101 is used to monitor and record the operation action events of the valve to obtain the action timing sequence;

[0155] The multi-signal synchronous acquisition module 102 is used to synchronously acquire the multi-modal response signals of the valve for each operation action event to obtain a multi-modal response sequence; the multi-modal response signals include at least vibration signals, acoustic emission signals, and pressure fluctuation signals;

[0156] The causal inference module 103 is used to perform time series causal inference on the response signal mode corresponding to each operation action event, obtain the causal effect intensity of the response signal, and form a causal effect intensity matrix composed of multiple causal effect intensity values;

[0157] The pattern comparison module 104 is used to compare the causal effect intensity matrix with the baseline causal pattern of a normal valve, analyze the causal deviation, and obtain the causal anomaly index.

[0158] The report generation module 105 is used to determine whether there is internal leakage in the valve based on causal anomaly indicators, and to generate an internal leakage detection report based on the judgment results.

[0159] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0160] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0161] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0162] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0163] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0164] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for performance testing of an industrial valve, characterized in that, The method comprises: S1, monitoring and recording the operation action event of the valve to obtain the action time sequence; S2, for each operation action event, synchronously collecting the multi-modal response signal of the valve to obtain the multi-modal response sequence; the multi-modal response signal at least includes vibration signal, acoustic emission signal and pressure fluctuation signal; S3, for each operation action event, time sequence causal inference is carried out on the corresponding response signal mode to obtain the causal effect intensity of the response signal, and a causal effect intensity matrix composed of multiple causal effect intensity values is obtained; S4, comparing the causal effect intensity matrix with the reference causal mode of the normal valve, analyzing the causal deviation degree, and obtaining the causal anomaly index; S5, judging whether the valve exists internal leakage based on the causal anomaly index, and generating an internal leakage detection report according to the judgment result.

2. A method for performance testing of an industrial valve as claimed in claim 1, wherein, The monitoring and recording of the operation action event of the valve to obtain the action time sequence comprises: monitoring the rotation action of the valve stem of the valve based on a high-precision encoder to obtain a valve stem rotation angle-time sequence; According to the valve stem rotation angle-time sequence, the continuous rotation angle change is divided into action segments to obtain independent operation action events in the unit of complete opening and closing cycles; For each independent operation action event, the current signal of the driving device is synchronously collected, and the time domain waveform of the current signal is time-aligned and marked with the valve stem rotation angle-time sequence to obtain an action time sequence carrying driving state characteristics.

3. A method for performance testing of an industrial valve as claimed in claim 1, wherein, The synchronously collecting of the multi-modal response signal of the valve for each operation action event to obtain the multi-modal response sequence comprises: Based on the start and end time of each operation action event in the action time sequence, a plurality of groups of sensors arranged at key positions of the valve body and the flow passage are triggered to synchronously collect to obtain multi-channel original signals corresponding to the event; According to the structural characteristics and signal transmission path of the valve, the vibration, acoustic emission and pressure fluctuation signals in the multi-channel original signals are spatially attributed and distributed to obtain response signals with position attributes; The response signals with position attributes belonging to the same operation action event are aligned along the time axis to obtain a multi-modal response sequence with a space-time correlation structure.

4. The method for performance testing of industrial valves as claimed in claim 1 wherein, The time sequence causal inference on the response signal mode corresponding to each operation action event to obtain the causal effect intensity of the response signal comprises: Cross-correlation calculation is performed on the vibration signal and the acoustic emission signal in the multi-modal response sequence to obtain the time shift corresponding to the maximum cross-correlation value, and the time shift is determined as the first dominant time lag of the acoustic emission signal leading the vibration signal; The vibration signal and the acoustic emission signal are aligned with the first dominant time lag, and the proportion of the energy variance of the vibration signal that can be linearly explained by the acoustic emission signal under the alignment state is calculated to obtain the first causal effect intensity value.

5. A method for performance testing of an industrial valve as claimed in claim 4, wherein, The causal effect intensity matrix composed of multiple causal effect intensity values comprises: The causal effect intensity values of the vibration signal on the pressure fluctuation signal and the pressure fluctuation signal on the vibration signal under the same operation action event are normalized with the causal effect intensity value of the vibration source from the acoustic emission to obtain a group of comparable standardized intensity values; According to the correspondence between the inherent order of the signal modal and the spatial position label, the normalized intensity values are arranged into a two-dimensional square matrix, the rows and columns of which represent the cause and the result respectively, so as to obtain a cause-effect intensity matrix.

6. A method for performance testing of an industrial valve as claimed in claim 1, wherein, The comparison of the cause-effect intensity matrix with the reference cause-effect pattern of the normal valve comprises: From the pre-established normal valve sample library, the cause-effect intensity matrix of each historical sample is extracted, and statistical characteristic analysis is performed on the intensity values in the same row and column positions of all the matrices to obtain a reference interval mapping relationship representing the allowable fluctuation range of each cause-effect edge under the normal state; Each element in the cause-effect intensity matrix generated by the valve to be detected is compared with the corresponding interval in the reference interval mapping relationship one by one to determine whether it falls within the allowable fluctuation range, and a consistency comparison matrix composed of Boolean values is obtained.

7. A method for performance testing of an industrial valve as claimed in claim 6, wherein, The analysis of the cause-effect deviation degree to obtain a cause-effect abnormality index comprises: According to the consistency comparison matrix, all abnormal intensity values falling outside the allowable fluctuation range are identified, and according to their row and column positions in the matrix, the corresponding specific signal modal cause-effect edge is traced back; For each abnormal intensity value, the standard deviation multiple of the same edge intensity mean in the historical normal data is calculated, and the standard deviation multiple is taken as a contribution factor; All abnormal intensity values and their corresponding contribution factors are weighted and aggregated to obtain a quantitative cause-effect abnormality index.

8. A method for performance testing of an industrial valve as claimed in claim 2, wherein, The judgment of whether the valve has internal leakage based on the cause-effect abnormality index comprises: According to the action time sequence, the continuous change curve of the cause-effect abnormality index on the entire operation action event time axis is extracted; It is identified whether the change curve presents a monotonous rapid rising trend in the valve opening stage and whether it is maintained at a high level with a fluctuation amplitude lower than a preset threshold in the subsequent valve stable stage; If both the monotonous rapid rising trend and the high level maintenance state are satisfied, it is determined that there is an internal leakage sign.

9. A method for performance testing of an industrial valve as claimed in claim 8, wherein, The generation of an internal leakage detection report according to the judgment result comprises: When it is determined that there is an internal leakage sign, the first N cause-effect edges with the largest intensity value contribution in the stable stage are screened out from the cause-effect intensity matrix and marked as key abnormal cause-effect paths; The internal leakage sign determination conclusion, the mean and variance of the change curve in the stable stage, and the specific signal modal combination involved in the key abnormal cause-effect paths are integrated into an internal leakage detection report.

10. A performance testing system for industrial valves for implementing a performance testing method for industrial valves according to any one of claims 1 to 9, characterized in that, The system comprises: An action monitoring module for monitoring and recording the operation action events of the valve to obtain an action time sequence; A multi-signal synchronous acquisition module for synchronously acquiring the multi-modal response signals of the valve for each operation action event to obtain a multi-modal response sequence; the multi-modal response signals at least include vibration signals, acoustic emission signals and pressure fluctuation signals; A cause-effect inference module for performing time sequence cause-effect inference on the response signal modal corresponding to each operation action event to obtain the cause-effect intensity of the response signal, and composing a cause-effect intensity matrix composed of multiple cause-effect intensity values; A pattern comparison module for comparing the cause-effect intensity matrix with the reference cause-effect pattern of the normal valve, analyzing the cause-effect deviation degree, and obtaining a cause-effect abnormality index; The report generation module is configured to determine whether the valve has internal leakage based on the causal abnormality index, and generate an internal leakage detection report according to a determination result.

Images