A petrochemical plant equipment health management and fault diagnosis method and system
By using real-time data collection and automatic division of operating conditions, a health baseline model was established and deviation indices were calculated. This solved the problem of health assessment and fault diagnosis of petrochemical equipment under varying operating conditions, and enabled accurate equipment health status assessment and fault type identification.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HEBEI MINTAI SAFETY EVALUATION CONSULTING CO LTD
- Filing Date
- 2026-03-18
- Publication Date
- 2026-06-16
AI Technical Summary
In the existing technology, the health status assessment and fault diagnosis methods of petrochemical plant equipment cannot adapt to the changing operating conditions of the plant, resulting in false alarms and missed alarms, and it is difficult to accurately identify the fault type.
By collecting equipment operating parameters and related process parameters in real time, the system automatically classifies operating conditions and establishes a health baseline model, calculates deviation indicators, assigns weights to form a comprehensive health index, and matches it with a pre-built fault mode library to determine the fault type.
It enables quantitative assessment of the health status of petrochemical plant equipment and accurate identification of fault types, improving the accuracy and timeliness of fault diagnosis.
Smart Images
Figure CN122221098A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of health management and fault diagnosis technology for petrochemical equipment, and more specifically, to a method and system for health management and fault diagnosis of petrochemical equipment. Background Technology
[0002] Currently, in the petrochemical industry, equipment health management and fault diagnosis technologies have evolved from traditional manual inspections to online monitoring. Early equipment condition monitoring relied primarily on periodic inspections and maintenance, with maintenance personnel using portable measuring instruments to collect parameters such as vibration and temperature, combining this with personal experience to judge equipment condition. With the development of sensor and computer technologies, online monitoring systems based on fixed threshold alarms emerged, providing real-time alerts for exceeding alarm limits for various monitoring parameters. In recent years, data-driven fault diagnosis methods have been widely applied, including control chart methods based on statistical analysis, time-frequency domain feature analysis methods based on signal processing, and support vector machines and neural network classification methods based on machine learning. Simultaneously, some advanced systems have begun to introduce hybrid modeling methods combining mechanistic models and data-driven approaches, constructing virtual mappings of equipment through digital twin technology to achieve condition assessment and fault simulation. Furthermore, fault diagnosis technologies based on case-based reasoning and expert systems have also found some application in industrial settings, providing decision support for on-site maintenance by establishing fault case libraries and rule bases.
[0003] In existing technologies, traditional fixed-threshold alarm methods cannot adapt to the characteristics of variable operating conditions in petrochemical plants. Under different loads and process stages, the range of normal operating parameters varies significantly, and fixed thresholds easily lead to a large number of false alarms or missed alarms. Monitoring methods based on single parameters cannot comprehensively reflect the overall health status of the equipment. Parameters are interrelated and have different sensitivities to faults; isolated analysis of a single parameter often fails to accurately identify early fault symptoms. Conventional data-driven diagnostic methods lack effective utilization of parameter deviation directions and combination patterns. Large amounts of continuous monitoring data are difficult to directly correlate with fault types, resulting in poor interpretability of diagnostic results. Therefore, how to achieve quantitative assessment of the health status of petrochemical plant equipment and accurate identification of fault types, and improve the accuracy and timeliness of fault diagnosis for petrochemical plant equipment, is a problem facing the industry. Summary of the Invention
[0004] This application provides a method and system for health management and fault diagnosis of petrochemical equipment, which can realize the quantitative assessment of the health status of petrochemical equipment and the accurate identification of fault types, thereby improving the accuracy and timeliness of fault diagnosis of petrochemical equipment.
[0005] In a first aspect, this application provides a method for health management and fault diagnosis of petrochemical equipment, the method comprising the following steps:
[0006] The system collects the equipment operating parameter set and associated process parameter set of the target petrochemical plant equipment in real time. Based on the associated process parameter set, the system automatically divides the equipment operating conditions and identifies and marks the operating state range of the equipment in different process stages.
[0007] For each defined operating condition, historical health operation data under the corresponding operating condition is extracted, and then a health baseline model of the equipment under each operating condition is established. The set of equipment operating parameters is compared with the corresponding equipment health baseline model to obtain various deviation indicators.
[0008] Weights are assigned to each deviation index to obtain the comprehensive health index of the target petrochemical equipment. When the comprehensive health index is lower than the preset alarm threshold, pattern encoding is performed on each deviation index to form the deviation pattern vector of the current state.
[0009] The deviation pattern vector is matched with the standard patterns in the pre-built fault deviation pattern library, and the fault type is determined based on the matching result.
[0010] In this embodiment, the operating parameters include vibration, temperature, and current, and the associated process parameters include medium flow rate, inlet and outlet pressure, and material temperature.
[0011] In this embodiment, based on the associated process parameter set, the automatic division of equipment operating conditions and the identification and marking of the equipment's operating state intervals at different process stages specifically include:
[0012] A multi-dimensional operating condition feature vector is constructed from the set of associated process parameters collected within a preset time window;
[0013] Using the aforementioned operating condition feature vector as input, cluster analysis is performed on historical operating data to divide the continuous process into multiple operating condition clusters with similar process characteristics, and the cluster center of each operating condition cluster is determined.
[0014] For the associated process parameters collected in real time, determine their membership degree relative to each cluster center, and determine the working condition category to which the current sampling point belongs based on the principle of maximum membership degree.
[0015] The operating condition identification results on the continuous time series are smoothed to obtain the intervals of each operating state.
[0016] In this embodiment, for each defined operating condition, historical health operation data under the corresponding condition is extracted, and a health baseline model for each operating condition is established. Specifically, this includes:
[0017] For each operating condition, the set of equipment operating parameters for the corresponding operating condition marked time period is extracted from the historical database, and the operating data of the equipment in a healthy state is used as the health sample set.
[0018] Normality tests were performed on the operating parameters of each device in the healthy sample set to determine the distribution characteristics of each device's operating parameters under the corresponding operating conditions.
[0019] Based on the distribution characteristics, the statistical features of each operating parameter under healthy conditions are extracted, and then a health baseline model of the equipment under each operating condition is established.
[0020] In this embodiment, the set of equipment operating parameters is compared with the corresponding equipment health baseline model to obtain the deviation index of each equipment operating parameter, which specifically includes:
[0021] Based on the currently identified operating conditions, the equipment health baseline model established under the corresponding operating conditions is invoked to obtain the baseline statistical characteristics of each equipment operating parameter under the corresponding operating conditions.
[0022] The measured operating parameters of each device are compared with the corresponding baseline central trend parameters, and the deviation of each device's operating parameters is determined by combining the dispersion parameters.
[0023] Based on the historical fluctuation characteristics of the operating parameters of each device and the operating status of the device, the deviation of each parameter is corrected to obtain the corresponding deviation index.
[0024] In this embodiment, the deviation index is used to quantify the degree to which the current state of a parameter deviates from its health benchmark.
[0025] In this embodiment, the deviation index of each parameter is weighted to obtain the comprehensive health index of the target petrochemical plant equipment, which specifically includes:
[0026] Based on the various deviation indicators within the preset time window, a deviation fusion matrix is constructed;
[0027] Based on the aforementioned deviation fusion matrix, the objective weight of each device operating parameter in the device health status assessment is determined;
[0028] By integrating the various deviation indicators at the current moment with their corresponding objective weights, a comprehensive health index for the target petrochemical plant equipment is obtained.
[0029] In this embodiment, the comprehensive health index is a single quantitative indicator used to quantitatively characterize the overall health status of petrochemical plant equipment.
[0030] In this embodiment, matching the deviation pattern vector with standard patterns in a pre-built fault deviation pattern library and determining the fault type based on the matching result specifically includes:
[0031] The deviation pattern vector is compared with each standard fault pattern vector stored in the fault deviation pattern library to determine the fault matching degree sequence.
[0032] Based on the fault matching degree sequence, standard fault modes with a fault matching degree exceeding a preset threshold are selected as fault types.
[0033] Secondly, this application provides a health management and fault diagnosis system for petrochemical equipment, used to execute a method for health management and fault diagnosis of petrochemical equipment, the health management and fault diagnosis system comprising:
[0034] The data acquisition module is used to collect the set of equipment operating parameters and related process parameters of the target petrochemical plant equipment in real time. Based on the set of related process parameters, the operating conditions of the equipment are automatically divided, and the operating status range of the equipment in different process stages is identified and marked.
[0035] The deviation determination module is used to extract historical health operation data for each of the divided operating conditions, and then establish equipment health baseline models for each operating condition. The set of equipment operating parameters is compared with the corresponding equipment health baseline models to obtain various deviation indicators.
[0036] The pattern encoding module is used to assign weights to each deviation index to obtain the comprehensive health index of the target petrochemical equipment. When the comprehensive health index is lower than the preset alarm threshold, the pattern encoding is performed on each deviation index to form the deviation pattern vector of the current state.
[0037] The type matching module is used to match the deviation pattern vector with the standard patterns in the pre-built fault deviation pattern library, and determine the fault type based on the matching result.
[0038] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects:
[0039] The system collects real-time equipment operating parameter sets and associated process parameter sets of the target petrochemical plant equipment. Based on the associated process parameter sets, it automatically divides the equipment operating conditions, identifies and marks the operating state intervals of the equipment in different process stages. For each divided operating condition, it extracts historical health operation data for that condition and establishes an equipment health baseline model for each condition. The equipment operating parameter sets are compared with the corresponding equipment health baseline models to obtain various deviation indicators. Weights are assigned to each deviation indicator to obtain a comprehensive health index for the target petrochemical plant equipment. When the comprehensive health index is lower than a preset alarm threshold, each deviation indicator is pattern-encoded to form a deviation pattern vector for the current state. The deviation pattern vector is matched with standard patterns in a pre-built fault deviation pattern library, and the fault type is determined based on the matching results.
[0040] Therefore, this application firstly collects equipment operating parameters and related process parameters in real time, and automatically classifies and marks equipment operating conditions based on the related process parameters. This enables subsequent equipment health assessments to use differentiated benchmark models for different operating conditions, avoiding false alarms and missed alarms caused by changes in operating conditions, and laying a data foundation for establishing an adaptive health management system. Secondly, by establishing an equipment health baseline model for each operating condition and calculating the deviation index of each parameter based on this baseline model, a quantitative comparison between the equipment operating status and the health benchmark under the corresponding operating condition is achieved. This unifies operating parameters with different physical meanings into comparable dimensionless deviation indices, avoiding false alarms caused by changes in operating conditions. This approach addresses misjudgments caused by deviations, improving the accuracy and reliability of health assessments. Then, by weighting and fusing various deviation indicators, multi-parameter status information is compressed into a single comprehensive health index, enabling an intuitive quantitative expression of the overall equipment health level. This facilitates rapid understanding of the equipment's status. When the health index falls below a threshold, pattern encoding is triggered, transforming continuous deviations into discrete symbolic pattern vectors. This provides standardized input for subsequent fault pattern matching, enhancing the targeting and efficiency of fault diagnosis. Finally, by matching the deviation pattern vector of the current state with standard patterns in a pre-built fault deviation pattern library, rapid and accurate identification of equipment fault types is achieved, improving the intelligence level and response speed of fault diagnosis.
[0041] In summary, the technical solution adopted in this application can realize the quantitative assessment of the health status of petrochemical plant equipment and the accurate identification of fault types, thereby improving the accuracy and timeliness of fault diagnosis of petrochemical plant equipment. Attached Figure Description
[0042] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this embodiment of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0043] Figure 1 This is an exemplary flowchart of a method for health management and fault diagnosis of petrochemical equipment provided in this application;
[0044] Figure 2 This is a flowchart illustrating the process of obtaining various deviation indices based on the information provided in this application;
[0045] Figure 3 This is a flowchart illustrating the process of obtaining the comprehensive health index of the target petrochemical plant equipment based on the information provided in this application.
[0046] Figure 4 This is a calculation flowchart based on vibration deviation provided in this application.
[0047] Figure 5 This is a flowchart of the calculation of a comprehensive health index based on a deviation index fusion provided in this application.
[0048] Figure 6 This is a modular structure diagram of a petrochemical plant equipment health management and fault diagnosis system provided in this application. Detailed Implementation
[0049] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.
[0050] This application provides a method and system for health management and fault diagnosis of petrochemical plant equipment. Its core is the real-time acquisition of equipment operating parameter sets and associated process parameter sets of the target petrochemical plant equipment. Based on the associated process parameter sets, the operating conditions of the equipment are automatically divided, and the operating state intervals of the equipment in different process stages are identified and marked. For each divided operating condition, historical health operation data under the corresponding condition is extracted, and a health baseline model for each condition is established. The equipment operating parameter sets are compared with the corresponding equipment health baseline models to obtain various deviation indicators. Weights are assigned to each deviation indicator to obtain a comprehensive health index for the target petrochemical plant equipment. When the comprehensive health index is lower than a preset alarm threshold, each deviation indicator is pattern-encoded to form a deviation pattern vector for the current state. The deviation pattern vector is matched with standard patterns in a pre-constructed fault deviation pattern library, and the fault type is determined based on the matching result.
[0051] Example 1: To better understand the above technical solution, the following will provide a detailed description of the technical solution in conjunction with the accompanying drawings and specific implementation methods. (Refer to...) Figure 1 As shown in the figure, this is an exemplary flowchart of a method for health management and fault diagnosis of petrochemical equipment according to this embodiment of the present application. The health management and fault diagnosis method includes the following steps:
[0052] In step S1, the equipment operating parameter set and associated process parameter set of the target petrochemical plant equipment are collected in real time. Based on the associated process parameter set, the equipment operating conditions are automatically divided, and the operating status range of the equipment in different process stages is identified and marked.
[0053] In practical implementation, firstly, the system can collect real-time sets of equipment operating parameters and related process parameters of the target petrochemical plant. Specifically, sensors can be deployed on key equipment. For rotating equipment such as compressors and pumps, temperature-vibration composite sensors can be installed at the bearing housing to collect effective values of vibration velocity and bearing temperature. For static equipment such as reactors and heat exchangers, acoustic emission sensors can be installed on the walls to monitor material damage signals. Secondly, related process parameters, including media flow rate, inlet and outlet pressure, and material temperature, are read from the plant's distributed control system via a communication interface. The sampling frequency is set to once per minute, and data synchronization processing is performed. All acquisition devices synchronize with the server time via the NTP protocol to ensure that each data point has a unified timestamp. High-frequency vibration data undergoes feature value calculation at the acquisition terminal, and only the feature values are uploaded. Operating parameters include vibration, temperature, and current, while related process parameters include media flow rate, inlet and outlet pressure, and material temperature.
[0054] In this embodiment, the automatic division of equipment operating conditions based on the associated process parameter set, and the identification and marking of the equipment's operating state intervals at different process stages, can be achieved through the following steps:
[0055] A multi-dimensional operating condition feature vector is constructed from the set of associated process parameters collected within a preset time window;
[0056] Using the aforementioned operating condition feature vector as input, cluster analysis is performed on historical operating data to divide the continuous process into multiple operating condition clusters with similar process characteristics, and the cluster center of each operating condition cluster is determined.
[0057] For the associated process parameters collected in real time, determine their membership degree relative to each cluster center, and determine the working condition category to which the current sampling point belongs based on the principle of maximum membership degree.
[0058] The operating condition identification results on the continuous time series are smoothed to obtain the intervals of each operating state.
[0059] In specific implementation, firstly, a multi-dimensional operating condition feature vector is constructed using the set of associated process parameters collected within a preset time window. That is, for each sampling moment, a time window of a preset length is taken with the current moment as the center, and the statistical characteristics of each parameter within the window are calculated to form a multi-dimensional operating condition feature vector. For each associated process parameter, its mean, standard deviation, and rate of change are extracted as features. The three-dimensional features of multiple parameters are combined to form a multi-dimensional operating condition feature vector. Secondly, using the operating condition feature vector as input, cluster analysis is performed on historical operating data to divide the continuous process into multiple... Clusters of operating conditions with similar process characteristics are identified, and the cluster centers of each cluster are determined. Specifically, fuzzy C-means clustering can be used to cluster the historical operating condition feature vectors to determine the number of clusters. The optimal number of clusters is selected using the silhouette coefficient method. Clustering is performed within a preset range, and the silhouette coefficients corresponding to each cluster number are calculated. The cluster number corresponding to the point with the largest silhouette coefficient is selected as the optimal number of clusters. After determining the number of clusters, the membership matrix and cluster centers are randomly initialized. Using the set of operating condition feature vectors as input, the objective function of fuzzy C-means clustering is iteratively optimized. This objective function is... The weighted sum of squared distances between samples and each cluster center is used. The cluster center and membership matrix are updated alternately during the iterative process until the objective function reaches its maximum iteration count, thus obtaining the cluster centers. Then, for the real-time acquired associated process parameters, their membership degree relative to each cluster center is determined. Based on the maximum membership principle, the operating condition category of the current sampling point is determined. That is, the Euclidean distance between the multi-dimensional operating condition feature vector and each cluster center can be calculated. Then, the membership degree calculation formula in fuzzy C-means clustering is used to convert the Euclidean distance into membership degree. Based on the maximum membership principle, the value of the Euclidean distance is selected. The operating condition category corresponding to the maximum membership degree is taken as the operating condition category to which the current sampling point belongs. Finally, the operating condition identification results on the continuous time series can be smoothed to obtain each operating state interval. That is, for each time point on the continuous time series, the frequency of each operating condition category in the window is counted, and the operating condition category with the highest frequency is taken as the final operating condition identification result at that time. After smoothing, the time points that are continuous on the time axis and have the same operating condition category are merged into one operating state interval. The start time and end time of each interval are recorded to obtain each operating state interval.
[0060] In step S2, for each of the defined operating conditions, historical health operation data under the corresponding operating conditions are extracted, and then a health baseline model for each operating condition is established. The set of equipment operating parameters is compared with the corresponding equipment health baseline model to obtain various deviation indicators.
[0061] In this embodiment, for each defined operating condition, historical health operation data under the corresponding condition is extracted, and then the equipment health baseline model under each condition is established. This can be achieved through the following steps:
[0062] For each operating condition, the set of equipment operating parameters for the corresponding operating condition marked time period is extracted from the historical database, and the operating data of the equipment in a healthy state is used as the health sample set.
[0063] Normality tests were performed on the operating parameters of each device in the healthy sample set to determine the distribution characteristics of each device's operating parameters under the corresponding operating conditions.
[0064] Based on the distribution characteristics, the statistical features of each operating parameter under healthy conditions are extracted, and then a health baseline model of the equipment under each operating condition is established.
[0065] In practical implementation, firstly, for each operating condition, the set of equipment operating parameters within the corresponding marked time period is extracted from the historical database. The operating data of the equipment in a healthy state is used as the health sample set. That is, based on the start and end times of the operating state interval, the time-series data of all equipment operating parameters within that time period is extracted. Then, the extracted data is cleaned, removing data from equipment failure periods, downtime maintenance periods, and non-steady-state operating periods. Specifically, by querying equipment maintenance records and alarm logs, data within a preset time window before and after a failure is marked and removed. By identifying the transition periods at both ends of the operating condition interval, data within a preset duration before and after the operating condition switch is removed, retaining only the data from the stable operating period. The remaining data is then used as the health sample set when the equipment is in a healthy state under the corresponding operating condition. For each operating condition, the health sample set is stored in matrix form, with rows corresponding to sampling times and columns corresponding to different equipment operating parameters. Then, it can be... Normality tests were performed on the operating parameters of each device in the healthy sample set to determine the distribution characteristics of each parameter under the corresponding operating conditions. Specifically, the Shapiro-Wilke test was used to determine the statistical distribution characteristics of each operating parameter under the current operating conditions. A pre-set significance level was used. For each parameter, the Shapiro-Wilke statistic and its corresponding p-value were calculated for the healthy sample. If the p-value was greater than the significance level, the null hypothesis was accepted, and the parameter was determined to follow a normal distribution under the current operating conditions. If the p-value was less than or equal to the significance level, the null hypothesis was rejected, and the parameter was determined to not follow a normal distribution. For parameters that failed the normality test, histograms were further plotted for visual verification to confirm whether their distribution was skewed or multimodal. The test results were recorded in the form of distribution type labels, thus obtaining the distribution characteristics of each device operating parameter under the corresponding operating conditions.Finally, based on the distribution characteristics, statistical features of each operating parameter under healthy conditions can be extracted, thereby establishing a health baseline model for equipment under various operating conditions. Specifically, appropriate statistical feature extraction methods can be used for parameters with different distribution types. For parameters determined to follow a normal distribution, parametric methods are used to extract statistical features, calculating the arithmetic mean and standard deviation of the healthy sample, and simultaneously calculating the minimum and maximum values of the sample as a reference for the actual observation range. For parameters determined not to follow a normal distribution, nonparametric methods are used to extract statistical features, calculating the median of the healthy sample as a measure of central tendency, calculating the interquartile range as a measure of dispersion, and further calculating the 5th and 95th percentiles as reference limits for the normal fluctuation range. For parameters exhibiting a clear multimodal distribution, kernel density estimation is used to fit its probability density function, and multiple central trend points are determined through the local maxima of the density curve. For multiple parameters with significant correlations, based on single-parameter analysis, the correlation coefficient matrix between parameters is further calculated to establish multidimensional joint distribution characteristics. For parameters following a multivariate normal distribution... The parameter set calculates the mean vector and covariance matrix. For parameter sets that do not follow a multivariate normal distribution, a joint probability density function is established using a multidimensional kernel density estimation method. Based on the above statistical characteristics, an equipment health baseline model is constructed for each operating condition. That is, the equipment health baseline model is stored in structured data form and includes the following core elements: operating condition identifier; parameter list, containing the names and units of all equipment operating parameters under that operating condition; for each parameter, storing its distribution type label, central tendency statistic, dispersion statistic, and normal fluctuation range boundary; for parameter sets, storing the correlation coefficient matrix or multidimensional distribution parameters; model metadata, including the number of samples used for modeling, data time range, and modeling timestamp. In the equipment health baseline model, for parameters that follow a normal distribution, the normal fluctuation range is defined as the mean plus or minus a preset multiple of the standard deviation; for parameters that do not follow a normal distribution, the normal fluctuation range is defined as a preset percentile interval. The baseline model also includes probability density function parameters, which can be used to calculate the probability of any measured value occurring in a healthy state, thus obtaining the equipment health baseline model for each operating condition.
[0066] Preferably, in this embodiment, the set of equipment operating parameters is compared with the corresponding equipment health baseline model to obtain various deviation indicators, which are then used as references. Figure 2 As shown in the figure, this is a flowchart illustrating the process of obtaining various deviation indices in some embodiments of this application. The specific steps for obtaining each deviation index in this embodiment are as follows:
[0067] In step S21, based on the currently identified operating conditions, the equipment health baseline model established under the corresponding operating conditions is invoked to obtain the baseline statistical characteristics of each equipment operating parameter under the corresponding operating conditions.
[0068] In step S22, the measured operating parameters of each device are compared with the corresponding baseline center trend parameters, and the deviation of each device's operating parameters is determined by combining the dispersion parameters.
[0069] In step S23, the deviation of each parameter is corrected based on the historical fluctuation characteristics of each device's operating parameters and the device's operating status to obtain the corresponding deviation index.
[0070] In practical implementation, firstly, based on the currently identified operating conditions, the corresponding equipment health baseline model can be invoked to obtain the baseline statistical characteristics of each equipment operating parameter under the corresponding operating conditions. That is, the baseline statistical characteristics of all required equipment operating parameters can be obtained from the equipment health baseline model. For parameters modeled using parametric methods and following a normal distribution, their baseline mean and baseline standard deviation are obtained; for parameters modeled using non-parametric methods, their median and interquartile range or specified percentiles are obtained; for parameter sets modeled using multivariate methods, their mean vector and covariance matrix are obtained. Simultaneously, the health status of each parameter is read from the baseline model. The normal fluctuation range under the current state is used as a reference limit for subsequent judgment. Then, the currently measured operating parameters of each device can be compared with the corresponding baseline central trend parameters, and combined with the dispersion parameter, the parameter deviation of each device operating parameter can be determined. That is, the currently collected real-time measured values of each device operating parameter can be compared with the corresponding baseline statistical characteristics to calculate the basic deviation of each parameter. According to the distribution characteristics of each parameter in the health baseline model, different formulas are used. For parameters that follow a normal distribution and are modeled using the parametric method, the current measured value is subtracted from the baseline mean, the absolute value of the result is taken, and then the result is divided by the baseline standard. The deviation is used as the baseline deviation. For parameters that do not follow a normal distribution and are modeled using nonparametric methods, the baseline deviation is calculated by subtracting the baseline median from the current measured value, dividing the absolute value of the result by the interquartile range of the baseline, and then using this result as the baseline deviation. For parameter sets modeled using multivariate methods, the baseline deviation is calculated using the Mahalanobis distance formula. While calculating the baseline deviation, the deviation direction of each parameter is recorded, i.e., whether the current measured value deviates positively or negatively from the baseline center, indicated by a positive or negative sign. Finally, the result is used as the parameter deviation of each device's operating parameter. Finally, the historical fluctuations of each device's operating parameters can be used as a basis for further analysis. Based on the characteristics and operating status of the equipment, the deviation of each parameter is corrected to obtain the corresponding deviation index. That is, for each equipment operating parameter, its fluctuation characteristics during historical healthy operation are analyzed, and the coefficient of variation of each parameter in the historical healthy data is calculated, which is the ratio of the standard deviation to the mean. The larger the coefficient of variation, the more drastic the fluctuation of the parameter during normal operation; the smaller the coefficient of variation, the more stable the parameter during normal operation. Based on the comparison of the coefficient of variation of each parameter with the average coefficient of variation of all parameters, the volatility correction coefficient is determined. For parameters with large historical volatility, a correction coefficient of less than 1 is assigned to appropriately reduce their deviation and avoid misjudging normal fluctuations as abnormal.For parameters with low historical volatility, a correction coefficient greater than 1 is assigned to appropriately increase their deviation and highlight their sensitivity to abnormal changes. The volatility correction coefficient is multiplied by the parameter's base deviation to obtain the volatility-corrected parameter deviation. This deviation is then adjusted based on the current operating condition identification confidence level. Specifically, the membership degree of the current sampling point relative to its assigned operating condition is obtained from the real-time operating condition identification results and used as the operating condition identification confidence level. An operating condition stability correction coefficient is calculated based on this confidence level; the correction coefficient is 1 when the confidence level is 1, and decreases accordingly when the confidence level is below a preset threshold. This correction coefficient is multiplied by the volatility-corrected deviation to obtain the operating condition stability-corrected parameter deviation. The parameter deviation after both volatility and operating condition stability corrections is used as the parameter's deviation index, thus obtaining various deviation indices. It should be noted that the deviation index is used to quantitatively characterize the degree to which the parameter's current state deviates from its healthy baseline.
[0071] like Figure 4 As shown, the horizontal axis represents the time series T1 to T10, and the vertical axis represents the vibration value. The figure shows the measured vibration value curve, the mean healthy baseline curve, the basic deviation curve, and the final deviation index curve. The basic deviation is calculated by comparing the measured vibration value of the equipment at each time with the mean healthy baseline. The final deviation index is obtained by combining the operating condition fluctuation correction and confidence level correction, thus reflecting the degree of deviation of the equipment vibration state from the healthy state and providing basic parameters for equipment condition assessment.
[0072] It should be noted that by establishing a health baseline model for each operating condition and calculating the deviation index of each parameter based on the health baseline model, a quantitative comparison between the equipment operating status and the health benchmark under the corresponding operating condition is realized. Operating parameters with different physical meanings are uniformly transformed into comparable dimensionless deviation indices, avoiding misjudgments caused by differences in operating conditions and improving the accuracy and reliability of health assessment.
[0073] In step S3, each deviation index is weighted to obtain the comprehensive health index of the target petrochemical plant equipment. When the comprehensive health index is lower than the preset alarm threshold, each deviation index is pattern encoded to form the deviation pattern vector of the current state.
[0074] Preferably, in this embodiment, weights are assigned to each deviation index to obtain the comprehensive health index of the target petrochemical plant equipment, with reference to... Figure 3 As shown in the figure, this is a flowchart illustrating the process of obtaining the comprehensive health index of the target petrochemical plant equipment in some embodiments of this application. In this embodiment, obtaining the comprehensive health index of the target petrochemical plant equipment can be achieved through the following steps:
[0075] In step S31, a deviation fusion matrix is constructed based on the various deviation indices within a preset time window;
[0076] In step S32, based on the deviation fusion matrix, the objective weight of each device operating parameter in the device health status assessment is determined;
[0077] In step S33, the deviation indicators at the current moment are fused with their corresponding objective weights to obtain the comprehensive health index of the target petrochemical plant equipment.
[0078] In practical implementation, firstly, a deviation fusion matrix can be constructed based on the various deviation indicators within a preset time window. That is, if the number of equipment operating parameters is defined as m and the time window length is n, the constructed deviation fusion matrix is an m-row, n-column matrix. Matrix elements represent the deviation indicator of the i-th equipment operating parameter at time j. The time window length is determined based on the equipment type and monitoring requirements. When constructing the matrix, the original deviation data is preprocessed to remove outliers and missing values. For a small number of missing data points, linear interpolation between adjacent time points is used to fill them. For obviously abnormal outliers, the median of the preceding and following time points is used for replacement. Then, the constructed matrix is standardized using the extreme value standardization method to obtain the deviation fusion matrix. Next, based on the deviation fusion matrix, the objective weight of each equipment operating parameter in the equipment health status assessment can be determined. That is, the variation richness of each equipment operating parameter can be calculated using the following formula:
[0079]
[0080] in, This represents the richness of variation in the operating parameters of the i-th device; This represents the matrix elements; variation richness is used to measure the degree of variation of equipment operating parameters. The smaller the variation richness, the greater the difference in deviation of the equipment operating parameters at different times, that is, the greater the degree of variation of the equipment operating parameters and the more information they contain. The objective weight of each equipment operating parameter in the equipment health status assessment can be obtained by the following formula:
[0081]
[0082] in, This represents the i-th objective weight; Let represent the richness of variation of the i-th variable, thus obtaining the objective weight of each equipment operating parameter in the equipment health status assessment; finally, the deviation indicators at the current moment can be fused with the corresponding objective weights to obtain the comprehensive health index of the target petrochemical plant equipment. That is, the deviation indicators can be weighted and summed with the corresponding objective weights, and the result can be used as the comprehensive health index of the target petrochemical plant equipment. The comprehensive health index is a single quantitative indicator used to quantitatively characterize the overall health status of the petrochemical plant equipment.
[0083] like Figure 5 As shown, vibration deviation, temperature deviation, and current deviation are input into the deviation fusion matrix as multi-source state parameters. The corresponding weights are obtained by statistical analysis of the deviation of each parameter, and weighted fusion processing is performed based on the weight calculation to finally obtain the comprehensive health index.
[0084] In practical implementation, when the comprehensive health index falls below a preset alarm threshold, pattern coding is performed on each deviation indicator to form a deviation pattern vector for the current state. That is, the alarm threshold is set based on historical fault statistics of the equipment. When the comprehensive health index is less than the alarm threshold, the equipment is determined to be in an abnormal state, automatically triggering the fault diagnosis process and executing the pattern coding step. When the comprehensive health index is greater than the alarm threshold, monitoring continues without triggering coding. During the pattern coding step, the deviation indicators of each equipment operating parameter at the current moment are classified into levels, discretizing continuous values into a finite number of level codes. Based on fault diagnosis experience with petrochemical equipment, each deviation level is set, and the threshold for level classification can be set based on historical experience. For example: Normal: Deviation less than 1, indicating that the parameter is within the normal fluctuation range, with a corresponding code value of 0; Deviation: Deviation greater than 1... A value less than 3 indicates a significant parameter abnormality, requiring inspection; the corresponding code value is 2. A serious deviation (greater than 3 but less than 4) indicates a serious parameter abnormality requiring intervention; the corresponding code value is 3. A dangerous deviation (greater than 4) indicates a parameter in a dangerous state requiring immediate action; the corresponding code value is 4. Based on the operating parameters of each device, their level codes are determined, and these level codes are merged into a vector to form the deviation pattern vector for the current state.
[0085] It should be noted that by weighting and fusing various deviation indicators, multi-parameter status information is compressed into a single comprehensive health index, enabling an intuitive quantitative expression of the overall health level of the equipment. This facilitates quick understanding of the equipment status. When the health index falls below a threshold, pattern coding is triggered, transforming continuous deviations into discrete symbolic pattern vectors. This provides standardized input for subsequent fault mode matching, improving the targeting and efficiency of fault diagnosis.
[0086] In step S4, the deviation pattern vector is matched with the standard patterns in the pre-built fault deviation pattern library, and the fault type is determined based on the matching result.
[0087] In this embodiment, the deviation pattern vector is matched with standard patterns in a pre-built fault deviation pattern library, and the fault type is determined based on the matching result. This can be achieved through the following steps:
[0088] The deviation pattern vector is compared with each standard fault pattern vector stored in the fault deviation pattern library to determine the fault matching degree sequence.
[0089] Based on the fault matching degree sequence, standard fault modes with a fault matching degree exceeding a preset threshold are selected as fault types.
[0090] In practical implementation, the deviation pattern vector can be compared with each standard fault pattern vector stored in the fault deviation pattern library to determine the fault matching degree sequence. That is, for each standard fault pattern vector, the proportion of the number of dimensions where the deviation pattern vector and the standard fault pattern vector have consistent encoding in each dimension can be calculated out of the total number of dimensions. The result is used as the fault matching degree of the standard fault pattern vector, thus obtaining each fault matching degree. All fault matching degrees are sorted in descending order to form a sequence, and the result is used as the fault matching degree sequence. Then, based on the fault matching degree sequence, standard fault patterns with fault matching degrees exceeding a preset threshold can be selected as fault types. That is, the standard fault pattern with the highest fault matching degree is selected as the fault type. If the highest fault matching degree is lower than the preset threshold, it is determined to be an unrecognizable abnormal pattern, and an "unrecognized fault" prompt is output.
[0091] Therefore, this application firstly collects equipment operating parameters and related process parameters in real time, and automatically classifies and marks equipment operating conditions based on the related process parameters. This enables subsequent equipment health assessments to use differentiated benchmark models for different operating conditions, avoiding false alarms and missed alarms caused by changes in operating conditions, and laying a data foundation for establishing an adaptive health management system. Secondly, by establishing an equipment health baseline model for each operating condition and calculating the deviation index of each parameter based on this baseline model, a quantitative comparison between the equipment operating status and the health benchmark under the corresponding operating condition is achieved. This unifies operating parameters with different physical meanings into comparable dimensionless deviation indices, avoiding false alarms caused by changes in operating conditions. This approach addresses misjudgments caused by deviations, improving the accuracy and reliability of health assessments. Then, by weighting and fusing various deviation indicators, multi-parameter status information is compressed into a single comprehensive health index, enabling an intuitive quantitative expression of the overall equipment health level. This facilitates rapid understanding of the equipment's status. When the health index falls below a threshold, pattern encoding is triggered, transforming continuous deviations into discrete symbolic pattern vectors. This provides standardized input for subsequent fault pattern matching, enhancing the targeting and efficiency of fault diagnosis. Finally, by matching the deviation pattern vector of the current state with standard patterns in a pre-built fault deviation pattern library, rapid and accurate identification of equipment fault types is achieved, improving the intelligence level and response speed of fault diagnosis.
[0092] In summary, the technical solution adopted in this application can realize the quantitative assessment of the health status of petrochemical plant equipment and the accurate identification of fault types, thereby improving the accuracy and timeliness of fault diagnosis of petrochemical plant equipment.
[0093] Example 2: This application provides a health management and fault diagnosis system for petrochemical equipment, referring to... Figure 6 As shown in the figure, this is a modular structure diagram of a petrochemical plant equipment health management and fault diagnosis system according to this embodiment of the present application. The health management and fault diagnosis system includes:
[0094] The data acquisition module 100 is used to collect the set of equipment operating parameters and the set of associated process parameters of the target petrochemical plant equipment in real time, and automatically divide the equipment operating conditions based on the set of associated process parameters, and identify and mark the operating state range of the equipment in different process stages.
[0095] The deviation determination module 200 is used to extract historical health operation data for each of the divided operating conditions, and then establish equipment health baseline models for each operating condition. The set of equipment operating parameters is compared with the corresponding equipment health baseline models to obtain various deviation indicators.
[0096] The pattern encoding module 300 is used to assign weights to each deviation index to obtain the comprehensive health index of the target petrochemical equipment. When the comprehensive health index is lower than the preset alarm threshold, the pattern encoding of each deviation index is performed to form the deviation pattern vector of the current state.
[0097] The type matching module 400 is used to match the deviation pattern vector with the standard patterns in the pre-built fault deviation pattern library, and determine the fault type based on the matching result.
[0098] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0099] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compactdisc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.
[0100] It should also be noted that 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. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.
Claims
1. A method for health management and fault diagnosis of petrochemical equipment, characterized in that, The health management and fault diagnosis method includes the following steps: The system collects the equipment operating parameter set and associated process parameter set of the target petrochemical plant equipment in real time. Based on the associated process parameter set, the system automatically divides the equipment operating conditions and identifies and marks the operating state range of the equipment in different process stages. For each defined operating condition, historical health operation data under the corresponding operating condition is extracted, and then a health baseline model of the equipment under each operating condition is established. The set of equipment operating parameters is compared with the corresponding equipment health baseline model to obtain various deviation indicators. Weights are assigned to each deviation index to obtain the comprehensive health index of the target petrochemical equipment. When the comprehensive health index is lower than the preset alarm threshold, pattern encoding is performed on each deviation index to form the deviation pattern vector of the current state. The deviation pattern vector is matched with the standard patterns in the pre-built fault deviation pattern library, and the fault type is determined based on the matching result.
2. The method for health management and fault diagnosis of petrochemical equipment as described in claim 1, characterized in that, The set of equipment operating parameters includes vibration, temperature, and current, while the set of associated process parameters includes medium flow rate, inlet and outlet pressure, and material temperature.
3. The method for health management and fault diagnosis of petrochemical equipment as described in claim 1, characterized in that, Based on the aforementioned set of associated process parameters, the operating conditions of the equipment are automatically divided, and the operating state intervals of the equipment in different process stages are identified and marked, specifically including: A multi-dimensional operating condition feature vector is constructed from the set of associated process parameters collected within a preset time window; Using the aforementioned operating condition feature vector as input, cluster analysis is performed on historical operating data to divide the continuous process into multiple operating condition clusters with similar process characteristics, and the cluster center of each operating condition cluster is determined. For the associated process parameters collected in real time, determine their membership degree relative to each cluster center, and determine the working condition category to which the current sampling point belongs based on the principle of maximum membership degree. The operating condition identification results on the continuous time series are smoothed to obtain the intervals of each operating state.
4. The method for health management and fault diagnosis of petrochemical equipment as described in claim 1, characterized in that, For each defined operating condition, historical health operation data for that condition is extracted, and a health baseline model for each condition is established. Specifically, this includes: For each operating condition, the set of equipment operating parameters for the corresponding operating condition marked time period is extracted from the historical database, and the operating data of the equipment in a healthy state is used as the health sample set. Normality tests were performed on the operating parameters of each device in the healthy sample set to determine the distribution characteristics of each device's operating parameters under the corresponding operating conditions. Based on the distribution characteristics, the statistical features of each operating parameter under healthy conditions are extracted, and then a health baseline model of the equipment under each operating condition is established.
5. The method for health management and fault diagnosis of petrochemical equipment as described in claim 1, characterized in that, The set of equipment operating parameters is compared with the corresponding equipment health baseline model to obtain the deviation index of each equipment operating parameter, which specifically includes: Based on the currently identified operating conditions, the equipment health baseline model established under the corresponding operating conditions is invoked to obtain the baseline statistical characteristics of each equipment operating parameter under the corresponding operating conditions. The measured operating parameters of each device are compared with the corresponding baseline central trend parameters, and the deviation of each device's operating parameters is determined by combining the dispersion parameters. Based on the historical fluctuation characteristics of the operating parameters of each device and the operating status of the device, the deviation of each parameter is corrected to obtain the corresponding deviation index.
6. The method for health management and fault diagnosis of petrochemical equipment as described in claim 1, characterized in that, The deviation index is used to quantify the degree to which the current state of a parameter deviates from its health benchmark.
7. The method for health management and fault diagnosis of petrochemical equipment as described in claim 1, characterized in that, By assigning weights to the deviation indices of each parameter, the comprehensive health index of the target petrochemical plant equipment is obtained, which specifically includes: Based on the various deviation indicators within the preset time window, a deviation fusion matrix is constructed; Based on the aforementioned deviation fusion matrix, the objective weight of each device operating parameter in the device health status assessment is determined; By integrating the various deviation indicators at the current moment with their corresponding objective weights, a comprehensive health index for the target petrochemical plant equipment is obtained.
8. The method for health management and fault diagnosis of petrochemical equipment as described in claim 1, characterized in that, The comprehensive health index is a single quantitative indicator used to quantitatively characterize the overall health status of petrochemical plant equipment.
9. The method for health management and fault diagnosis of petrochemical equipment as described in claim 1, characterized in that, The deviation pattern vector is matched with standard patterns in a pre-built fault deviation pattern library, and the fault type is determined based on the matching results, specifically including: The deviation pattern vector is compared with each standard fault pattern vector stored in the fault deviation pattern library to determine the fault matching degree sequence. Based on the fault matching degree sequence, standard fault modes with a fault matching degree exceeding a preset threshold are selected as fault types.
10. A petrochemical plant equipment health management and fault diagnosis system, used to execute the petrochemical plant equipment health management and fault diagnosis method as described in any one of claims 1 to 9, characterized in that, The health management and fault diagnosis system includes: The data acquisition module is used to collect the set of equipment operating parameters and related process parameters of the target petrochemical plant equipment in real time. Based on the set of related process parameters, the operating conditions of the equipment are automatically divided, and the operating status range of the equipment in different process stages is identified and marked. The deviation determination module is used to extract historical health operation data for each of the divided operating conditions, and then establish equipment health baseline models for each operating condition. The set of equipment operating parameters is compared with the corresponding equipment health baseline models to obtain various deviation indicators. The pattern encoding module is used to assign weights to each deviation index to obtain the comprehensive health index of the target petrochemical equipment. When the comprehensive health index is lower than the preset alarm threshold, the pattern encoding is performed on each deviation index to form the deviation pattern vector of the current state. The type matching module is used to match the deviation pattern vector with the standard patterns in the pre-built fault deviation pattern library, and determine the fault type based on the matching result.