An intelligent automobile fault diagnosis system

Abstract

The present application relates to the technical field of fault diagnosis, in particular to an automobile fault intelligent diagnosis system, the system comprises: a collection sequence verification module, an intensity anomaly grouping module, a trend path reconstruction module, a frequency priority empowerment module and a reverse estimation judgment module.In the present application, the time label and the sampling point number sequence verification signal are checked in sequence and the intensity mutation condition, the collection time sequence disorder phenomenon is discriminated and clustered, the grouping identification of the jump abnormal signal is completed combined with the connection path sequence of the signal node in the wiring structure diagram, the signal evolution path is screened with the help of the trend consistency in the time sequence and is mapped with the running state, the triggering frequency of each path in different periods is searched and the increasing trend is determined, the signal source point is located reversely through the path space distribution, the accurate identification ability of the potential fault source point can be improved, and the problem that the diagnosis accuracy is easily decreased under the condition of fault node space-time dislocation is solved.

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis technology, and in particular to an intelligent fault diagnosis system for automobiles. Background Technology

[0002] The field of fault diagnosis technology involves the identification, location, and diagnosis of abnormal states that occur in various equipment and systems during operation. Core aspects include the collection, identification, and judgment of fault information; the detection and classification of faults based on equipment operating parameters or state variables; and the analysis of fault development trends. This technology is widely used in various industries such as manufacturing, power, and transportation. Especially in the transportation sector, fault diagnosis technology, as a crucial means to ensure equipment safety, improve system reliability, and enhance maintenance efficiency, is gradually shifting from manual experience-based judgment to intelligent analysis. It encompasses key aspects such as sensor data acquisition, fault mode recognition, feature extraction, and system diagnostic rule construction, forming a complete process for fault detection and location based on electronic digital data processing systems. Traditional automotive intelligent fault diagnosis systems refer to systems that detect fault symptoms in key components such as automotive electronic control systems, engines, and transmissions during use. These systems collect vehicle operating data by deploying onboard sensors and use threshold comparison methods based on logical judgments or empirical rules to determine abnormal states and preliminarily locate fault locations. Traditional automotive intelligent fault diagnosis systems use a controller area network centered on the engine control unit to receive operating parameter information transmitted by onboard sensors. They then compare the vehicle status item by item using a set fault code and its corresponding relationship table to identify the specific fault type in the system.

[0003] Existing technologies rely on preset thresholds and fixed logic rules for fault diagnosis, lacking the identification and analysis of correlation paths between multiple signals. Especially in scenarios where there are errors in sensor sampling timing or delays in data transmission, it is easy to cause misalignment between fault information and the actual source point, affecting the accuracy of the location results. Traditional methods fail to combine operational trends, node distribution, and frequency changes for comprehensive judgment, resulting in a delayed response to phenomena such as path jumps and signal evolution in complex scenarios. This can easily lead to the omission of source points in some cross wiring or nonlinear conduction structures. Especially in areas with complex system structures or multiple overlapping nodes, misjudgments or location deviations are likely to occur, affecting maintenance decisions and subsequent troubleshooting efficiency. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing an intelligent diagnostic system for automotive faults.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: an intelligent vehicle fault diagnosis system includes:

[0006] The acquisition sequence verification module obtains the fault code timestamps and sampling point number sequences uploaded by the sensor nodes in the vehicle. Based on the timetamp order, it determines whether there are signal items with misaligned order and sudden intensity changes among the sampling points. If the conditions are met, they are included in the misaligned signal set, generating an acquisition sequence abnormal signal set.

[0007] The intensity anomaly grouping module, based on the collection sequence anomaly signal set, calls the automotive wiring structure diagram, extracts the connection path order of the signal nodes in the structure diagram, performs a connection continuity judgment with the previous signal path, and generates a path jump association fault group.

[0008] The trend path reconstruction module, based on the time series of signals in the path jump associated fault group, filters trend change sequences with consistent directions within continuous segments, classifies them into the same signal evolution path number, maps and binds them with the vehicle's operating status, and generates a trend linkage path identification set.

[0009] The frequency priority weighting module calls the path number in the trend linkage path identification set, retrieves the triggering situation in the vehicle operation status record segment, and statistically analyzes the repetitive triggering frequency of the signal corresponding to each path number according to the operation stage, generates a critical path priority response table, and performs frequency and criticality quantification and sorting of vehicle fault diagnosis.

[0010] As a further embodiment of the present invention, the collection sequence abnormal signal set includes misalignment signal item number, time tag difference, and intensity mutation index; the path jump associated fault group includes jump node number, jump segment signal value, and difference in path number before and after; the trend linkage path identification set includes trend path number, direction consistent change sequence, and operation status mapping information; and the critical path priority response table includes path number, operation stage, and increasing frequency identifier.

[0011] As a further aspect of the present invention, the acquisition sequence verification module includes:

[0012] The fault code sorting submodule obtains the fault code timestamps and sampling point number sequences uploaded by the sensor nodes in the vehicle, extracts the sampling point numbers corresponding to each set of timestamps, sorts them in ascending order according to the timestamps, calculates the index difference between adjacent numbers in the original sequence and the sorted sequence, and generates a sampling order offset value sequence.

[0013] The misalignment judgment submodule calls the sampling order offset value sequence, extracts continuous offset segments in the same direction based on the change in offset direction of adjacent sampling points, calculates the misalignment signal change rate value by analyzing the intensity change amplitude, and combines and compares the misalignment signal change rate value with the degree of change of segment number to generate a misalignment signal segment index group.

[0014] The abnormal signal extraction submodule extracts the sampling number and time tag corresponding to the segment based on the segment information located by the misaligned signal segment index group, completes segment reconstruction and sampling order mark update, and obtains the collection order abnormal signal set.

[0015] As a further aspect of the present invention, the intensity anomaly grouping module includes:

[0016] The connection path extraction submodule, based on the abnormal signal set of the acquisition sequence, calls the signal node connection information in the automotive wiring structure diagram, locates the connection relationship of each pair of adjacent abnormal signals in the structure diagram, extracts the real-time path sequence number, arranges the node index according to the order of signal acquisition, and generates path index sequence value.

[0017] The path jump judgment submodule calls the path index sequence value, compares the numbering order of any two consecutive signals according to the signal acquisition order, and if the two nodes do not form a continuous path in the wiring structure diagram, it is determined to be a structural jump connection. The number of occurrences of structural jumps is counted, and the frequency value of jump signal pairs is generated.

[0018] The jump fault marking submodule calls the frequency value of the jump signal pair, extracts the signal acquisition location in sequence according to the real-time signal number, calculates the path jump intensity distribution value, extracts the signal point pair combination with prominent numerical proportion according to the path jump intensity distribution value, and generates the path jump associated fault group.

[0019] As a further aspect of the present invention, the trend path reconstruction module includes:

[0020] The jump signal filtering submodule detects the continuity of time periods within the time series based on the time series of signals in the path jump associated fault group, calculates the value increase of adjacent time nodes, determines whether there is a trend change in the direction of change, filters signal segments with the same continuous increase direction, and generates a set of jump signal segments with consistent trend.

[0021] The direction sequence classification submodule calls the trend-consistent jump signal segment set, analyzes the change direction markings of the signal segments, classifies them according to the trend direction, identifies the correspondence between the classification number and the signal path position, and obtains the trend classification signal path sequence.

[0022] The state path binding submodule extracts time period and vehicle operation status data based on the trend classification signal path sequence, constructs a mapping combination of path sequence and status data, calculates the trend linkage strength value of each path status combination, and filters out path status combinations with prominent linkage to obtain a trend linkage path identification set.

[0023] As a further aspect of the present invention, the frequency priority weighting module includes:

[0024] The path trigger statistics submodule calls the path number in the trend linkage path identification set, retrieves the signal timestamp according to the path number index in the running status record segment, and groups the timestamps according to their respective running stages. It then counts the number of signal triggers for each group of path numbers in the corresponding stage and generates a stage signal trigger frequency value.

[0025] The phase trend identification submodule extracts the frequency change sequence of the path number corresponding to the differentiated phase based on the phase signal trigger frequency value, arranges the path number by path number, determines whether there is any path number whose frequency value increases in the continuous operation phase, filters the set of path numbers that meet the trend requirements, and generates a path frequency increasing trend sequence.

[0026] The critical path extraction submodule calls the set of path numbers in the path frequency increasing trend sequence, extracts the operation stage labels corresponding to the path numbers, and organizes the mapping structure between path numbers and stage labels in the order of stage numbers to generate a critical path priority response table.

[0027] As a further aspect of the present invention, the system also includes a reverse positioning judgment module:

[0028] The reverse positioning judgment module uses the priority path number in the critical path priority response table to extract the spatial distribution order of the signal nodes corresponding to the path in the structure diagram, performs reverse tracking on the nodes that have position jumps, and if there is no continuous connection path between the node and the previous signal node, the previous signal node is marked as the source point, and the fault source point prediction result is generated.

[0029] The fault source prediction results include the potential source node number, corresponding spatial location, and path disconnection marker.

[0030] As a further aspect of the present invention, the reverse positioning judgment module includes:

[0031] The spatial order extraction submodule calls the priority path number in the critical path priority response table, retrieves the signal node index associated with the path number, obtains the corresponding spatial coordinate position in the structure diagram according to the node label, constructs a spatial arrangement array of signal nodes according to the path number order, and generates a path signal spatial order list.

[0032] The connectivity investigation submodule extracts the connection relationship between two adjacent signal nodes in the structure graph based on the path signal spatial order list, performs connection path judgment, records the set of node pairs that have not established connections, marks them as path interruption nodes, and generates a signal jump node index group.

[0033] The source point tracing and identification submodule calls the signal jump node index group to perform reverse path indexing on the previous signal node in the structure graph, and determines whether the previous node is a path terminal or an unconnected node. If the determination result is a path breakpoint, it is recorded and a fault source point prediction result is generated.

[0034] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0035] In this invention, the sequence of signals and intensity abrupt changes are verified by using time tags and sampling point number sequences to identify and cluster disordered acquisition timing phenomena. By combining the connection path sequence of signal nodes in the wiring structure diagram, the jump-type abnormal signals are grouped and identified. The consistency of trend changes in the time series is used to screen the signal evolution path and establish a mapping with the operating state. The trigger frequency of each path in different time periods is searched in stages and the increasing trend is determined. The signal source point is located in reverse through the spatial distribution of the path. A multi-level linkage judgment mechanism is constructed between key features such as sampling anomalies, path jumps, signal trend evolution, frequency distribution and spatial disconnection. This can improve the accurate identification capability of potential fault sources and effectively make up for the problem that the diagnostic accuracy is easy to decrease under the condition of fault nodes being misaligned across time and space. Attached Figure Description

[0036] Figure 1 This is a system flowchart of the present invention;

[0037] Figure 2 This is a flowchart of the data acquisition sequence verification module in this invention;

[0038] Figure 3 This is a flowchart of the intensity anomaly grouping module in this invention;

[0039] Figure 4 This is a flowchart of the trend path reconstruction module in this invention;

[0040] Figure 5 This is a flowchart of the frequency priority weighting module in this invention;

[0041] Figure 6 This is a flowchart of the reverse positioning judgment module in this invention. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0043] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0044] Please see Figure 1 An intelligent vehicle fault diagnosis system includes:

[0045] The acquisition sequence verification module obtains the fault code timestamps and sampling point number sequences uploaded by the sensor nodes in the vehicle. Based on the timetamp order, it determines whether there are signal items with misaligned order and sudden intensity changes among the sampling points. If the conditions are met, they are included in the misaligned signal set, generating an acquisition sequence abnormal signal set.

[0046] The intensity anomaly grouping module is based on the collection sequence anomaly signal set, calls the automotive wiring structure diagram, extracts the connection path sequence of the signal nodes in the structure diagram, and judges the connection continuity with the previous signal path. If there is a node jump, the real-time signal and the previous signal are combined into a jump mark group to generate a path jump associated fault group.

[0047] The trend path reconstruction module, based on the time series of signals in the path jump associated fault group, filters trend change sequences with consistent directions within continuous segments, classifies them into the same signal evolution path number, maps and binds them with the vehicle's operating status, and generates a trend linkage path identification set.

[0048] The frequency priority weighting module calls the path number in the trend linkage path identification set, searches for triggering situations in the vehicle operation status record segment, and statistically analyzes the repeated triggering frequency of the signal corresponding to each path number according to the operation stage. It analyzes whether there are signal groups where the number of triggers of the path number shows an increasing trend in continuous operation stages, binds the path number to the operation stage as the stage key judgment path, and generates a key path priority response table.

[0049] The reverse positioning judgment module uses the priority path number in the critical path priority response table to extract the spatial distribution order of the signal nodes corresponding to the path in the structure diagram, performs reverse tracing on the nodes that have position jumps, and if there is no continuous connection path between the node and the previous signal node, the previous signal node is marked as the source point, and the fault source point prediction result is generated.

[0050] The collection sequence abnormal signal set includes misaligned signal item number, time tag difference, and intensity change index; the path jump associated fault group includes jump node number, jump segment signal value, and difference in path number before and after; the trend linkage path identification set includes trend path number, consistent direction change sequence, and operation status mapping information; the critical path priority response table includes path number, operation stage, and increasing frequency identifier; and the fault source inference results include potential source node number, corresponding spatial location, and path disconnection mark.

[0051] Please see Figure 2 The data acquisition sequence verification module includes:

[0052] The fault code sorting submodule obtains the fault code timestamps and sampling point number sequences uploaded by the sensor nodes in the vehicle, extracts the sampling point numbers corresponding to each set of timestamps, sorts them in ascending order according to the timestamps, calculates the index difference between adjacent numbers in the original sequence and the sorted sequence, and generates a sampling order offset value sequence.

[0053] The process retrieves the fault code timestamps and sampling point number sequences uploaded by the sensor nodes in the vehicle. This involves calling the fault code data frames recorded by the sensor nodes, extracting sub-segments containing timestamp and sampling point sequence fields, parsing the raw UTC format values ​​in the timestamp field and converting them to standard time-series values ​​in seconds. For example, extracting the timestamp "2025-08-14 10:03:01" converts it to a second value of 1713004981 seconds. The sampling point number field is extracted to show the 256th group number, which is recorded using array indexing. Simultaneously, the sampling points are sorted in ascending order based on the timestamp. The sequence is sorted using bubble sort, which compares time stamp items one by one and swaps the corresponding sampling point numbers. For example, if the previous time stamp is 1713004979 seconds and the next time stamp is 1713004975 seconds, they are swapped. This process is repeated to complete the sorting of the entire sequence. After sorting, an index sequence array is created for the original sequence and the sorted sequence for each group of sampling points. The index difference between the two at the same sampling point number is calculated and recorded as the offset value. For example, if the index of the 5th sampling point is 5 in the original sequence and 3 in the sorted sequence, then the sampling order offset value of that point is 2, and a sampling order offset value sequence is generated.

[0054] The misalignment detection submodule calls the sampling order offset value sequence, extracts continuous offset segments in the same direction based on the change in offset direction between adjacent sampling points, and analyzes the intensity change amplitude using the formula:

[0055] ;

[0056] Calculate the rate of change of the misaligned signal, and compare the rate of change of the misaligned signal with the degree of change of the segment number to generate a misaligned signal segment index group.

[0057] in, Indicates the first The rate of change of the segment misalignment signal. Indicates the first Section 1 Signal strength at each sampling point Indicates the first Section 1 Signal strength at each sampling point Indicates the first Section 1 Time stamp of each sampling point Indicates the first Section 1 Time stamp of each sampling point Indicates the first The difference between the start and end time tags of the segment Indicates the number of sampling points;

[0058] Formula calculation logic: Extracting continuous signals from signal segments For each sampling point, the difference in signal strength between every two adjacent sampling points is calculated and summed to obtain the total signal change. The sum of the squares of the differences in time stamps between every two sampling points, and the sum of the squares of the differences in time stamps between the first and last sampling points of the segment, are then added together and the square root is taken to form a time change normalization factor. The total signal change is divided by this normalization factor to obtain the signal change rate per unit time. This value can quantify the rate of change of intensity over time in a sampled signal, and is suitable for determining whether a signal has a shift trend.

[0059] The rate of change of misaligned signals represents the average rate of change of signal intensity over time in a sampling segment. It is calculated by normalizing the sum of the differences between adjacent signal intensities and the corresponding time change amplitude. The larger the rate value, the more drastic the signal change in that segment. It is used to identify abnormal sampling segments or sampling sequence misalignment segments.

[0060] Extract consecutive offset segments in the same direction based on the change in offset direction between adjacent sampling points. Then, by traversing adjacent items in the sampling order offset sequence in a double-layer process, let the i-th... The item is offset value 5, the first If the item has an offset value of 7, then the current offset trend is recorded as positive. If the subsequent items also show a positive trend, then they will be grouped into one segment. For example, if the offset value sequence is... The first five items are divided into a group of segments. The amplitude of the change in the sampling signal intensity is analyzed in each segment, and the rate of change of the signal intensity in the segment is calculated. ;

[0061] Meaning of parameters and calculation process:

[0062] For the first Section 1 The signal strength value of each sampling point is normalized by the voltage value obtained by the ADC acquisition channel of the original signal. For example, when the input voltage is 3.3V, the maximum value is 1023. Then, the sampling value is 561, and the corresponding signal strength is 561 / 1023≈0.548.

[0063] The time stamp for this point is collected in seconds. For example, if the sampling time points are 1.001 seconds, 1.002 seconds, and 1.003 seconds, the calculation process is as follows:

[0064] The values ​​assigned to the formula parameters are as follows:

[0065] , , ;

[0066] , , ;

[0067] Calculate the numerator:

[0068] ;

[0069] Calculate the denominator:

[0070] ;

[0071] ;

[0072] Combine denominators and take the square root:

[0073] ;

[0074] Substitute into the formula to calculate:

[0075] ;

[0076] The result indicates that the rate of change of the misalignment signal in this segment is 4.082, which can be used as a benchmark for segment division. The value is compared with the average rate of change sequence recorded in the segment. For example, if the current segment rate value is 4.082 and the average rate of change value of the corresponding segment is 2.315, then the current value is greater than its upper limit deviation range and is judged as an abnormal fluctuation segment. The index position of this segment is generated as the abnormal signal index segment, and the index segment number is the index position of the original segment sequence array.

[0077] The advantage of the formula is that it achieves a quantitative characterization of the signal change rate by combining the signal change amount with the sampling time interval.

[0078] The parameter settings are explained in the following instructions:

[0079] The time stamp is measured in microsecond-level sampling intervals, with the interval generally not exceeding 0.002 seconds;

[0080] The signal strength value is set to the ratio after ADC normalization, and the range is within... between;

[0081] To enhance the accuracy of anomaly detection, a change rate threshold is set as follows: Indicates a segment of violent fluctuation;

[0082] Table 1: Example Table of Fragment Rate Calculation

[0083] ;

[0084] Table 1 lists the specific values ​​of the sampled signal strength and sampling time points in a certain segment for use in rate calculation.

[0085] The abnormal signal extraction submodule extracts the sampling number and time tag corresponding to the segment based on the segment information located by the misaligned signal segment index group, completes segment reconstruction and sampling order mark update, and obtains the collection order abnormal signal set;

[0086] Iterate through each index value in the index group, corresponding to the number array where segmentation has been completed in the original signal sampling record. Locate the start and end sampling point numbers of the segment using the index value. Let the index of the current erroneous segment be 5. In the segment index record array, the 5th segment corresponds to the start sampling number 140 and the end sampling number 187. Then, read the sampling data from the 140th to the 187th item in the original sampling record, extract the time tag field recorded therein, and construct a time window array. The array content is as follows: Then, this array is used as the time location information for the abnormal signal segments. The original signal strength values ​​corresponding to the time period are read sequentially to construct the abnormal segment signal set. For example, the strength array is... By re-establishing the sequence index for this data segment and reconstructing the sampling order table according to chronological order, a continuous time series structure can be obtained after error correction. The abnormal signal sequence is then inserted into the abnormal signal set array, resulting in an array sequence containing multiple abnormal signal units. ,in Indicates the first Each abnormal signal unit It contains three sets of data: a timestamp array, an intensity array, and a sampling number array. The unified structure facilitates segment-level signal diagnosis and analysis in subsequent processing, and allows for the acquisition of signal sets with abnormal acquisition order.

[0087] Please see Figure 3 The intensity anomaly grouping module includes:

[0088] The connection path extraction submodule, based on the abnormal signal set of the acquisition sequence, calls the signal node connection information in the automotive wiring structure diagram, locates the connection relationship of each pair of adjacent abnormal signals in the structure diagram, extracts the real-time path sequence number, arranges the node index according to the order of signal acquisition, and generates path index sequence values.

[0089] The system retrieves the signal node connection information from the vehicle wiring diagram. For each signal point in the abnormal signal set, it sequentially calls its corresponding node number in the diagram. A lookup table is used to read the mapping relationship and establish the conversion from signal number to node index. For example, the abnormal signal number "E145" corresponds to a wiring node index value of 32. After mapping the abnormal signals, a node sequence is formed. Then, the paths are rearranged in ascending order according to the signal acquisition time sequence to ensure that the constructed paths match the actual acquisition time sequence. For example, if the timestamp corresponding to node 32 is 1.010 seconds, node 35 is 1.025 seconds, and node 42 is 1.032 seconds, the sorted path sequence will maintain its original order. The path will then be numbered sequentially according to the time sequence and recorded as a path index sequence. This is used for subsequent jump judgment and fault identification processes after the path sequence is established.

[0090] After completion, each pair of adjacent nodes is searched for connection edges in the structure graph to generate path index sequence values.

[0091] The path jump judgment submodule calls the path index sequence value and compares the numbering order of any two consecutive signals according to the signal acquisition order. If the two nodes do not form a continuous path in the wiring structure diagram, it is determined to be a structural jump connection. The number of occurrences of structural jumps is counted, and the frequency value of jump signal pairs is generated.

[0092] By reading the pre-constructed adjacency matrix in the structure graph, each adjacent node pair is checked for the existence of connecting edges. For example, if the value of the adjacency matrix corresponding to the 32nd row and 35th column of nodes 32 and 35 is 1, then a connection exists. If the value of the adjacency matrix corresponding to the 42nd row and 50 is 0, then this pair is recorded as a skip node pair. The connection status of each node pair formed in the path index is checked, and the unconnected items are summarized and recorded in the skip pair list. The frequency of the jump pair is statistically analyzed based on the number of repetitions in the same path. If a vehicle's jump pair (42, 50) appears 3 times in 5 samplings, its frequency is recorded as 3. A jump frequency table is established for the jump node pairs and frequencies to prepare for the next step of path jump intensity calculation. The structure is in the form of key-value pairs, where the key is the node pair and the value is the frequency, such as {(42, 50):3, (58, 61):1}. Each jump pair is synchronized with the sampling time and wiring position to ensure the consistency of the context node structure and generate the jump signal pair frequency value.

[0093] The skip fault marker submodule calls the skip signal frequency value and extracts the signal acquisition location sequentially according to the real-time signal number, using the following formula:

[0094] ;

[0095] Calculate the path hop intensity distribution value, extract signal point pairs with prominent numerical proportions based on the path hop intensity distribution value, and generate path hop associated fault groups.

[0096] in, This represents the path jump intensity distribution value. Representing the The frequency of jumps in the jump signal. , The first For the signal and the first Numbering the signals, For the first The total number of jumps on the path to the signal point. , The first Regarding the first The path connection index number for the jump signal. This represents the total number of jump signal pairs;

[0097] Formula calculation logic: Establish a numbering sequence for jump node pairs, starting from the 1st jump signal to the Zth jump signal, and perform calculations item by item. The numerator in the calculation is the frequency of the current jump pair. Product of the difference between the signal number and the signal number This reflects whether the jump occurred at a location with a large spatial distance from the signal. The denominator is the difference between the number of jumps in the jump path plus 1 and the jump index position in the path. The summation is used to normalize the adjustment factor and prevent local frequencies from shifting due to short paths. Each pair of skip nodes is substituted into the above structure and summed to obtain the overall path RF value. For example, if the first pair has a frequency of 4, an index difference of 20, a path hop count of 3, and an index difference of 2, then its contribution is... The values ​​are summed to obtain value;

[0098] The path jump intensity distribution value is used to evaluate the concentration and amplitude of jump signals in the path. Combined with jump frequency, number span, and path index difference, it reflects the discontinuous distribution characteristics of signals in the wiring structure. The larger the value, the more jump signals there are in the path and the uneven distribution of them. It is often used as a basis for judging abnormal structural paths.

[0099] The parameter descriptions and calculation logic are as follows:

[0100] Used to measure the magnitude of change in the coded number of a jump signal;

[0101] Used to characterize the relative spacing of jump positions in a path sequence;

[0102] : Represents the intensity of the jump signal;

[0103] Regularize the jump path to prevent the denominator from being zero;

[0104] The overall structure is the sum of the ratios of frequency-weighted differences to path span, reflecting the intensity of path jump distribution;

[0105] Let: jump node pair (12) 17). , , , , , ;

[0106] Jump node pairs (23) 45), , , , , , ;

[0107] Then substitute the values ​​into the calculation as follows:

[0108] Item 1:

[0109] ;

[0110] Item 2:

[0111] ;

[0112] Substitute into the formula to calculate:

[0113] ;

[0114] This result indicates that there is a significant distribution of skip nodes in the current path. The value is 31.0, significantly higher than the baseline range [5.0]. The normal fluctuation range of 10.0];

[0115] The innovation of the formula lies in the fact that it uses the frequency multiplied by the difference in node number for weighting and combines it with the path span for normalization, thus forming a quantitative assessment method for the degree of jump structure anomaly.

[0116] Table 2: Example Parameter Table for Path Jump Nodes

[0117] ;

[0118] Table 2 lists the detailed parameters of example node pairs participating in the path jump intensity calculation, for use in formula substitution and calculation.

[0119] Please see Figure 4 The trend path reconstruction module includes:

[0120] The skip signal filtering submodule detects the continuity of time periods within the time series based on the time series of signals in the path skip associated fault group, calculates the value increase of adjacent time nodes, determines whether there is a trend change in the direction of change, filters signal segments with the same continuous increase direction, and generates a set of skip signal segments with consistent trends.

[0121] The sampling time of each node in the path of the group is called and rearranged in chronological order to transform the original sampling sequence into a monotonically non-decreasing sequence. It then sequentially checks whether the time increment between two adjacent time points remains continuously positive. For example, if the first time point is 1.020 seconds and the second is 1.024 seconds, it is considered a positive increase. This process continues for subsequent points; if the third is 1.028 seconds and the fourth is 1.030 seconds, it indicates a continuous increase. If a node's time point is less than the previous time point, the growth trend of that segment is considered interrupted. The current signal segment is then divided into independent segment sequences, generating an array of trend-consistent segments. The direction of signal change within each trend segment is determined based on the signal strength value. Let the signal strength change value of the first segment be... If the change value is determined to be a negative trend, then... If the trend is positive, only segments with consistent direction are retained as trend-consistent jump segment groups to generate a trend-consistent jump signal segment set.

[0122] The direction sequence classification submodule calls the trend-consistent jump signal segment set, analyzes the change direction markings of the signal segments, classifies them according to the trend direction, identifies the correspondence between the classification number and the signal path position, and obtains the trend-classified signal path sequence.

[0123] The signal change direction of each segment in the set is called, and direction difference classification processing is performed. A change vector is constructed for each signal segment. Let the change vector of the i-th segment be... The j-th segment is The direction consistency of the two signals is calculated based on the cosine angle. If the cosine value is greater than 0.9, they are considered to have the same direction and are classified into the same category. If the angle is 180°, meaning the directions are completely opposite, they are classified into opposite categories and not merged. After assigning classification labels to the classification signal segments, the path jump position information is called. For each jump path segment, the classification signal segment is matched with its path index to identify the position of the jump segment. For example, if a segment numbered "SEG23" belongs to path "P7", and this signal segment is classified as "Trend Group A", it is mapped to "P7-SEG23-A". This completes the trend classification labeling within the path and outputs the trend classification path segment sequence, with the following structure: .

[0124] The state path binding submodule extracts time period and vehicle operating status data based on trend classification signal path sequences, and constructs a mapping combination between path sequences and status data using the following formula:

[0125] ;

[0126] Calculate the trend linkage strength value for each path state combination, filter out the path state combinations with prominent linkage, and obtain the trend linkage path identification set.

[0127] in, Representing the The trend linkage strength value of the combination of path states. Indicates the first Path number The amplitude of signal value change in each segment Indicates the first Path number The duration of each segment Indicates the first Path number The magnitude of change in the state data of each segment Indicates the number of segments in the path;

[0128] Formula calculation logic: The numerator is the signal change value of each segment. With duration The absolute value of the product reflects the intensity fluctuation capability of each segment per unit time; the denominator represents the amplitude of the state data change. The sum of absolute values ​​represents the total change in vehicle state parameters in the path segment, used to normalize the total numerator intensity and form an index of the degree of response of the trend signal to the state variables.

[0129] The trend linkage strength value reflects the comprehensive influence of signal trends on changes in vehicle status data. It is used to measure the tightness of trend linkage in path status combinations. By multiplying the amplitude and duration of signal changes, the total fluctuation of state variables is normalized to construct a relative response index. The larger the value, the stronger the linkage between the signal trend and the status data of the path segment. It is an important reference for identifying key trend segments.

[0130] In practice, each path contains multiple segments, and each segment is calculated item by item. Suppose path b has 3 segments, and the data is as follows:

[0131] Segment 1: , , ;

[0132] Segment 2: , , ;

[0133] Segment 3: , , ;

[0134] but:

[0135] ;

[0136] ;

[0137] Substitute into the formula to calculate:

[0138] ;

[0139] The result shows that the trend linkage strength value in path b is 2.56, indicating that the trend change has a strong response relationship compared with the state change.

[0140] Table 3: Examples of Path Trend Status Changes

[0141] ;

[0142] Table 3 lists the participating factors and intermediate results for each of the three segments of path b in the trend linkage strength calculation, which facilitates verification. The construction logic and accuracy.

[0143] Please see Figure 5 The frequency priority assignment module includes:

[0144] The path trigger statistics submodule calls the path number in the trend linkage path identification set, retrieves the signal timestamp by indexing the path number in the running status record segment, and groups the timestamps according to their respective running stages. It then counts the number of signal triggers for each path number in the corresponding stage and generates a stage signal trigger frequency value.

[0145] The system iteratively calls the index mapping relationship of each path number within the vehicle operation status record segment to extract the signal number set associated with each path number. Then, based on the timestamp sequence generated during the operation of each group of signal numbers, it establishes a relationship structure between path numbers and signal timestamps. It obtains the operation stage number corresponding to each time point in the timestamp set. The operation stage numbers are uniformly divided according to the start and end time periods of the record file. For example, stage 1 is set to seconds 0 to 100, stage 2 to seconds 101 to 200, and so on. Signal trigger times are then assigned to stages accordingly, completing the grouping of trigger events for each path number within each operation stage. The array counter is used to perform frequency statistics on each group of data to obtain the number of signals triggered for each path number in each stage. A two-dimensional table structure is constructed with the path number as the row and the running stage as the column. The frequency statistics results are filled in item by item. For example, the trigger frequency of path number P01 in stage 1, stage 2 and stage 3 is 8, 12 and 15, and the trigger frequency of path number P02 is 5, 5 and 7. The above values ​​are filled in the corresponding row of the frequency table in sequence. If there is no corresponding signal record in a certain stage, zero value is filled in as the default item. After the statistics are completed, the frequency values ​​will be used for subsequent inter-stage trend judgment processing to obtain the stage signal trigger frequency value.

[0146] The phase trend identification submodule extracts the frequency change sequence of path numbers corresponding to the differentiated phases based on the phase signal trigger frequency value and arranges them by path number. It then determines whether any path number shows an increasing frequency value in the continuous operation phase, filters the set of path numbers that meet the trend requirements, and generates a path frequency increasing trend sequence.

[0147] The frequency values ​​of each stage of each path are extracted using the path number as the primary index, forming a frequency sequence array. For example, if the path number is P05, and its frequency values ​​in stages 1, 2, 3, and 4 are 6, 9, 11, and 13 respectively, then its frequency sequence is: The frequency difference between two consecutive stages is calculated through a difference operation. A first-order difference operation is performed on the sequence, and the result is... The process involves determining whether each difference is positive. If all frequency differences are greater than 0, the path number is recorded as part of the increasing trend path set. If any frequency difference is negative or zero, the path number is removed, completing one trend filtering process. For the path numbers participating in the filtering, consistency of frequency sequence length is checked. If any path frequency sequence is missing intermediate stage data, the path is marked as an abnormal path and removed from the calculation range to ensure the integrity of subsequent trend judgment data. Among the remaining paths, a set of path numbers that meet the increasing trend condition is counted, and a corresponding trend identification table is established, recording the path number and its corresponding frequency sequence together. For example, the frequency sequence corresponding to path P07 is... The frequency corresponding to path P09 is Then, P07 and P09 are used as trend path markers to obtain a trend sequence with increasing path frequency.

[0148] The critical path extraction submodule calls the path number set in the increasing trend sequence of path frequency, extracts the running stage label corresponding to the path number, and organizes the mapping structure between path number and stage label according to the stage number order to generate the critical path priority response table.

[0149] The system retrieves the vehicle operation phase annotation records and performs reverse matching on the phase numbers where the frequency values ​​appear for each path number. This confirms the active phase range of the path number during operation. The operation phase array for each path number is then sorted in ascending order by phase number. A one-to-one mapping table structure is then established between path numbers and phase numbers, with the path number as the index and the phase number as the value. For example, if the trend frequency of path P08 corresponds to phases 3, 4, and 5, then this path number and its phase number form a mapping record. After the path numbers and phase numbers are organized, the mapping structure is flattened and converted into a list of path-phase key-value pairs. Then, the key-value pairs are sorted by stage number to ensure that the path number order remains unchanged in the output while the stage order remains consistent. After the construction is completed, the mapping structure is written into the critical path matrix as input, and the field order is standardized. The output standard format structure is a two-dimensional table with path number column and stage number column, resulting in the critical path priority response table.

[0150] Please see Figure 6 The reverse positioning judgment module includes:

[0151] The spatial order extraction submodule calls the priority path number in the critical path priority response table, retrieves the signal node index associated with the path number, obtains the corresponding spatial coordinate position in the structure diagram according to the node number, constructs a spatial arrangement array of signal nodes according to the path number order, and generates a path signal spatial order list.

[0152] Based on the path number index, extract all signal node numbers within the corresponding path, map the node numbers to the node index table in the structure diagram, and retrieve the three-dimensional spatial coordinate data of each node number in the structure diagram using the node index table. The coordinate data is in the format... This represents the location information of nodes within the vehicle layout. All acquired node coordinates are reorganized in path number order to construct a spatial node sequence array. The node coordinate sequence corresponding to each path number is stored in a two-dimensional array structure, forming an index list. For example, path P03 corresponds to signal node numbers N1, N2, N5, and N9, and their spatial coordinates are as follows: Then the spatial order array of path P03 is Then, the array is labeled with path number to form a key-value binding structure between path number and spatial sequence. At the same time, the overall path sequence array is arranged in ascending order of path number to ensure that the path structure remains consistent in the logical chain. The sorted structure is written into the path spatial sequence mapping set for subsequent connectivity verification operations to generate a path signal spatial sequence list.

[0153] The connectivity investigation submodule extracts the connection relationship between two adjacent signal nodes in the structure graph based on the spatial order list of path signals, performs connection path judgment, records the set of node pairs that have not established connections, marks them as path interruption nodes, and generates a signal jump node index group.

[0154] For each path, the adjacent node pairing operation is performed on the spatial coordinate sequence of nodes. The current node and the next node are extracted in sequence to form a set of node pairs. For each node pair, the path connectivity query is performed on the connection edge data in the structure graph. The query method is to search in the adjacency matrix of the structure graph to see if there is a connection value marked as 1 between the node pairs. If the query result is 0, it is judged as an unconnected node pair and recorded as a path interruption record. If the query result is 1, it is skipped and not recorded. After each pairing query of a set of path numbers is completed, a corresponding list of unconnected node pairs is generated. This list is organized and classified according to the path number and uniformly marked with the interruption status to construct a set of disconnected node pairs. For example, in path P04, node pair (N2) N5), (N5) If there are no connected terms in N9), then construct a pair of disconnected nodes as follows: The set is converted into an index number format and recorded as a node index jump array. The node numbers in the array are summarized according to the path number to ensure that the broken segments in the path can be clearly located at a specific position in the node sequence for reverse tracing identification, and a signal jump node index group is generated.

[0155] The source point tracing and identification submodule calls the signal jump node index group, performs reverse path indexing on the previous signal node in the structure graph, and determines whether the previous node is a path terminal or an unconnected node. If the determination result is a path breakpoint, it is recorded and a fault source point prediction result is generated.

[0156] For each jump node pair, extract the path connection information within its structure graph from the preceding signal node. Read the list of extended path numbers of the node in the connection direction of the structure graph and determine whether it is the terminal point of a certain structural path. The judgment criterion is whether the number of connections in the corresponding row in the adjacency matrix is ​​1. If it is 1, mark the node as a terminal node. If the number of connections is 0 or there is an edge but it cannot be connected to the subsequent jump node, mark the node as a breakpoint node. Index and record the node numbers that meet the above conditions. Deduplicate the breakpoint nodes that appear repeatedly under the same path number to ensure that the source point under a single path is not output repeatedly. After completing the breakpoint analysis of the jump node pair, organize the path number and its corresponding breakpoint node number, record the potential path source position, and generate the fault source prediction result.

[0157] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. An intelligent diagnostic system for automotive faults, characterized in that, The system includes: The acquisition sequence verification module obtains the fault code timestamps and sampling point number sequences uploaded by the sensor nodes in the vehicle. Based on the timetamp order, it determines whether there are signal items with misaligned order and sudden intensity changes among the sampling points. If the conditions are met, they are included in the misaligned signal set, generating an acquisition sequence abnormal signal set. The intensity anomaly grouping module, based on the collection sequence anomaly signal set, calls the automotive wiring structure diagram, extracts the connection path order of the signal nodes in the structure diagram, performs a connection continuity judgment with the previous signal path, and generates a path jump association fault group. The trend path reconstruction module, based on the time series of signals in the path jump associated fault group, filters trend change sequences with consistent directions within continuous segments, classifies them into the same signal evolution path number, maps and binds them with the vehicle's operating status, and generates a trend linkage path identification set. The frequency priority weighting module calls the path number in the trend linkage path identification set, retrieves the triggering situation in the vehicle operation status record segment, and statistically analyzes the repetitive triggering frequency of the signal corresponding to each path number according to the operation stage, generates a critical path priority response table, and performs frequency and criticality quantification and sorting of vehicle fault diagnosis. The reverse positioning judgment module uses the priority path number in the critical path priority response table to extract the spatial distribution order of the signal nodes corresponding to the path in the structure diagram, performs reverse tracking on the nodes that have position jumps, and if there is no continuous connection path between the node and the previous signal node, the previous signal node is marked as the source point, and the fault source point prediction result is generated. The fault source prediction results include the potential source node number, corresponding spatial location, and path disconnection marker.

2. The intelligent vehicle fault diagnosis system according to claim 1, characterized in that, The collection sequence abnormal signal set includes misaligned signal item number, time tag difference, and intensity change index; the path jump associated fault group includes jump node number, jump segment signal value, and difference in path number before and after; the trend linkage path identification set includes trend path number, consistent direction change sequence, and operation status mapping information; and the critical path priority response table includes path number, operation stage, and increasing frequency identifier.

3. The intelligent vehicle fault diagnosis system according to claim 1, characterized in that, The data acquisition sequence verification module includes: The fault code sorting submodule obtains the fault code timestamps and sampling point number sequences uploaded by the sensor nodes in the vehicle, extracts the sampling point numbers corresponding to each set of timestamps, sorts them in ascending order according to the timestamps, calculates the index difference between adjacent numbers in the original sequence and the sorted sequence, and generates a sampling order offset value sequence. The misalignment judgment submodule calls the sampling order offset value sequence, extracts continuous offset segments in the same direction based on the change in offset direction of adjacent sampling points, calculates the misalignment signal change rate value by analyzing the intensity change amplitude, and combines and compares the misalignment signal change rate value with the degree of change of segment number to generate a misalignment signal segment index group. The abnormal signal extraction submodule extracts the sampling number and time tag corresponding to the segment based on the segment information located by the misaligned signal segment index group, completes segment reconstruction and sampling order mark update, and obtains the collection order abnormal signal set.

4. The intelligent vehicle fault diagnosis system according to claim 3, characterized in that, The intensity anomaly grouping module includes: The connection path extraction submodule, based on the abnormal signal set of the acquisition sequence, calls the signal node connection information in the automotive wiring structure diagram, locates the connection relationship of each pair of adjacent abnormal signals in the structure diagram, extracts the real-time path sequence number, arranges the node index according to the order of signal acquisition, and generates path index sequence value. The path jump judgment submodule calls the path index sequence value, compares the numbering order of any two consecutive signals according to the signal acquisition order, and if the two nodes do not form a continuous path in the wiring structure diagram, it is determined to be a structural jump connection. The number of occurrences of structural jumps is counted, and the frequency value of jump signal pairs is generated. The jump fault marking submodule calls the frequency value of the jump signal pair, extracts the signal acquisition location in sequence according to the real-time signal number, calculates the path jump intensity distribution value, extracts the signal point pair combination with prominent numerical proportion according to the path jump intensity distribution value, and generates the path jump associated fault group.

5. The intelligent vehicle fault diagnosis system according to claim 4, characterized in that, The trend path reconstruction module includes: The jump signal filtering submodule detects the continuity of time periods within the time series based on the time series of signals in the path jump associated fault group, calculates the value increase of adjacent time nodes, determines whether there is a trend change in the direction of change, filters signal segments with the same continuous increase direction, and generates a set of jump signal segments with consistent trend. The direction sequence classification submodule calls the trend-consistent jump signal segment set, analyzes the change direction markings of the signal segments, classifies them according to the trend direction, identifies the correspondence between the classification number and the signal path position, and obtains the trend classification signal path sequence. The state path binding submodule extracts time period and vehicle operation status data based on the trend classification signal path sequence, constructs a mapping combination of path sequence and status data, calculates the trend linkage strength value of each path status combination, and filters out path status combinations with prominent linkage to obtain a trend linkage path identification set.

6. The intelligent vehicle fault diagnosis system according to claim 5, characterized in that, The frequency priority weighting module includes: The path trigger statistics submodule calls the path number in the trend linkage path identification set, retrieves the signal timestamp according to the path number index in the running status record segment, and groups the timestamps according to their respective running stages. It then counts the number of signal triggers for each group of path numbers in the corresponding stage and generates a stage signal trigger frequency value. The phase trend identification submodule extracts the frequency change sequence of the path number corresponding to the differentiated phase based on the phase signal trigger frequency value, arranges the path number by path number, determines whether there is any path number whose frequency value increases in the continuous operation phase, filters the set of path numbers that meet the trend requirements, and generates a path frequency increasing trend sequence. The critical path extraction submodule calls the set of path numbers in the path frequency increasing trend sequence, extracts the operation stage labels corresponding to the path numbers, and organizes the mapping structure between path numbers and stage labels in the order of stage numbers to generate a critical path priority response table.

7. The intelligent vehicle fault diagnosis system according to claim 1, characterized in that, The reverse positioning judgment module includes: The spatial order extraction submodule calls the priority path number in the critical path priority response table, retrieves the signal node index associated with the path number, obtains the corresponding spatial coordinate position in the structure diagram according to the node label, constructs a spatial arrangement array of signal nodes according to the path number order, and generates a path signal spatial order list. The connectivity investigation submodule extracts the connection relationship between two adjacent signal nodes in the structure graph based on the path signal spatial order list, performs connection path judgment, records the set of node pairs that have not established connections, marks them as path interruption nodes, and generates a signal jump node index group. The source point tracing and identification submodule calls the signal jump node index group to perform reverse path indexing on the previous signal node in the structure graph, and determines whether the previous node is a path terminal or an unconnected node. If the determination result is a path breakpoint, it is recorded and a fault source point prediction result is generated.

Images