A device fault early warning detection method based on digital radar holographic scanning

Abstract

The application discloses a kind of equipment failure early warning detection methods based on digital radar holography scanning, it is related to the technical field of failure early warning, including, the abnormal propagation direction and parameter relationship transmission characteristics between the analysis secondary abnormal distribution set and primary abnormal distribution set, form abnormal evolution chain, extract abnormal structure mode of abnormal evolution chain by structure analysis, and execute fault fitting processing, generate multidimensional fault fitting set;Estimate the abnormal evolution trend of multidimensional fault fitting set, generate operation risk index set, execute abnormal evolution consistency determination to operation risk index set, generate failure early warning result set.The application deduces and constructs abnormal evolution chain by counterfactual state, realizes the explicit modeling to equipment "normal operation relationship mode", makes abnormal change from "discrete identification" to "structured evolution analysis", improves failure early warning foresight and reliability.

Description

Technical Field

[0001] This invention relates to the field of fault early warning technology, and in particular to a device fault early warning and detection method based on digital radar holographic scanning. Background Technology

[0002] As industrial equipment becomes larger, more complex, and more intelligent, equipment operation status monitoring and fault early warning technologies are gradually becoming key technologies for ensuring production safety and improving operational efficiency. Current technologies typically collect multi-source operational monitoring data during equipment operation through distributed control devices or various online monitoring equipment. This data includes analog parameters such as temperature, pressure, vibration, current, and flow rate, and the equipment's operating status is assessed based on statistical analysis, threshold judgment, or simple trend analysis methods. Furthermore, with the development of data-driven methods, some technical solutions incorporate machine learning or data modeling methods to identify and predict abnormal states by modeling and analyzing historical operating data.

[0003] However, existing technologies have certain limitations. Most rely on fixed thresholds or simple statistical features for anomaly detection, lacking a modeling and comparison mechanism for "normal operation relationship patterns." This makes it difficult to accurately distinguish between normal fluctuations and genuine anomalies when equipment operating conditions fluctuate or the environment changes. Furthermore, existing technologies typically only identify "single-point anomalies" or "local anomalies," lacking the ability to analyze the propagation process and structural evolution of anomalies across time and parameter dimensions. This makes it difficult to achieve a holistic characterization of the fault formation process and provide early warnings. In such cases, even if an anomaly can be detected, it is difficult to further determine its development trend and the degree of risk of it evolving into a fault, thus limiting the accuracy and reliability of fault warnings. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a device fault early warning and detection method based on digital radar holographic scanning to solve the problems of insufficient characterization of multi-source operational data coupling relationship and limited ability to identify complex anomaly evolution.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a method for equipment fault early warning and detection based on digital radar holographic scanning, comprising: collecting multi-source operation monitoring data during equipment operation and performing time-series organization processing to generate a parameter collaborative scanning sequence; extracting analog parameters from the parameter collaborative scanning sequence and analyzing the correlation of parameter changes to form a parameter relationship representation set; statistically analyzing historical stable operation segments in the parameter relationship representation set and performing counterfactual state inference to generate a counterfactual reference set; calculating the parameter relationship deviation value of the parameter collaborative scanning sequence relative to the counterfactual reference set, outputting a first-level anomaly distribution set; identifying structural anomalies within the first-level anomaly distribution set through anomaly nesting analysis to generate a second-level anomaly distribution set; analyzing the anomaly propagation direction and parameter relationship transmission characteristics between the second-level anomaly distribution set and the first-level anomaly distribution set to form an anomaly evolution chain; extracting the anomaly structure pattern of the anomaly evolution chain through structural analysis and performing fault fitting processing to generate a multi-dimensional fault fitting set; evaluating the anomaly evolution trend of the multi-dimensional fault fitting set to generate an operation risk index set; performing anomaly evolution consistency judgment on the operation risk index set to generate a fault early warning result set.

[0007] As a preferred embodiment of the equipment fault early warning and detection method based on digital radar holographic scanning according to the present invention, the specific steps for generating the parameter cooperative scanning sequence are as follows: Multi-source operational monitoring data are resampled according to a unified time reference to generate a synchronous observation data sequence. Data consistency governance is performed on outliers and missing values ​​in the synchronous observation data sequence, and sliding time slicing is performed according to continuous time windows to generate a parameter co-scanning sequence.

[0008] As a preferred embodiment of the equipment fault early warning and detection method based on digital radar holographic scanning described in this invention, the specific steps for forming the parameter relationship representation set are as follows: Read the values ​​of temperature, pressure, vibration, current and flow parameters at continuous time positions corresponding to each time window in the parameter collaborative scanning sequence to form an analog parameter sequence; Analyze the synchronous change relationship, response delay relationship, and consistency of change direction among the analog parameters in the analog parameter sequence to form a parameter relationship characterization set.

[0009] As a preferred embodiment of the equipment fault early warning and detection method based on digital radar holographic scanning described in this invention, the specific steps for generating the counterfactual reference set are as follows: Read the parameter relationship representation set formed within the historical operating cycle, and extract the correlation between the analog parameters corresponding to each time window to form a historical parameter relationship sequence; Perform relation trajectory reconstruction processing on the historical parameter relationship sequence, and connect them in chronological order to form a set of historical relation structure trajectories, using the correlation between each analog parameter as nodes. Identify running segments with similar trajectory structures in the historical relationship structure trajectory set, and perform counterfactual state inference by combining the analog quantity parameter sequence in the current detection cycle to generate a counterfactual reference set.

[0010] As a preferred embodiment of the equipment fault early warning and detection method based on digital radar holographic scanning described in this invention, the specific steps for outputting the first-level anomaly distribution set are as follows: The parametric relationship deviation between the parameter cooperative scan sequence and the counterfactual reference set is calculated to form a parametric relationship deviation sequence; Based on the parameter relationship deviation sequence, the abnormal time locations of parameter relationship deviation values ​​within each time window are identified and organized according to the time window to generate a first-level abnormal distribution set.

[0011] As a preferred embodiment of the equipment fault early warning and detection method based on digital radar holographic scanning described in this invention, the specific steps for generating the secondary anomaly distribution set are as follows: The continuous distribution of abnormal deviations in the first-level anomaly distribution set is statistically analyzed in terms of time and parameter dimensions to form a set of abnormal local regions. By analyzing the changing trends and expansion directions of abnormal deviation values ​​in the set of abnormal local regions over continuous time, structural abnormal nodes that appear during the evolution of abnormal deviations are identified, and a secondary abnormal distribution set is generated.

[0012] As a preferred embodiment of the equipment fault early warning and detection method based on digital radar holographic scanning described in this invention, the specific steps for forming the abnormal evolution chain are as follows: The structural anomaly nodes in the secondary anomaly distribution set are associated with the corresponding anomaly deviation segments in the primary anomaly distribution set according to the unified time position and parameter position, forming a set of anomaly associated node pairs; Analyze the distribution relationship of each abnormal associated node in the set in terms of time sequence and the transmission relationship in the direction of parameter change to identify the direction of abnormal propagation; The abnormal related nodes are organized in a chain according to the direction of abnormal propagation to form an abnormal evolution chain.

[0013] As a preferred embodiment of the equipment fault early warning and detection method based on digital radar holographic scanning described in this invention, the specific steps for generating the multidimensional fault fitting set are as follows: The abnormal evolution chain is structurally expanded according to time sequence, and the propagation path and connection relationship between each abnormal associated node are counted to form an abnormal structure path set. Analyze the coordinated change relationship between the analog parameters corresponding to each abnormal associated node in the abnormal structure path set to generate abnormal structure patterns; Based on the abnormal structure pattern, the abnormal associated nodes in the abnormal evolution chain are combined and fitted to generate a multidimensional fault fitting set.

[0014] As a preferred embodiment of the equipment fault early warning and detection method based on digital radar holographic scanning described in this invention, the specific steps for generating the operational risk indicator set are as follows: The fault fitting structures in the multidimensional fault fitting set are read in chronological order, and the frequency and persistence distribution of each fault fitting structure in a continuous time window are statistically analyzed to form a fault structure evolution sequence. The evolution characteristics of fault structures are obtained by analyzing the direction of change, expansion range and propagation speed of each fault fitting structure at continuous time position in the fault structure evolution sequence. Based on the evolution characteristics of the fault structure, the risk of each fault fitting structure is quantified to generate a set of operational risk indicators.

[0015] As a preferred embodiment of the equipment fault early warning detection method based on digital radar holographic scanning described in this invention, the specific steps for generating the fault early warning result set are as follows: Read the operational risk indicators in the operational risk indicator set in chronological order, and statistically analyze the direction and magnitude of change of each operational risk indicator in a continuous time window to form a risk indicator change sequence. Identify persistent risk segments in the risk indicator change sequence, and perform early warning judgment processing based on persistent risk segments to generate a fault early warning result set.

[0016] The beneficial effects of this invention are as follows: By generating a counterfactual reference set through counterfactual state deduction, explicit modeling of the "normal operation relationship pattern" of equipment is achieved, and a comparable structural reference is established in the current detection cycle. This transforms anomaly identification from a single parameter deviation judgment to a structural deviation judgment at the parameter relationship level, improving the ability to distinguish between normal fluctuations and real anomalies in complex working conditions and fluctuating environments. By constructing an anomaly evolution chain, extracting anomaly structural patterns, and performing fault fitting processing, a systematic characterization of the propagation path and evolution law of anomalies in the time and parameter dimensions is achieved. This transforms anomaly identification from "discrete identification" to "structured evolution analysis," continuously assessing fault development trends and supporting risk quantification and early warning judgment, thereby improving the foresight and reliability of fault early warning. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a device fault early warning and detection method based on digital radar holographic scanning.

[0019] Figure 2 This is a schematic diagram of a fault early warning structure based on counterfactual reasoning.

[0020] Figure 3 A schematic diagram of time-parameter coupling modeling for anomalous structures.

[0021] Figure 4 A schematic diagram of the abnormal evolution chain network model. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides a device fault early warning and detection method based on digital radar holographic scanning, including the following steps: S1. Collect multi-source operation monitoring data during equipment operation, perform time-series organization and processing, and generate parameter collaborative scanning sequences.

[0026] S1.1. Perform time resampling processing on multi-source operation monitoring data according to a unified time benchmark to generate synchronous observation data sequences.

[0027] It should be noted that the multi-source operation monitoring data includes temperature parameters, pressure parameters, vibration parameters, current parameters, and flow parameters that are collected in real time by various types of sensors during equipment operation. Among them, temperature parameters are collected by temperature sensors installed at the heat-generating parts of the equipment, pressure parameters are collected by pressure sensors installed in pipelines or closed cavities, vibration parameters are collected by vibration sensors installed in the structural parts of the equipment, current parameters are collected by current detection devices connected to the power supply circuit, and flow parameters are collected by flow sensors installed in the fluid channels. The corresponding sampling time and location are recorded during the collection process.

[0028] Read the sampling time position corresponding to each sub-data in the multi-source operation monitoring data and establish a unified time benchmark; perform time resampling processing on each sub-data in the multi-source operation monitoring data according to the unified time benchmark. In the time resampling process, by selecting the value of the multi-source operation monitoring data in the corresponding time neighborhood at the unified time position as the representative value of that time position, a one-to-one correspondence is established between each sub-data in the multi-source operation monitoring data at the same time position, forming a synchronous observation data sequence.

[0029] S1.2 Perform data consistency management on abnormal and missing values ​​in the synchronous observation data sequence, and perform sliding time slicing according to continuous time windows to generate parameter co-scanning sequences.

[0030] It should be noted that the process involves reading sub-data corresponding to each time position in the synchronous observation data sequence and establishing a time value sequence for each sub-data in chronological order; calculating the value difference between adjacent time positions within a time interval consisting of multiple consecutive time positions (e.g., three or more time positions), and statistically analyzing the average difference and range of all value differences within the current time interval; comparing the value difference corresponding to the current time position with the average difference and the range of difference; if the value difference at the current time position is outside the range of difference, the observed value at the corresponding time position is determined to be an abnormal value, and the abnormal value is replaced by reading the observed values ​​of adjacent time positions before and after that time position; for time positions in the synchronous observation data sequence where no observation record has been formed, supplementary values ​​are generated by reading the observed values ​​of adjacent time positions before and after that time position, constructing a fixed-length continuous time window (e.g., ten consecutive time positions) with a unified time reference, and gradually moving the time window along the time axis in chronological order; reading and combining multi-source operational monitoring data for the corresponding time position within each time window to generate a parameter collaborative scanning sequence.

[0031] S2. Extract analog parameters from the parameter co-scanning sequence and analyze the correlation of parameter changes to form a parameter relationship representation set. Statistically analyze the historical stable operating segments in the parameter relationship representation set and perform counterfactual state deduction to generate a counterfactual reference set.

[0032] S2.1 Read the values ​​of temperature, pressure, vibration, current and flow parameters at continuous time positions corresponding to each time window in the parameter collaborative scanning sequence to form an analog parameter sequence.

[0033] It should be noted that the observed values ​​of temperature, pressure, vibration, current and flow parameters at each time position are read from the parameter collaborative scanning sequence for each continuous time window, and arranged in chronological order so that the temperature, pressure, vibration, current and flow parameters at each time position form a corresponding parameter value sequence, thereby generating an analog parameter sequence.

[0034] S2.2 Analyze the synchronous change relationship, response delay relationship and consistency of change direction among the analog parameters in the analog parameter sequence to form a parameter relationship characterization set.

[0035] It should be noted that the analog parameters are organized according to a unified time position; within a time interval consisting of multiple consecutive time positions (such as ten consecutive time positions), the difference in value of each analog parameter between adjacent time positions is calculated, and the proportion of time positions in which the change direction of the analog parameter value difference is consistent at the same time position is counted, which is used as the synchronous change relationship between the analog parameters; the value difference of one analog parameter at the current time position in the analog parameter sequence is compared one by one with the value difference of another analog parameter at a subsequent time position, and the number of time position intervals in which the change direction is consistent is recorded to characterize the response delay relationship between the analog parameters; within the same time interval, the number of time positions in which the change direction of the analog parameter value difference is consistent or the response delay relationship is consistent is counted, and the ratio between the number of such time positions and the total number of time positions in the corresponding time interval is used as the correlation strength of the analog parameters; based on the synchronous change relationship, response delay relationship, and correlation strength, the analog parameters in each time window are combined and organized to generate a parameter relationship representation set.

[0036] S2.3 Read the parameter relationship representation set formed within the historical operating cycle, and extract the correlation between the analog parameters corresponding to each time window to form a historical parameter relationship sequence.

[0037] It should be noted that the correlation records corresponding to each continuous time window in the parameter relationship representation set formed within the historical operating cycle are read, including the synchronous change relationship, response delay relationship and correlation strength value between analog parameters within each time window; the correlation records are continuously organized according to the time window order within the historical operating cycle to generate a historical parameter relationship sequence.

[0038] S2.4 Perform relationship trajectory reconstruction processing on the historical parameter relationship sequence, using the correlation between each analog parameter as nodes, and connect them in chronological order to form a set of historical relationship structure trajectories.

[0039] It should be noted that the synchronous change relationship, response delay relationship, and correlation strength value corresponding to each continuous time window in the historical parameter relationship sequence are read and combined to form the corresponding parameter relationship record; within the historical operation cycle, the parameter relationship record corresponding to each time window is read in chronological order, and the parameter relationship record between adjacent time windows is continuously connected according to the time position, so that the parameter relationship record within the continuous time window forms a continuous change path; within the continuous time window, each parameter relationship record is used as a trajectory node, and the connection relationship between the nodes is established according to the chronological order of each trajectory node in the time dimension to generate a historical relationship structure trajectory set.

[0040] S2.5 Identify running segments with similar trajectory structures in the historical relationship structure trajectory set, and perform counterfactual state deduction by combining the analog parameter sequence in the current detection cycle to generate a counterfactual reference set.

[0041] It should be noted that the parameter relationship records corresponding to each trajectory node in the historical relationship structure trajectory set are read, and the synchronous change relationship, response delay relationship, and association strength value corresponding to each trajectory node are extracted. Within a trajectory segment consisting of multiple consecutive time windows (such as five consecutive time windows), the combined values ​​of the synchronous change relationship, response delay relationship, and association strength corresponding to each trajectory node are statistically analyzed, and the number of consistent time positions of the combined values ​​in different trajectory segments is compared. The ratio between the number of consistent time positions and the total number of time positions is used as the trajectory structure consistency. When the trajectory structure consistency reaches the condition that most time positions in the trajectory segment are consistent (such as the number of consistent time positions exceeding half of the total number of time positions in the trajectory segment), the corresponding trajectory segment is identified as a running segment with similar trajectory structure. Within the identified running segment with similar trajectory structure, the analog parameter sequence at the corresponding time position is read, and based on the correspondence between the analog parameter sequence in the current detection period and the analog parameter sequence in the historical running segment, the parameter relationship values ​​corresponding to each time position are deduced in the current detection period to generate the counterfactual parameter relationship sequence for the current detection period, which is used as the counterfactual reference set.

[0042] It should also be noted that the parameter relationship values ​​specifically include: the synchronous change relationship values, response delay relationship values, and correlation strength values ​​between the analog parameters at the corresponding time positions.

[0043] The simulation process is based on the synchronous change relationships, response delay relationships, and correlation strength values ​​corresponding to each time position in the historical operating segment. In the current detection cycle, the change difference of each analog parameter at adjacent time positions is obtained. At each time position, the change direction between the current analog parameters is compared one by one with the synchronous change relationship of the corresponding time position in the historical operating segment, and the cases where the change direction is consistent are counted to determine the synchronous change relationship value. The change difference of the current analog parameters between different time positions is compared with the response delay relationship corresponding to the historical operating segment to determine the response delay relationship value. In the continuous time segment, the proportion of time positions that satisfy the synchronous change relationship or response delay relationship is counted as the correlation strength value, thereby determining the parameter relationship value corresponding to each time position.

[0044] S3. Calculate the deviation of the parameter relationship between the parameter collaborative scanning sequence and the counterfactual reference set, output the first-level anomaly distribution set, and identify the structural anomalies within the first-level anomaly distribution set through anomaly nesting analysis to generate the second-level anomaly distribution set.

[0045] S3.1 Calculate the parameter relationship deviation between the parameter cooperative scan sequence and the counterfactual reference set to form a parameter relationship deviation sequence.

[0046] It should be noted that, based on a unified time reference, the parameter deviation values ​​corresponding to each time position are read sequentially. These deviation values ​​are then arranged in chronological order to form a parameter deviation sequence.

[0047] The expression for calculating the deviation of the parameter relationship is: ; ; ; ; ; in, Indicates the first Synchronization deviation characterization value at each time position Indicates the first The delay deviation characterization value at each time location, Indicates the first The intensity deviation from the characterization value at each time location Indicates the first Deviation of parameter relationship at each time point Indicates the first Deviation from discrete adjustment factor at each time location This indicates the current detection cycle is in the [number]th [number]. The synchronous change relationship of each time position. This indicates the current detection cycle is in the [number]th [number]. The synchronous change relationship of each time position. The counterfactual reference set is in the first... The synchronous change relationship of each time position. The counterfactual reference set is in the first... The synchronous change relationship of each time position. This indicates the current detection cycle is in the [number]th [number]. The response delay relationship at each time point is represented by the values. This indicates the current detection cycle is in the [number]th [number]. The response delay relationship at each time point is represented by the values. The counterfactual reference set is in the first... The response delay relationship at each time point is represented by the values. The counterfactual reference set is in the first... The response delay relationship at each time point is represented by the values. This indicates the current detection cycle is in the [number]th [number]. The correlation strength values ​​at each time location, This indicates the current detection cycle is in the [number]th [number]. The correlation strength values ​​at each time location, The counterfactual reference set is in the first... The correlation strength values ​​at each time location, The counterfactual reference set is in the first... The correlation strength values ​​at each time location, This represents a very small positive number, and its purpose is to prevent the denominator from being zero, such as 10. -6 .

[0048] S3.2 Based on the parameter relationship deviation sequence, identify the abnormal time location of the parameter relationship deviation value within each time window, and organize them according to the time window to generate a first-level abnormal distribution set.

[0049] It should be noted that the parameter deviation values ​​corresponding to each time window in the parameter deviation sequence are read, and a one-to-one correspondence between time positions and parameter deviation values ​​is established according to a unified time benchmark. Within each time window, the average value of the parameter deviation and the range between the maximum and minimum values ​​are statistically analyzed. Based on the corresponding average value and range, a deviation fluctuation judgment interval is constructed. The sum of the average value and half of the range is used as the upper bound for deviation fluctuation judgment, and the difference between the average value and half of the range is used as the lower bound. The parameter deviation value corresponding to each time position is compared with the corresponding deviation fluctuation judgment interval. When the parameter deviation value is higher than the corresponding upper bound for multiple consecutive time windows (e.g., more than three consecutive time windows), the corresponding time position is identified as an abnormal time position. Within consecutive time windows, the identified abnormal time positions are organized, and time segments where abnormal time positions exist in multiple consecutive time windows are statistically analyzed and used as abnormal concentration segments. Within abnormal concentration segments, the abnormal time positions and parameter positions are jointly organized in conjunction with the parameter positions corresponding to each time position to generate a first-level abnormal distribution set.

[0050] S3.3 Statistically analyze the continuous distribution of abnormal deviation values ​​in the first-level abnormal distribution set in the time and parameter dimensions to form a set of abnormal local regions.

[0051] It should be noted that the abnormal deviation values, corresponding time positions, and parameter positions for each time window in the first-level anomaly distribution set are read; the abnormal deviation values ​​are arranged in chronological order on the time axis, and the number of time windows in which abnormal deviation values ​​exist simultaneously between adjacent time positions is counted. Time segments in which abnormal deviation values ​​exist in multiple consecutive time windows (such as five consecutive time windows) are identified as time-continuous anomaly segments; the parameter positions corresponding to each abnormal deviation value are counted in the parameter dimension, and the parameter sets in which multiple parameter positions simultaneously exhibit abnormal deviation values ​​within the same time segment are counted. The parameter combinations in which multiple parameter positions simultaneously exhibit abnormal deviation values ​​within the same time segment are identified as parameter-continuous anomaly sets; the time-continuous anomaly segments and parameter-continuous anomaly sets are organized accordingly, so that the abnormal deviation values ​​corresponding to multiple parameter positions within the same time segment form a two-dimensional distribution region, generating an anomaly local region set.

[0052] S3.4 By analyzing the changing trend and expansion direction of abnormal deviation values ​​in the set of abnormal local regions over continuous time, structural abnormal nodes that appear in the process of abnormal deviation evolution are identified, and a secondary abnormal distribution set is generated.

[0053] It should be noted that the process involves reading the time and parameter positions corresponding to each abnormal deviation value in the abnormal local region set, and extracting the abnormal deviation value sequence corresponding to consecutive time positions at the same parameter position. The difference between adjacent time positions is calculated in the abnormal deviation value sequence, and these differences are arranged chronologically to form an abnormal deviation change sequence. The number of time positions with values ​​greater than zero and the number of time positions with values ​​less than zero are counted. When the number of time positions with values ​​greater than zero exceeds the number of time positions with values ​​less than zero, the corresponding time segment is determined as an abnormal deviation enhancement segment. When the time interval is defined as the abnormal deviation reduction interval, it is used to characterize the changing trend of abnormal deviation values ​​in the time dimension. The parameter position distribution corresponding to the newly added abnormal deviation values ​​between adjacent time windows is compared. When the newly added parameter position appears outside the original parameter position set, the newly added parameter position is regarded as the expansion position of abnormal deviation in the parameter dimension, and the abnormal expansion direction is recorded in the order of parameter number. In the same abnormal local area, the time position and parameter position where the abnormal deviation change trend and parameter expansion direction change are defined as structural abnormal nodes, and each structural abnormal node is organized in chronological order to generate a secondary abnormal distribution set.

[0054] S4. Analyze the anomaly propagation direction and parameter relationship transmission characteristics between the secondary anomaly distribution set and the primary anomaly distribution set to form an anomaly evolution chain. Extract the anomaly structure pattern of the anomaly evolution chain through structural analysis and perform fault fitting processing to generate a multidimensional fault fitting set.

[0055] S4.1 Associate the structural anomaly nodes in the secondary anomaly distribution set with the corresponding anomaly deviation segments in the primary anomaly distribution set according to the unified time position and parameter position to form a set of anomaly association node pairs.

[0056] It should be noted that, on a unified timeline, the temporal positions of structural anomaly nodes in the secondary anomaly distribution set are compared with the time segments of primary anomaly deviation segments. When the temporal position of a structural anomaly node falls within the time segment of its corresponding deviation segment, a temporal correspondence between the structural anomaly node and the deviation segment is established. In the parameter dimension, the parameter positions of structural anomaly nodes are compared with the parameter segments of their corresponding deviation segments. When the parameter positions of structural anomaly nodes fall within the parameter segments of their corresponding deviation segments, a parameter correspondence between the structural anomaly node and the deviation segment is established. Under the condition that both temporal and parameter correspondences are satisfied, the structural anomaly node is associated with its corresponding deviation segment to form an anomaly-associated node pair. All anomaly-associated node pairs are then integrated to generate a set of anomaly-associated node pairs.

[0057] S4.2 Analyze the distribution relationship of each abnormal associated node in the set in terms of time sequence and the transmission relationship in the direction of parameter change to identify the direction of abnormal propagation.

[0058] It should be noted that the abnormal associated node pairs in the set are arranged in chronological order. The temporal relationship between the structural abnormal node's position and the time range of the abnormal deviation segment is compared between adjacent abnormal associated node pairs. When the structural abnormal node's position in a later abnormal associated node pair is later than the starting position of the abnormal deviation segment's time range in a previous abnormal associated node pair, the corresponding relationship is recorded as a temporal succession relationship. The direction of parameter position change between adjacent abnormal associated node pairs is statistically analyzed. When the parameter position in a later abnormal associated node pair moves relative to the parameter position in a pair along the direction of increasing parameter number, the corresponding relationship is recorded. The relationship is recorded as a forward propagation relationship. When the parameter position in the subsequent abnormal associated node pair moves relative to the parameter position in the previous abnormal associated node pair in the direction of decreasing parameter number, the corresponding relationship is recorded as a reverse propagation relationship. When the parameter position in the subsequent abnormal associated node pair expands simultaneously in the direction of increasing parameter number and the direction of decreasing parameter number, the corresponding relationship is recorded as a bidirectional propagation relationship. The temporal succession relationship is organized with the forward, reverse, or bidirectional propagation relationships in the parameter dimension. When adjacent abnormal associated node pairs simultaneously satisfy the temporal succession relationship and the parameter propagation relationship, the connection direction between the corresponding node pairs is determined as the abnormal propagation direction.

[0059] S4.3. Organize the nodes associated with each anomaly in a chain according to the direction of anomaly propagation to form an anomaly evolution chain.

[0060] It should be noted that all abnormal associated nodes are arranged in a uniform chronological order, and the abnormal propagation direction where the time position of the preceding abnormal associated node is earlier than that of the subsequent abnormal associated node is recorded as a valid connection relationship. In terms of parameters, the parameter positions corresponding to the preceding and subsequent abnormal associated nodes are read, and the valid connections with positive, negative, or bidirectional propagation relationships are organized according to the continuous change order of parameter positions, so that abnormal associated nodes with continuous propagation relationships are connected sequentially to form an abnormal propagation path. Within the same abnormal propagation path, the abnormal associated node with the earliest starting time position and no preceding connection relationship is determined as the starting node, and the abnormal associated node with the latest ending time position and no subsequent connection relationship is determined as the ending node. All abnormal associated nodes connected between the starting node and the ending node in chronological order are regarded as an abnormal evolution chain.

[0061] S4.4. Expand the structure of each abnormal associated node in the abnormal evolution chain in chronological order, and count the propagation path and connection relationship between each abnormal associated node to form an abnormal structure path set.

[0062] It should be noted that within each anomaly evolution chain, the anomaly-related nodes are sequentially expanded according to their order in the propagation sequence, forming a sequential connection between adjacent anomaly-related nodes. For each pair of adjacent anomaly-related nodes, the time and parameter positions of the preceding and following anomaly-related nodes are recorded, and the connection between the preceding and following nodes is recorded as a propagation path. Within the same anomaly evolution chain, all propagation path records are continuously organized in chronological order, forming an anomaly structure path corresponding to the current anomaly evolution chain. The anomaly structure paths corresponding to each anomaly evolution chain are summarized to generate an anomaly structure path set.

[0063] S4.5 Analyze the coordinated change relationship between the analog parameters corresponding to each abnormal associated node in the abnormal structure path set, and generate abnormal structure patterns.

[0064] It should be noted that the time and parameter positions of the corresponding abnormal associated nodes in each propagation path record of the abnormal structure path set are read. Based on the time-dimensional arrangement of each abnormal associated node, the values ​​of the analog parameter sequence at the corresponding time position are extracted, and the parameter position identifiers corresponding to each value are recorded. The difference in values ​​between adjacent time positions of the analog parameter sequence is calculated, and the consistency of the change direction of different analog parameters at the same time position is statistically analyzed. When multiple analog parameters simultaneously show an increasing or decreasing trend at the same time position, the corresponding time position is recorded as a coordinated change time position. The number of coordinated change time positions within the same time interval is counted, and the ratio between this number and the total number of time positions within the corresponding time interval is taken as the degree of coordinated change. The degree of coordinated change of each time interval is arranged in chronological order within the same abnormal structure path, and the propagation structure where the degree of coordinated change maintains a consistent change direction within multiple consecutive time intervals is statistically analyzed. The corresponding analog parameter change direction and parameter position identifiers are recorded to form an abnormal structure pattern.

[0065] S4.6. Based on the abnormal structure pattern, perform combined fitting processing on the abnormal related nodes in the abnormal evolution chain to generate a multidimensional fault fitting set.

[0066] It should be noted that the time position, parameter position, anomaly propagation direction, and anomaly deviation value of each anomaly-related node in the anomaly evolution chain are read, and the anomaly-related nodes are arranged in chronological order. The time position difference and parameter position change direction are compared between adjacent anomaly-related nodes. When the time positions of two adjacent anomaly-related nodes continue to increase and the parameter position change direction is consistent with the anomaly propagation direction, the connection relationship between the corresponding nodes is recorded as a continuous propagation relationship. The continuous propagation relationship between adjacent anomaly-related nodes is checked one by one in chronological order in the anomaly evolution chain. When the continuous propagation relationship remains uninterrupted, the corresponding node sequence is divided into the same propagation segment. When the continuous propagation relationship is interrupted, the interruption position is used as the segment boundary position, forming multiple continuous propagation segments.

[0067] Read the analog parameter change direction and parameter position identifier corresponding to each mode record in the abnormal structure pattern, and extract the parameter position change direction and abnormal deviation value change direction corresponding to the abnormal associated nodes in each continuous propagation segment; compare the parameter position change direction corresponding to each abnormal associated node in the continuous propagation segment with the parameter position identifier in the abnormal structure pattern item by item, and compare the consistency between the abnormal deviation value change direction and the analog parameter change direction in the abnormal structure pattern, and count the number of nodes with consistent parameter positions and the number of nodes with consistent change directions to form a pattern matching record; within the same continuous propagation segment, perform pattern matching records corresponding to different abnormal structure patterns... The model is compared, and the pattern matching record with the most matching nodes is selected as the fault fitting result corresponding to the continuous propagation segment. The parameter position identifier and the combination of analog parameter change direction in the abnormal structure pattern corresponding to the pattern matching record are extracted as the structural feature description of the corresponding fault fitting result. The structural feature description of the fault fitting result corresponding to each continuous propagation segment is subjected to structural classification processing. Fault fitting results with the same parameter position identifier combination and analog parameter change direction are classified into the same fault structure type, and a corresponding fault structure type identifier is assigned to each type of fault structure. The fault fitting results with fault structure type identifiers are organized in chronological order to generate a multidimensional fault fitting set.

[0068] S5. Evaluate the abnormal evolution trend of the multidimensional fault fitting set, generate an operational risk index set, perform an abnormal evolution consistency judgment on the operational risk index set, and generate a fault early warning result set.

[0069] S5.1 Read the fault fitting structures in the multidimensional fault fitting set in chronological order, and count the frequency and duration of each fault fitting structure in the continuous time window to form a fault structure evolution sequence.

[0070] It should be noted that the fault fitting structures in the multidimensional fault fitting set are read. These fault fitting structures are formed by fitting abnormal structural patterns with corresponding abnormal evolution chain segments, and include time position segments, parameter position identifiers, and fault structure type identifiers. All fault fitting structure records are arranged in chronological order, forming a continuous record sequence based on their time positions. Fixed-length continuous time windows (e.g., five consecutive time positions constitute one time window) are constructed using a unified time benchmark. Within each time window, the number of occurrences of fault fitting structures belonging to the same fault structure type identifier is counted and recorded as the frequency of fault structure occurrence in the corresponding time window. The frequency changes of fault structures corresponding to the same fault structure type identifier are compared between adjacent time windows. When multiple adjacent time windows (e.g., three consecutive time windows) contain fault fitting structures corresponding to the same fault structure type identifier, the corresponding time window segment is recorded as the continuous distribution segment of the current fault fitting structure. The frequency of fault structure occurrence and continuous distribution segments corresponding to each time window are continuously organized in chronological order to generate a fault structure evolution sequence.

[0071] S5.2 Analyze the direction of change, expansion range and propagation speed of each fault fitting structure in the fault structure evolution sequence at continuous time positions to obtain the fault structure evolution characteristics.

[0072] It should be noted that when comparing the direction of change in the frequency of fault structure occurrence between adjacent time windows in the fault structure evolution sequence, if the frequency of fault structure occurrence in the later time window is greater than that in the earlier time window, the corresponding change relationship is recorded as the direction of enhancement; if the frequency of fault structure occurrence in the later time window is less than that in the earlier time window, the corresponding change relationship is recorded as the direction of weakening; and if the frequency of fault structure occurrence in adjacent time windows is consistent, the corresponding change relationship is recorded as the direction of stable change.

[0073] The parameter position identifiers corresponding to each time window in the fault structure evolution sequence are read, and the number of parameter positions in each time window that show fault fitting structures within the same time interval is counted. The difference between the maximum and minimum number of parameter positions that show fault fitting structures within the same time interval is used as the parameter expansion amount, representing the expansion range of the fault fitting structure in the parameter dimension. The number of time windows in consecutive time windows that show fault fitting structures within the same time interval is counted, and the ratio between the current number of time windows and the total number of time windows within the corresponding time interval is used as the propagation rate of the fault fitting structure in the time dimension. The change direction of the fault fitting structure in the time dimension, the expansion range in the parameter dimension, and the propagation rate in the time dimension are combined and organized to obtain the fault structure evolution characteristics.

[0074] S5.3. Based on the evolution characteristics of the fault structure, perform risk quantification on each fault fitting structure to generate a set of operational risk indicators.

[0075] It should be noted that, within each time interval, the number of fault-fitted structures belonging to the same fault-fitted structure type is counted, and the ratio between the number of fault-fitted structures and the total number of fault-fitted structures in the current time interval is used as the structure occurrence ratio. Within the current time interval, the corresponding change direction record is read. When the change direction is an enhancing change direction, the product of the structure occurrence ratio and the propagation rate value is used as the risk change reference quantity. When the change direction is a weakening change direction or a stable change direction, only the propagation rate value is recorded as the risk change reference quantity. The parameter expansion quantity is combined with the structure occurrence ratio to obtain the structure expansion characterization quantity. Based on the maximum and minimum values ​​of the structure expansion characterization quantity, propagation rate value, and risk change reference quantity within the current time interval, a structure expansion characterization quantity is constructed. A unified evaluation interval is established, and the structural expansion characterization, propagation rate values, and risk change reference values ​​are mapped according to this unified evaluation interval. Within the same time period, the changes in the structural expansion characterization, propagation rate values, and risk change reference values ​​after interval mapping are statistically analyzed, and their corresponding weights are determined based on the proportion of each change to the total change. A weighted summation is then performed on the structural expansion characterization, propagation rate values, and risk change reference values ​​according to their corresponding weights to obtain a joint risk characterization value. The product of the joint risk characterization value and the proportion of structural occurrence is used as the quantitative value of operational risk for the corresponding time period. The quantitative values ​​of operational risk for each time period are arranged in chronological order to generate a set of operational risk indicators.

[0076] S5.4 Read the operational risk indicators in the operational risk indicator set in chronological order, and statistically analyze the direction and magnitude of change of each operational risk indicator in a continuous time window to form a risk indicator change sequence.

[0077] It should be noted that, by comparing the differences in the values ​​of operational risk indicators at adjacent time points in the operational risk indicator set in chronological order, when the value of the operational risk indicator at a later time point is greater than that at a previous time point, the corresponding change is recorded as an upward risk direction; when the value of the operational risk indicator at a later time point is less than that at a previous time point, the corresponding change is recorded as a downward risk direction; and when the values ​​of the operational risk indicators at adjacent time points are consistent, the corresponding change is recorded as a stable risk direction. Within the same time period, the absolute value of the difference in the values ​​of operational risk indicators between adjacent time points is calculated, and the difference between the maximum and minimum absolute values ​​of all differences is taken as the risk change magnitude. The risk change direction and risk change magnitude corresponding to each time period are continuously organized in chronological order to form a risk indicator change sequence.

[0078] S5.5 Identify persistent risk segments in the risk indicator change sequence, and perform early warning judgment processing based on persistent risk segments to generate a fault early warning result set.

[0079] It should be noted that the risk change direction record and risk change magnitude value corresponding to each time position in the risk indicator change sequence are read, and all time positions are arranged in chronological order. Within a time segment consisting of multiple consecutive time positions (e.g., five consecutive time positions), the distribution of risk change direction records is statistically analyzed. If the risk change direction corresponding to most time positions in the current time segment (e.g., more than half of the total number of time positions in the time segment) is an upward risk direction, the corresponding time segment is identified as a risk-rising segment. Within the risk-rising segment, the risk change magnitude value is further statistically analyzed. When the risk change magnitude gradually increases within multiple consecutive time positions (e.g., three consecutive time positions), the corresponding time segment is identified as a continuous risk segment. Within the identified continuous risk segment, the operational risk indicator value at the corresponding time position is read. When the operational risk indicator maintains a continuous upward change within the continuous risk segment, the corresponding continuous risk segment is recorded as a fault warning segment, and the corresponding time position segment and parameter position identifier are extracted to form a fault warning result set.

[0080] It should also be noted that the "early warning judgment processing" refers to the process of determining whether a risk segment meets the characteristics of fault evolution based on the identified continuous risk segment and the changing trend of the operational risk indicators at the corresponding time position. Specifically, within a continuous risk segment, when the operational risk indicators maintain an upward trend at multiple consecutive time positions, the corresponding continuous risk segment is determined as a fault early warning segment; otherwise, it is not considered a fault early warning segment.

[0081] In summary, this invention achieves explicit modeling of the "normal operation relationship pattern" of equipment through counterfactual state deduction and the generation of a counterfactual reference set. It also establishes a comparable structural reference within the current detection cycle, transforming anomaly identification from single-parameter deviation judgment to structural deviation judgment at the parameter relationship level. This enhances the ability to distinguish normal fluctuations from true anomalies in complex operating conditions and fluctuating environments. Furthermore, by constructing anomaly evolution chains, extracting anomaly structural patterns, and performing fault fitting processing, it achieves a systematic characterization of the propagation path and evolution law of anomalies in both time and parameter dimensions. This transforms anomaly identification from "discrete identification" to "structured evolution analysis," enabling continuous assessment of fault development trends and supporting risk quantification and early warning judgment, thereby improving the foresight and reliability of fault early warning.

[0082] 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, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for early warning and detection of equipment faults based on digital radar holographic scanning, characterized in that, include: Collect multi-source operation monitoring data during equipment operation, perform time-series organization and processing, and generate parameter collaborative scanning sequences; The analog parameters of the parameter co-scanning sequence are extracted and the correlation of parameter changes is analyzed to form a parameter relationship characterization set. The historical stable operation segments in the parameter relationship characterization set are statistically analyzed and counterfactual state inferences are performed to generate a counterfactual reference set. The parameter relationship deviation of the collaborative scanning sequence relative to the counterfactual reference set is calculated, and a first-level anomaly distribution set is output. Through anomaly nesting analysis, structural anomalies within the first-level anomaly distribution set are identified, and a second-level anomaly distribution set is generated. The abnormal propagation direction and parameter relationship transmission characteristics between the secondary abnormal distribution set and the primary abnormal distribution set are analyzed to form an abnormal evolution chain. The abnormal structure pattern of the abnormal evolution chain is extracted through structural analysis, and fault fitting processing is performed to generate a multidimensional fault fitting set. The abnormal evolution trend of the multidimensional fault fitting set is evaluated, an operational risk index set is generated, an abnormal evolution consistency judgment is performed on the operational risk index set, and a fault early warning result set is generated.

2. The equipment fault early warning and detection method based on digital radar holographic scanning as described in claim 1, characterized in that, The specific steps for generating the parameter-coordinated scanning sequence are as follows: Multi-source operational monitoring data are resampled according to a unified time reference to generate a synchronous observation data sequence. Data consistency governance is performed on outliers and missing values ​​in the synchronous observation data sequence, and sliding time slicing is performed according to continuous time windows to generate a parameter co-scanning sequence.

3. The equipment fault early warning and detection method based on digital radar holographic scanning as described in claim 2, characterized in that, The specific steps for forming the parameter relationship representation set are as follows: Read the values ​​of temperature, pressure, vibration, current and flow parameters at continuous time positions corresponding to each time window in the parameter collaborative scanning sequence to form an analog parameter sequence; Analyze the synchronous change relationship, response delay relationship, and consistency of change direction among the analog parameters in the analog parameter sequence to form a parameter relationship characterization set.

4. The equipment fault early warning and detection method based on digital radar holographic scanning as described in claim 3, characterized in that, The specific steps for generating the counterfactual reference set are as follows: Read the parameter relationship representation set formed within the historical operating cycle, and extract the correlation between the analog parameters corresponding to each time window to form a historical parameter relationship sequence; Perform relation trajectory reconstruction processing on the historical parameter relationship sequence, and connect them in chronological order to form a set of historical relation structure trajectories, using the correlation between each analog parameter as nodes. Identify running segments with similar trajectory structures in the historical relationship structure trajectory set, and perform counterfactual state inference by combining the analog quantity parameter sequence in the current detection cycle to generate a counterfactual reference set.

5. The equipment fault early warning and detection method based on digital radar holographic scanning as described in claim 2 or 4, characterized in that, The specific steps for outputting the first-level anomaly distribution set are as follows: The parametric relationship deviation between the parameter cooperative scan sequence and the counterfactual reference set is calculated to form a parametric relationship deviation sequence; Based on the parameter relationship deviation sequence, the abnormal time locations of parameter relationship deviation values ​​within each time window are identified and organized according to the time window to generate a first-level abnormal distribution set.

6. The equipment fault early warning and detection method based on digital radar holographic scanning as described in claim 5, characterized in that, The specific steps for generating the secondary anomaly distribution set are as follows: The continuous distribution of abnormal deviations in the first-level anomaly distribution set is statistically analyzed in terms of time and parameter dimensions to form a set of abnormal local regions. By analyzing the changing trends and expansion directions of abnormal deviation values ​​in the set of abnormal local regions over continuous time, structural abnormal nodes that appear during the evolution of abnormal deviations are identified, and a secondary abnormal distribution set is generated.

7. The equipment fault early warning and detection method based on digital radar holographic scanning as described in claim 6, characterized in that, The specific steps for forming the abnormal evolution chain are as follows: The structural anomaly nodes in the secondary anomaly distribution set are associated with the corresponding anomaly deviation segments in the primary anomaly distribution set according to the unified time position and parameter position, forming a set of anomaly associated node pairs; Analyze the distribution relationship of each abnormal associated node in the set in terms of time sequence and the transmission relationship in the direction of parameter change to identify the direction of abnormal propagation; The abnormal related nodes are organized in a chain according to the direction of abnormal propagation to form an abnormal evolution chain.

8. The equipment fault early warning and detection method based on digital radar holographic scanning as described in claim 7, characterized in that, The specific steps for generating the multidimensional fault fitting set are as follows: The abnormal evolution chain is structurally expanded according to time sequence, and the propagation path and connection relationship between each abnormal associated node are counted to form an abnormal structure path set. Analyze the coordinated change relationship between the analog parameters corresponding to each abnormal associated node in the abnormal structure path set to generate abnormal structure patterns; Based on the abnormal structure pattern, the abnormal associated nodes in the abnormal evolution chain are combined and fitted to generate a multidimensional fault fitting set.

9. The equipment fault early warning and detection method based on digital radar holographic scanning as described in claim 1, characterized in that, The specific steps for generating the set of operational risk indicators are as follows: The fault fitting structures in the multidimensional fault fitting set are read in chronological order, and the frequency and persistence distribution of each fault fitting structure in a continuous time window are statistically analyzed to form a fault structure evolution sequence. The evolution characteristics of fault structures are obtained by analyzing the direction of change, expansion range and propagation speed of each fault fitting structure at continuous time position in the fault structure evolution sequence. Based on the evolution characteristics of the fault structure, the risk of each fault fitting structure is quantified to generate a set of operational risk indicators.

10. The equipment fault early warning and detection method based on digital radar holographic scanning as described in claim 1, characterized in that, The specific steps for generating the fault warning result set are as follows: Read the operational risk indicators in the operational risk indicator set in chronological order, and statistically analyze the direction and magnitude of change of each operational risk indicator in a continuous time window to form a risk indicator change sequence. Identify persistent risk segments in the risk indicator change sequence, and perform early warning judgment processing based on persistent risk segments to generate a fault early warning result set.

Images