Method and apparatus for detecting anomalous shaking of vehicle, device, vehicle, medium, and program product

Abstract

The present disclosure relates to a method and apparatus for detecting anomalous shaking of a vehicle, a device, a vehicle, a medium, and a program product, the method comprising: acquiring a target time sequence corresponding to an original time sequence signal, the original time sequence signal being a time sequence signal corresponding to a vehicle status signal, and the target time sequence being a stable time sequence obtained from preprocessing the original time sequence signal; randomly selecting a plurality of sample points from the target time sequence, constructing a plurality of isolation trees, selecting from the target time sequence a data point to be detected, traversing isolation trees among the plurality of isolation trees according to said data point, and, on the basis of the depths of said data point in the isolation trees, determining anomaly scores of said data point; on the basis of the anomaly scores of said point, determining an anomalous point list composed of anomalous points that have been detected; and, according to a time interval between the anomalous points in the anomalous point list, identifying an anomalous shaking interval. According to the present method, the problem of missed detection during anomaly detection on vehicle status information can be reduced.

Description

Vehicle abnormal vibration detection method, device, equipment, vehicle, medium and program product

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This disclosure claims priority to Chinese patent application No. 202311662801.X filed on December 6, 2023, with applicant Beijing Rockwell Technology Co., Ltd., and application name “Vehicle Abnormal Vibration Detection Method, Device, Electronic Device, Medium and Vehicle”, the full text of which is incorporated by reference into this disclosure. Technical Field

[0003] The present disclosure relates to the field of vehicle detection technology, and in particular to a method, device, equipment, vehicle, medium, and program product for detecting abnormal vehicle vibration. Background Art

[0004] When developing a new vehicle control strategy, road testing is an essential verification step. By analyzing key signals including vehicle vibration status, the implementation effect of the vehicle control strategy can be evaluated.

[0005] Continuous abnormal vibration during vehicle operation is a very dangerous condition. Continuous abnormal vibration related to torque or speed not only affects drivability but can also endanger the driver's safety. Therefore, accurately detecting continuous abnormal vehicle vibration is extremely important. However, the complex and diverse operating conditions faced by vehicles make vehicle status signals highly random. For example, for vehicle status signals such as battery bus actual power, motor torque, and motor speed, their statistical indicators such as mean and variance are functions of time. Therefore, vehicle status signals are typical non-stationary time series signals, and anomaly detection in non-stationary time series signals is very difficult.

[0006] In related technologies, when using prediction error-based anomaly detection methods for non-stationary signals, a prediction model is used to predict outliers. However, the sample data used during the training process does not cover all possible abnormal operating conditions. Therefore, the prediction model trained on this sample data only detects anomalies under a subset of operating conditions, leading to missed detections. Therefore, how to reduce missed detections of vehicle vibration in vehicle status information is a pressing technical issue. Summary of the Invention

[0007] In order to solve the above technical problems, the present disclosure provides a method, device, equipment, vehicle, medium and program product for detecting abnormal vehicle vibration.

[0008] The present disclosure provides a method for detecting abnormal vehicle vibration, comprising:

[0009] Obtaining a target time series corresponding to an original time series signal; the original time series signal is a time series signal corresponding to a vehicle state signal, and the target time series is a stationary time series obtained by preprocessing the original time series signal;

[0010] Randomly selecting multiple sample points from the target time series, constructing multiple isolation trees, and selecting a data point to be detected from the target time series, traversing an isolation tree from the multiple isolation trees according to the data point to be detected, and determining an anomaly score of the data point to be detected based on the depth of the data point to be detected in each of the isolation trees;

[0011] Determining an outlier list consisting of detected outliers based on the outlier scores of the data points to be detected;

[0012] The abnormal jitter interval is identified according to the time interval of each abnormal point in the abnormal point list.

[0013] The present disclosure also provides a device for detecting abnormal vehicle vibration, including:

[0014] an acquisition module configured to acquire a target time series corresponding to an original time series signal; the original time series signal is a time series signal corresponding to a vehicle state signal, and the target time series is a stationary time series obtained by preprocessing the original time series signal;

[0015] a detection module configured to randomly select multiple sample points from the target time series, construct multiple isolation trees, select a data point to be detected from the target time series, traverse an isolation tree from the multiple isolation trees according to the data point to be detected, and obtain an anomaly score for the data point to be detected based on the depth of the data point to be detected in each of the isolation trees;

[0016] a determination module configured to determine an outlier list consisting of detected outliers based on the outlier scores of the data points to be detected;

[0017] The identification module is configured to identify the abnormal jitter interval according to the time interval of each abnormal point in the abnormal point list.

[0018] An embodiment of the present disclosure further provides an electronic device, including: one or more processors;

[0019] a storage device configured to store one or more programs,

[0020] When the one or more programs are executed by the one or more processors, the one or more processors implement any of the above-described methods for detecting abnormal vehicle vibration.

[0021] An embodiment of the present disclosure also provides a vehicle, comprising the electronic device as described above.

[0022] The embodiment of the present disclosure further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for detecting abnormal vehicle vibration as described above is implemented.

[0023] An embodiment of the present disclosure further provides a computer program, which includes computer-readable code. When the computer-readable code runs in an electronic device, the processor of the electronic device executes the method for detecting abnormal vehicle vibration as described above.

[0024] An embodiment of the present disclosure also provides a computer program product, which includes a computer-readable code, or a non-volatile computer-readable storage medium carrying the computer-readable code. When the computer-readable code runs in a processor of an electronic device, the processor in the electronic device implements any of the above-described methods for detecting abnormal vehicle vibration.

[0025] The technical solution provided by the embodiments of the present disclosure has the following advantages over the prior art:

[0026] A target time series corresponding to an original time series signal is obtained, where the original time series signal is a vehicle status signal, and the target time series is a stationary time series obtained after preprocessing the original time series signal. Multiple sample points are randomly selected from the target time series, and multiple isolation trees are constructed. A data point to be detected is selected from the target time series, and the multiple isolation trees are traversed based on the data point to be detected. Based on the depth of the data point to be detected in each isolation tree, an anomaly score is determined for the data point to be detected. Based on the anomaly score of the data point to be detected, an anomaly list consisting of detected anomalies is determined, and abnormal jitter intervals are identified based on the time intervals of each anomaly point in the anomaly list. For a non-stationary time series signal of a vehicle, a stationary target time series is obtained after preprocessing. Multiple sample points are randomly selected from the target time series, and multiple isolation trees are constructed to form an isolation forest. Since each isolation tree is independently generated from a randomly selected sample point in the target time series, it can cover more abnormal data. Furthermore, the isolation forest has excellent large data processing capabilities, enabling parallel processing of multiple data points to be detected, quickly identifying multiple anomalies, and thereby reducing missed detections during abnormal vehicle jitter detection in vehicle status information. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0028] In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0029] FIG1 is a flow chart of a method for detecting abnormal vehicle vibration according to an embodiment of the present disclosure;

[0030] FIG2 is a flow chart of another method for detecting abnormal vehicle vibration provided by an embodiment of the present disclosure;

[0031] FIG3 is a schematic structural diagram of a vehicle abnormal vibration detection device provided by an embodiment of the present disclosure;

[0032] FIG4 is a schematic structural diagram of an electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0033] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described below. It should be noted that the embodiments of the present disclosure and the features therein can be combined with each other in the absence of conflict.

[0034] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.

[0035] Relational terms such as “first” and “second” in the description and claims of this disclosure are merely used to distinguish one entity or operation from another entity or operation, but do not necessarily require or imply any actual relationship or order between these entities or operations.

[0036] In the embodiments of the present disclosure, words such as "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present disclosure should not be interpreted as being more preferred or advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner. In addition, in the description of the embodiments of the present disclosure, unless otherwise specified, the meaning of "multiple" refers to two or more.

[0037] Glossary:

[0038] 1) Fast Fourier Transform (FFT): FFT is a general term for efficient and fast calculation methods for discrete Fourier transform using computers.

[0039] 2) Isolation Tree (iTree) Algorithm: The basic principle of the Isolation Forest algorithm is to detect outliers by constructing multiple isolation trees (iTrees). Each iTree recursively splits the dataset into two subsets by randomly selecting features and split values ​​until a predetermined stopping condition is reached (such as a tree height limit or only one data point remaining in the subset). Ultimately, the depth of each tree reflects the degree of anomaly of the data point: data points with shallower depth are more likely to be outliers.

[0040] 3) Binary Search Tree (BST): Also known as a binary search tree, binary sorted tree, or binary search tree, the principle is as follows: Assume a tree is either empty or a binary tree with the following properties: if its left subtree is not empty, then the values ​​of all nodes in the left subtree are less than the value of its root node; if its right subtree is not empty, then the values ​​of all nodes in the right subtree are greater than the value of its root node; and its left and right subtrees are also binary sorted trees. As a classic data structure, the binary search tree combines the fast insertion and deletion operations of a linked list with the fast search advantages of an array.

[0041] 4) Harmonic series: A harmonic series is a divergent series, which is the sum of the elements of a harmonic sequence. When n approaches infinity, its partial sum has no limit (or the partial sum is infinite).

[0042] For non-stationary signals in vehicle status signals, relevant technologies need to narrow the range of normal data points in non-stationary signals when using anomaly detection methods based on statistical information, which leads to false detection problems.

[0043] At the same time, related technologies also provide methods for detecting anomalies in vehicle status signals through predictive models. However, the selection of predictive models is very critical, and improper model selection can also lead to false detection and missed detection problems. For example, for the detection of outliers in non-stationary signals, related technologies also use algorithms such as Local Outlier Factor (LOF) and K-MEANS clustering to detect non-stationary signals; outliers are obviously abnormal data points in the observed data set, or in other words, the data distribution of outliers is different from the overall distribution of the data set. The purpose of outlier detection is to detect data points that are significantly different from normal data, and then perform further processing based on specific problems. When using the above algorithms for outlier detection, it is necessary to calculate the density of each point in the data set, which leads to high computational complexity and low detection efficiency.

[0044] Furthermore, the sample data used during the prediction model training process doesn't cover all possible abnormal operating conditions. Therefore, the prediction model trained on this sample data only detects anomalies under a subset of operating conditions, leading to missed detections. Therefore, reducing missed detections of vehicle jitter in vehicle status information is a pressing technical issue.

[0045] In response to the above problems, the present invention discloses a method for detecting non-stationary time series signals of vehicles. First, the isolation forest algorithm is used to identify all abnormal points corresponding to the original time series signal, and multiple abnormal jitter intervals are generated, thereby reducing the problem of missed detection in the process of detecting abnormalities in vehicle status information. The advantage of the isolation forest algorithm is that it does not require the calculation of point-to-point distance or density to determine abnormal data, thereby greatly reducing the computational complexity of the algorithm and improving the detection efficiency of abnormal points. At the same time, since each forest tree in the isolation forest is generated independently of each other, the isolation forest supports parallel training on a distributed system and can process vehicle-side big data well. The fast Fourier transform algorithm is further used to filter the false detection intervals in the multiple abnormal jitter intervals to obtain the final target abnormal jitter interval, further reducing the false detection problem in the process of detecting abnormalities in vehicle status information, thereby being able to accurately detect continuous abnormal jitter in the vehicle status signal, providing data support for R&D personnel to promptly discover potential problems in the control strategy, and also providing data support for the stability of the vehicle control strategy.

[0046] [Corrected 10.04.2025 according to Rule 91] As shown in FIG1 , the present disclosure provides a method for detecting abnormal vehicle vibration, which may include the following steps S11-S13:

[0047] S11. Obtain the target time series corresponding to the original time series signal.

[0048] Among them, the original time series signal is the time series signal corresponding to the vehicle status signal; the target time series is the stationary time series obtained after preprocessing the original time series signal.

[0049] In some embodiments, the pre-processing includes low-pass filtering and differencing.

[0050] In some embodiments, the original time series signal may include non-stationary time series signals such as battery bus actual power, motor torque, and motor speed.

[0051] In some embodiments, as shown in FIG2 , obtaining the target time series corresponding to the original time series signal can be achieved by the following steps:

[0052] S111. Obtain original time series signals.

[0053] Specifically, the original time series signal can be periodically collected in real time by the vehicle end, for example, with a collection period of 10 milliseconds.

[0054] S112: Perform low-pass filtering on the original time series signal to obtain a first time series.

[0055] Specifically, since high-frequency noise can be observed in the time series signals such as the actual power of the battery bus, motor torque, and motor speed collected at the vehicle end, a low-pass filter is set to preprocess the signal.

[0056] In some embodiments, performing low-pass filtering on the original time series signal to obtain the first time series can be achieved by the following steps:

[0057] a. Perform global spectrum analysis on the original time series signal to determine the cutoff frequency of the low-pass filter.

[0058] Specifically, perform global spectrum analysis on the original time series signal to determine the cutoff frequency f of the low-pass filter c .

[0059] For example, taking the actual power signal of the vehicle-side battery bus as an example, the time series signal of the actual power of the battery bus can be expressed by formula (1):

[0060] {P t ,t∈T}={P1,P2,P3,…,P t ,…} (1)

[0061] Among them, {P t} represents a time series with a length of t, reflecting the change of the actual power of the battery bus over time.

[0062] Since the analysis process targets a time series signal of finite length, the actual power time series of the battery bus {P N}:

[0063] {P N}={P1,P2,P3,…,P N} (2)

[0064] Using FFT to obtain time series signal {P N} global spectrum, determine the cutoff frequency f of the low-pass filter c For the time series of the actual power of the vehicle battery bus, the cutoff frequency f c Can be set to 30 Hz.

[0065] b. Design a low-pass filter based on the cutoff frequency, remove burrs and redundant peaks in the original time series signal through the low-pass filter, and obtain the first time series.

[0066] Optionally, a low-pass filter is designed based on the cutoff frequency, and burrs and redundant peaks in the original time series signal are removed by the low-pass filter to obtain a first time series, including:

[0067] The sampling rate is designed to be f according to the cutoff frequency s , a low-pass filter of order n; the original time series signal is calculated according to the difference equation of the low-pass filter to obtain the first time series; wherein the difference equation is shown in formula (3):

[0068] Among them, Y(k) represents the sequence value at the kth moment, P(k) is the sequence value at the kth moment in the original time series signal; n is the order of the filter, b0~b n is the input signal coefficient, a0~a n is the output signal coefficient, b0~b n and a0~a n The order n and cutoff frequency f of the filter can be c Sure.

[0069] Specifically, a low-pass filter is used to filter the time series signal {P N} Perform low-pass filtering and design the sampling rate to be f according to the cutoff frequency determined in step a. s , a low-pass filter of order n is used to remove the time series signal {P N The burrs and redundant peaks in the} can be calculated by using the difference equation of the low-pass filter. That is, the first time series {Y N}, the calculation process of the sequence value Y(k) at the kth moment is expanded as shown in formula (3).

[0070] For a clearer description, the following embodiments of the present disclosure uniformly use the set shown in formula (4) to represent the first time series after low-pass filtering:

[0071] {Y N}={Y1,Y2,Y3,…,Y N} (4)

[0072] S113: Perform differential processing on the first time series to obtain a target time series.

[0073] Specifically, the difference method can be used to calculate the first time series {Y N} perform differential processing and remove the first time series {Y N} in the trend component.

[0074] In some embodiments, performing differential processing on the first time series to obtain the target time series can be achieved by:

[0075] Performing first-order difference processing on the first time series to obtain a first-order difference series;

[0076] or;

[0077] Perform second-order difference processing on the first time series to obtain a second-order difference series.

[0078] The target time series includes a first-order difference series or a second-order difference series.

[0079] Specifically, the order of the difference processing needs to be determined according to the characteristics of the signal. In practical applications, feature selection is performed based on the effect of anomaly detection, and the first-order difference or second-order difference is selected as the feature of the isolation forest algorithm.

[0080] The calculation formula of the first-order difference sequence and sequence elements is shown in formula (5):

[0081] The calculation formula of the second-order difference sequence and sequence elements is shown in formula (6):

[0082] For example, for the time series of the actual power of the vehicle battery bus, what needs to be paid attention to is the rate of change of the instantaneous rate of change of power, that is, the second-order difference of the actual power of the battery bus. That is, when there is a sudden change in the rate of change of the instantaneous rate of change of power, it can be considered that the vehicle has experienced continuous abnormal shaking; and for the time series of the actual speed of the vehicle motor, it is necessary to pay attention to the instantaneous rate of change of the motor speed, that is, the angular acceleration. By analyzing whether there is a sudden change in the first-order difference sequence of the actual speed of the motor, if there is a sudden change, it can be considered that the vehicle has experienced abnormal shaking. In the subsequent embodiments of this disclosure, the set shown in formula (7) is uniformly used to represent the target time series after detrending:

[0083] {D R}={D1,D2,D3,…,D R} (7)

[0084] In addition, it should be noted that since high-order difference sequences will lose too much information, when using the difference method to construct anomaly detection features, it is not necessary to perform differences of order two or above.

[0085] S12. Randomly select multiple sample points from the target time series, construct multiple isolation trees, and select data points to be detected from the target time series. Traverse the isolation trees in the multiple isolation trees according to the data points to be detected, determine the anomaly score of the data points to be detected based on the depth of the data points to be detected in each isolation tree, and determine an anomaly point list consisting of detected anomaly points based on the anomaly scores of the data points to be detected.

[0086] Among them, the threshold cutting point is used to determine whether the sample point is an outlier.

[0087] In some embodiments, a plurality of sample points are randomly selected from a target time series, a plurality of isolation trees are constructed, and a data point to be detected is selected from the target time series. An isolation tree in the plurality of isolation trees is traversed according to the data point to be detected. An abnormality score of the data point to be detected is determined based on the depth of the data point to be detected in each isolation tree. An abnormality point list consisting of detected abnormal points is determined based on the abnormality score of the data point to be detected. This can be achieved by the following method:

[0088] a1. Randomly select multiple sample points from the target time series to generate a subsample set. Randomly determine the threshold cut point among the sample points between the maximum and minimum values ​​in the subsample set. Based on the threshold cut point, the multiple sample points in the subsample set are divided into a left branch and a right branch. Repeat the above steps in the left branch and the right branch until the termination condition is met, and construct multiple isolation trees.

[0089] Among them, the target time series is the difference series after difference processing.

[0090] For example, an isolation tree can be constructed as follows:

[0091] Multiple sample points are randomly selected from the target time series to generate a subsample set. A threshold cut point is randomly determined among the sample points between the maximum and minimum values ​​in the subsample set. Based on the threshold cut point, the multiple sample points in the subsample set are divided into a left branch and a right branch. The above steps are repeated in the left branch and the right branch until the termination condition is met and an isolation tree is constructed.

[0092] Exemplarily, the above steps can be implemented through the following process:

[0093] ①, from the difference sequence {D R} randomly select ψ sample points to generate the subsample set {D ψ} and put it into the root node of the current isolated tree.

[0094] ②, in the subsample set {D ψ} randomly generates a cutting point a(a∈[D min D max ]), where D min and D max Represent the minimum and maximum values ​​in the subsequence respectively.

[0095] ③. Divide the current node data space into two subspaces according to the cutting point a, place the data points less than a on the left branch of the current node, and place the data points greater than or equal to a on the right branch of the current node, completing one branch of the isolated tree.

[0096] ④. Recursively repeat steps ② and ③ on the left and right branches of the child node, continuously constructing new leaf nodes until the termination conditions are met. The termination conditions include: Condition 1, the data itself cannot be divided any further, and the leaf node only contains one sample; Condition 2, the depth of the iTree tree reaches log2ψ.

[0097] ⑤. Repeat steps ①-④ to obtain an isolated forest iFores consisting of q isolated trees iTree.

[0098] b1. Select the data point to be detected from the target time series, traverse each isolated tree according to the data point to be detected, and calculate the depth of the data point to be detected in each isolated tree.

[0099] Specifically, a data point to be detected is taken from the target time series, at least one isolated tree is traversed according to the data point to be detected, and the depth of the data point to be detected in each isolated tree is calculated.

[0100] c1. Calculate the abnormality score of the data point to be detected based on the depth of each isolated tree.

[0101] For example, from the differential sequence {D R}, take a data point x to be detected, traverse each iTree tree, calculate the depth h(x) of the data point x to be detected in each tree, and calculate the abnormal score of the data point x to be detected. The calculation formula is shown in formula (8):

[0102] Where E(h(x)) is the average depth of q isolated trees iTree; c(ψ) is the average path length of constructing a BST for ψ samples, which can be calculated using formula (9):

[0103] Here, H(i) represents the "harmonic number", which is the sum of the first i terms of the harmonic series. Harmonic number H(i) = ln(i) + ξ, where ξ represents Euler's constant, which is 0.5772156649.

[0104] For example, based on the anomaly scores of the data points to be detected, determining an anomaly point list consisting of detected anomaly points can be achieved by the following steps:

[0105] d. Based on the anomaly scores of the data points to be detected, determine the anomalies in the target time series.

[0106] Specifically, the exponential portion of the anomaly score calculation formula has a range of (-∞, 0), so s(x, ψ) has a range of (0, 1). The smaller the path length, the closer s(x, ψ) is to 1, indicating a greater probability that the sample point is an outlier. In summary, if the average path length of a target data point in the target time series across multiple isolated trees is shorter, the anomaly score is closer to 1, indicating a greater likelihood that the target data point is an outlier. If the average path length of a target data point across multiple isolated trees is longer, the anomaly score is closer to 0, indicating a lower likelihood that the target data point is an outlier. If the average path length of a target data point across multiple isolated trees is close to the overall mean, the anomaly score is around 0.5. In other words, when the anomaly score is close to 1, the path length is very small, indicating that the target data point is easily isolated and can be classified as an outlier. When the anomaly score is less than 0.5, the path length becomes longer, indicating a higher likelihood that the target data point is a normal point. If the anomaly scores of all data points are around 0.5, it means that there are no outliers in the entire target time series.

[0107] For example, the closer the anomaly score is to 1, the greater the probability that the data point to be detected is an outlier; when the anomaly score is evenly distributed around 0.5, the probability that the data point to be detected is a normal value is higher. For example, when the anomaly score of the data point to be detected is 0.95, it means that the current data point to be detected is easily isolated and is determined to be an outlier; when the anomaly score of the data point to be detected is 0.45, it means that the current data point to be detected is likely to be a normal point; when the anomaly score of the data point to be detected is 0.55, it means that the current data point to be detected is likely to be a normal point. It can be understood that, assuming there are 10 data points to be detected, eight of them have anomaly scores evenly distributed around 0.5, such as 0.46, 0.47, 0.52, 0.51, 0.48, 0.45, 0.53, and 0.49, and only two of them have anomaly scores of 0.97 and 0.99. At this time, these two anomaly scores close to 1 are determined to be anomalies.

[0108] e. Arrange the outliers in the target time series in ascending order according to the index of the target time series to obtain a list of outliers.

[0109] Specifically, all the identified abnormal points are sorted into the original sequence {D R} in ascending order, and finally get a list of all abnormal points detected by the isolation forest algorithm {X M}.

[0110] The advantage of the Isolation Forest algorithm is that it does not require calculating point-to-point distances or densities to identify abnormal data. This greatly reduces the algorithm's computational complexity and improves the efficiency of outlier detection. Furthermore, since each tree in the Isolation Forest is generated independently, the Isolation Forest supports parallel training on distributed systems and can effectively process vehicle-side big data.

[0111] S13. Identify abnormal jitter intervals according to the time intervals of the abnormal points in the abnormal point list.

[0112] In some embodiments, identifying an abnormal jitter interval based on the time intervals of each abnormal point in the abnormal point list can be achieved by the following steps:

[0113] (1) The first point in the abnormal point list is used as the starting point of the target abnormal jitter interval, and the starting point is added to the first list.

[0114] For example, the outlier list {X M} is used as the starting point of the first abnormal jitter interval Ω1, and the starting point is added to the first list.

[0115] (2) Starting from the second point in the abnormal point list, calculate the time interval between the current abnormal point and the previous abnormal point. If the time interval is greater than the preset time interval, determine the previous abnormal point as the end point of the target abnormal jitter interval, and determine the current abnormal point as the starting point of the next target abnormal jitter interval. Add the end point to the second list, repeat the judgment logic, and obtain multiple abnormal jitter intervals.

[0116] For example, from the list of outliers {X M}Starting from the second abnormal point in the image, calculate the interval between the current abnormal point X(i) and the previous abnormal point X(i-1). If the interval between the two exceeds the threshold θ interval , then determine the current abnormal jitter interval Ω j It has ended and needs to be truncated, so the previous abnormal point X(i-1) is used as the current abnormal jitter interval Ω j End point, add to the end point list {E m}, and at the same time, the current abnormal point X(i) is used as the starting point of the next abnormal jitter interval and added to the starting point list {S m The judgment logic is repeatedly executed to obtain multiple abnormal jitter intervals.

[0117] The vehicle abnormal vibration detection method provided by the present disclosure obtains a stable target time series after preprocessing for the non-stationary time series signal of the vehicle, and constructs multiple isolation trees by randomly selecting multiple sample points from the target time series to obtain an isolation forest. Since each isolation tree is independently generated by randomly selecting sample points from the target time series, it can cover more abnormal data. In addition, the isolation forest has an excellent ability to process big data and can process multiple data points to be detected in parallel, quickly identifying multiple abnormal points, thereby reducing the problem of missed detection in the process of detecting abnormal vehicle status information.

[0118] In related technologies, when using anomaly detection methods based on statistical information, it is necessary to narrow the range of normal data points, which may lead to false detection problems; while for anomaly detection based on prediction errors, the selection of prediction models is extremely critical, and improper model selection may also cause false detection problems.

[0119] Based on the problem of false detection in the current vehicle abnormality detection process, the embodiment of the present disclosure further discloses the following steps to reduce the false detection problem. That is, based on the FFT algorithm, multiple abnormal jitter intervals are filtered to obtain at least one target abnormal jitter interval. The specific implementation method is as follows:

[0120] A. Calculate the effective length of each abnormal jitter interval based on the starting point in the first list and the ending point in the second list; determine whether each abnormal jitter interval in the multiple abnormal jitter intervals is a continuous abnormal jitter interval based on the relationship between the effective length of each abnormal jitter interval and the effective length threshold, and determine the continuous abnormal jitter interval in the multiple abnormal jitter intervals as a valid abnormal jitter interval.

[0121] The effective abnormal jitter interval is used to represent an abnormal jitter interval whose effective length is greater than or equal to an effective length threshold among multiple abnormal jitter intervals.

[0122] For example, according to the starting point list of the abnormal jitter interval obtained in the previous step {S m} and the termination point list {E m}Calculate the abnormal jitter range Ω1~Ω m If the current abnormal jitter interval Ω j The length is less than the effective length threshold, and it is determined that the interval is not a continuous abnormal jitter interval, so it is directly selected from the starting point list {S m} and the termination point list {E m}Remove this pair of starting and ending points to obtain at least one valid abnormal jitter interval.

[0123] B. Obtain at least one differential sequence corresponding to at least one valid abnormal jitter interval.

[0124] In some embodiments, in the solutions provided in the embodiments of the present application, after step B is completed, the following steps may be further performed:

[0125] C. Perform FFT processing on at least one differential sequence to obtain spectrum data corresponding to the at least one differential sequence.

[0126] For example, the current abnormal jitter interval Ω is taken out j The corresponding local differential sequence [D(S(j)), D(S(j)+1), …, D(E(j)-1), D(E(j))] is processed by FFT to obtain the current abnormal jitter interval Ω. j spectral data.

[0127] D. Determine whether each valid abnormal jitter interval is a false detection interval based on the size relationship between the spectrum data corresponding to at least one differential sequence and the threshold spectrum data; determine the abnormal jitter interval in each valid abnormal jitter interval whose spectrum data is less than the threshold spectrum data as a false detection interval, remove the false detection interval, and repeatedly execute the false detection interval judgment logic until all valid abnormal jitter intervals are traversed and at least one target abnormal jitter interval is obtained.

[0128] For example, the search spectrum peak Amax , if the spectrum peak A max Less than the amplitude threshold θ Amplitude , determine the current abnormal jitter interval Ω j The interval belongs to the false detection, and then from the starting point list {S m} and the termination point list {E m}Remove this set of starting points S(j) and ending points E(j); repeat the previous step of calculating the local spectrum and this step of removing the false detection interval until all abnormal jitter intervals are traversed.

[0129] The vehicle abnormal jitter detection method provided by the present disclosure uses FFT to filter the false detection intervals in multiple abnormal jitter intervals to obtain the final target abnormal jitter interval, thereby reducing the false detection problem that occurs during the vehicle status information abnormality detection process, thereby accurately detecting continuous abnormal jitters in the vehicle status signal.

[0130] Furthermore, in some embodiments, after executing the above steps, the following method may also be executed:

[0131] The length of each target abnormal jitter interval is used as the corresponding abnormal jitter duration, and each abnormal jitter duration is output.

[0132] Specifically, according to the filtered starting point list {S w} and the termination point list {E w}, regenerate the abnormal jitter range Ω1~Ω w The system calculates the length of each abnormal jitter interval as the abnormal jitter duration, outputs each abnormal jitter duration, and visualizes the results. This provides data support for R&D personnel to promptly identify potential problems in the control strategy and also provides data support for the stability of the vehicle control strategy.

[0133] In some embodiments, as shown in FIG3 , a vehicle abnormal vibration detection device 300 is provided, comprising:

[0134] The acquisition module 310 is configured to acquire a target time series corresponding to an original time series signal; the original time series signal is a time series signal corresponding to the vehicle state signal, and the target time series is a stationary time series obtained by preprocessing the original time series signal;

[0135] The detection module 320 is configured to randomly select multiple sample points from the target time series, construct multiple isolation trees, select a data point to be detected from the target time series, traverse the multiple isolation trees according to the data point to be detected, and determine the anomaly score of the data point to be detected based on the depth of the data point to be detected in each isolation tree;

[0136] A determination module 330 is configured to determine an outlier list consisting of detected outliers based on the outlier scores of the points to be detected;

[0137] The identification module 340 is configured to identify abnormal jitter intervals according to the time intervals of the abnormal points in the abnormal point list.

[0138] In some embodiments, the acquisition module 310 includes:

[0139] an acquisition unit, configured to acquire an original time series signal;

[0140] a filtering unit configured to perform low-pass filtering on the original time series signal to obtain a first time series;

[0141] a processing unit configured to perform differential processing on the first time series to obtain a target time series;

[0142] In some embodiments, the filtering unit is further configured to perform a global spectrum analysis on the original time series signal to determine the cutoff frequency of the low-pass filter; design a low-pass filter based on the cutoff frequency, and remove glitches and redundant peaks in the original time series signal through the low-pass filter to obtain a first time series.

[0143] In some embodiments, the filtering unit is further configured to design a sampling rate of f according to the cutoff frequency. s , a low-pass filter with an order of n; calculating the original signal sequence according to the difference equation of the low-pass filter to obtain a first time series.

[0144] In some embodiments, the processing unit is further configured to perform first-order difference processing on the first time series to obtain a first-order difference sequence; or; perform second-order difference processing on the first time series to obtain a second-order difference sequence; the target time series includes a first-order difference sequence or a second-order difference sequence.

[0145] In some embodiments, the detection module 320 is further configured to randomly select multiple sample points from the target time series to generate a sub-sample set; randomly determine a threshold cut point among the sample points between the maximum value and the minimum value in the sub-sample set, and based on the threshold cut point, divide the multiple sample points in the sub-sample set into a left branch and a right branch, and repeat the above steps in the left branch and the right branch respectively until the termination condition is met, thereby constructing multiple isolated trees; select a data point to be detected from the target time series, traverse each isolated tree according to the data point to be detected, and calculate the depth of the data point to be detected in each isolated tree; and calculate the abnormality score of the data point to be detected according to the depth in each isolated tree.

[0146] In some embodiments, the determination module 330 is further configured to calculate the abnormality score of the data point to be detected by formula (8).

[0147] In some embodiments, the determination module 330 is further configured to determine the outliers in the target time series based on the outlier scores of the data points to be detected; and to sort the outliers in the target time series in ascending order according to the index of the target time series to obtain an outlier list.

[0148] In some embodiments, the identification module 340 is configured to use the first point in the abnormal point list as the starting point of the target abnormal jitter interval and add the starting point to the first list; starting from the second point in the abnormal point list, calculate the time interval between the current abnormal point and the previous abnormal point. If the time interval is greater than the preset time interval, determine the previous abnormal point as the end point of the target abnormal jitter interval, and determine the current abnormal point as the starting point of the next target abnormal jitter interval, add the end point to the second list, and repeat the judgment logic to obtain multiple abnormal jitter intervals.

[0149] In some embodiments, the vehicle abnormal vibration detection device also includes a filtering module, which is configured to calculate the effective length of each abnormal vibration interval based on the starting point in the first list and the ending point in the second list; judge whether each abnormal vibration interval in multiple abnormal vibration intervals is a continuous abnormal vibration interval based on the size relationship between the effective length of each abnormal vibration interval and the effective length threshold, and determine the continuous abnormal vibration interval in multiple abnormal vibration intervals as an effective abnormal vibration interval; the effective abnormal vibration interval is used to represent the abnormal vibration interval in multiple abnormal vibration intervals whose effective length is greater than or equal to the effective length threshold.

[0150] In some embodiments, the filtering module is further configured to obtain at least one differential sequence corresponding to at least one valid abnormal jitter interval; perform FFT processing on the at least one differential sequence to obtain spectrum data corresponding to the at least one differential sequence; determine whether each valid abnormal jitter interval belongs to a false detection interval based on the size relationship between the spectrum data corresponding to the at least one differential sequence and the threshold spectrum data; determine the abnormal jitter interval in each valid abnormal jitter interval whose spectrum data is less than the threshold spectrum data as a false detection interval, remove the false detection interval, and repeatedly execute the false detection interval judgment logic until all valid abnormal jitter intervals are traversed to obtain at least one target abnormal jitter interval.

[0151] In some embodiments, the vehicle abnormal vibration detection device further includes an output module configured to use the length of each target abnormal vibration interval as the corresponding abnormal vibration duration and output each abnormal vibration duration.

[0152] The vehicle abnormal vibration detection device provided by the present disclosure obtains a target time series corresponding to an original time series signal, wherein the original time series signal is a vehicle status signal, and the target time series is a stationary time series obtained by preprocessing the original time series signal. Multiple sample points are randomly selected from the target time series, multiple isolation trees are constructed, and data points to be detected are selected from the target time series. Each isolation tree is traversed according to the data points to be detected, and an abnormality score of the data points to be detected is obtained based on the depth of the data points to be detected in each isolation tree. Based on the abnormality score of the point to be detected, an abnormal point list consisting of detected abnormal points is determined, and the abnormal vibration interval is identified according to the time interval of each abnormal point in the abnormal point list. For the non-stationary time series signal of the vehicle, a stationary target time series is obtained after preprocessing. By randomly selecting multiple sample points from the target time series and constructing multiple isolation trees, an isolation forest is obtained. Since each isolation tree is independently generated by randomly selecting sample points from the target time series, it can cover more abnormal data. In addition, the isolation forest has a good ability to process big data. It can process multiple data points to be detected in parallel and quickly identify multiple abnormal points, thereby reducing the problem of missed detection in the process of abnormal detection of vehicle status information.

[0153] The specific limitations of the abnormal vehicle vibration detection device can be found in the limitations of the abnormal vehicle vibration detection method described above and will not be repeated here. Each module in the abnormal vehicle vibration detection device described above can be implemented in whole or in part through software, hardware, or a combination thereof. Each of the modules can be embedded in or independent of the processor of the electronic device in hardware form, or can be stored in the processor of the electronic device in software form, so that the processor can call and execute the corresponding operations of each module.

[0154] The present disclosure also provides an electronic device. Figure 4 is a schematic diagram of the structure of the electronic device provided by the present disclosure. As shown in Figure 4, the electronic device provided by the present disclosure includes: a memory 41 and a processor 42. The memory 41 is configured to store a computer program; the processor 42 is configured to execute the steps of any of the above-mentioned methods for detecting abnormal vehicle vibration when the computer program is invoked.

[0155] The electronic device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. When the computer program is executed by the processor, a method for detecting abnormal vehicle vibration is implemented. The display screen of the electronic device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the electronic device can be a touch layer covering the display screen, or a button, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0156] Those skilled in the art will understand that the structure shown in FIG4 is merely a block diagram of a portion of the structure related to the solution of the present disclosure, and does not constitute a limitation on the computer device to which the solution of the present disclosure is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0157] In some embodiments, the abnormal vehicle vibration detection device provided by the present disclosure can be implemented in the form of a computer, and the computer program can be run on the electronic device shown in Figure 4. The memory of the electronic device can store the various program modules that make up the abnormal vehicle vibration detection device of the electronic device, such as the acquisition module 310, detection module 320, determination module 330, and identification module 340 shown in Figure 3. The computer program composed of these various program modules causes the processor to execute the steps of the abnormal vehicle vibration detection method according to various embodiments of the present disclosure as described in this specification.

[0158] The embodiment of the present disclosure also provides a vehicle, which may include the electronic device as described above.

[0159] An embodiment of the present disclosure further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the vehicle abnormal vibration detection method provided by the above method embodiment is implemented.

[0160] An embodiment of the present disclosure also provides a computer program, which includes computer-readable code. When the computer-readable code runs in an electronic device, the processor of the electronic device executes the method for detecting abnormal vehicle vibration as described above.

[0161] An embodiment of the present disclosure also provides a computer program product, which includes a computer-readable code, or a non-volatile computer-readable storage medium carrying the computer-readable code. When the computer-readable code runs in a processor of an electronic device, the processor in the electronic device implements any of the above-described methods for detecting abnormal vehicle vibration.

[0162] Those skilled in the art will appreciate that embodiments of the present disclosure may be provided as methods, systems, or computer program products. Thus, the present disclosure may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present disclosure may take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code.

[0163] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0164] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0165] Computer-readable media includes both permanent and non-permanent, removable and non-removable storage media. Storage media can implement any method or technology for storing information, which can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media, such as modulated data signals and carrier waves.

[0166] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0167] The foregoing description is intended only to provide specific embodiments of the present disclosure, intended to enable those skilled in the art to understand and implement the present disclosure. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure is not intended to be limited to the embodiments described herein, but rather to be construed in the broadest manner consistent with the principles and novel features disclosed herein. Industrial Applicability

[0168] The present disclosure relates to a method, apparatus, device, vehicle, medium, and program product for detecting abnormal vehicle jitter. The method comprises: obtaining a target time series corresponding to an original time series signal; the original time series signal is a time series signal corresponding to a vehicle status signal, and the target time series is a stationary time series obtained by preprocessing the original time series signal; randomly selecting multiple sample points from the target time series, constructing multiple isolation trees, selecting a data point to be detected from the target time series, traversing isolated trees in the multiple isolation trees based on the data point to be detected, and determining an abnormality score of the data point to be detected based on the depth of the data point to be detected in each isolated tree; determining an abnormal point list consisting of detected abnormal points based on the abnormality score of the data point to be detected; and identifying an abnormal jitter interval based on the time interval of each abnormal point in the abnormal point list.

Claims

1. A method for detecting abnormal vehicle vibration, comprising: Obtaining a target time series corresponding to an original time series signal; the original time series signal is a time series signal corresponding to a vehicle state signal, and the target time series is a stationary time series obtained by preprocessing the original time series signal; Randomly select multiple sample points from the target time series, construct multiple isolated trees, and select a data point to be detected from the target time series, traverse an isolated tree in the multiple isolated trees according to the data point to be detected, and determine an abnormality score of the data point to be detected based on the depth of the data point to be detected in each of the isolated trees; Based on the anomaly scores of the data points to be detected, determining an anomaly point list consisting of the detected anomaly points; The abnormal jitter interval is identified according to the time interval of each abnormal point in the abnormal point list.

2. The method according to claim 1, wherein: The step of obtaining a target time series corresponding to the original time series signal includes: Acquire the original time series signal; Performing low-pass filtering on the original time series signal to obtain a first time series; Perform difference processing on the first time series to obtain the target time series.

3. The method according to claim 2, wherein: The low-pass filtering of the original time series signal to obtain a first time series includes: Performing global spectrum analysis on the original time series signal to determine the cutoff frequency of the low-pass filter; A low-pass filter is designed based on the cut-off frequency, and burrs and redundant peaks in the original time series signal are removed by the low-pass filter to obtain the first time series.

4. The method according to claim 3, wherein: The step of designing a low-pass filter based on the cutoff frequency, removing burrs and redundant peaks in the original time series signal by using the low-pass filter, and obtaining the first time series includes: According to the cut-off frequency, the sampling rate is designed to be f s , a low-pass filter of order n; The original time series signal is calculated according to the differential equation of the low-pass filter to obtain the first time series; wherein the differential equation is: Y(k) represents the sequence value at the kth moment, P(k) is the sequence value at the kth moment in the original time series signal, n is the order of the filter, b0~b n is the input signal coefficient, a0~a n is the output signal coefficient, b0~b n and a0~a n The order n and cutoff frequency f of the filter are passed c Sure.

5. The method according to claim 2, wherein: The performing differential processing on the first time series to obtain the target time series includes: Performing first-order difference processing on the first time series to obtain a first-order difference series; or; Performing second-order difference processing on the first time series to obtain a second-order difference sequence; wherein the target time series includes the first-order difference sequence or the second-order difference sequence.

6. The method according to claim 1, wherein: The method randomly selects multiple sample points from the target time series, constructs multiple isolated trees, selects a data point to be detected from the target time series, traverses an isolated tree in the multiple isolated trees according to the data point to be detected, and determines an abnormal score of the data point to be detected based on the depth of the data point to be detected in each of the isolated trees, including: Randomly select multiple sample points from the target time series to generate a sub-sample set, randomly determine a threshold cut point among sample points between the maximum value and the minimum value in the sub-sample set, divide the multiple sample points in the sub-sample set into a left branch and a right branch based on the threshold cut point, and repeat the above steps in the left branch and the right branch respectively until a termination condition is met, thereby constructing multiple isolated trees; Selecting a data point to be detected from the target time series, traversing each of the isolated trees according to the data point to be detected, and calculating the depth of the data point to be detected in each of the isolated trees; According to the depth of each of the isolated trees, the abnormality score of the data point to be detected is calculated.

7. The method according to claim 1, wherein: The abnormal score of the data point to be detected is calculated according to the following method: Wherein, x is the data point to be detected, s(x,ψ) is the anomaly score of the data point to be detected, E(h(x)) is the average depth of q isolated trees iTree; c(ψ) is the average path length of a binary search tree constructed for ψ sample points; q is an integer greater than or equal to 2.

8. The method according to claim 1, wherein: The determining, based on the abnormality scores of the data points to be detected, an abnormality point list consisting of detected abnormal points comprises: Determining anomalies in the target time series based on anomaly scores of the data points to be detected; According to the index of the target time series, the abnormal points in the target time series are arranged in ascending order to obtain the abnormal point list.

9. The method according to claim 1, wherein: The identifying the abnormal jitter interval according to the time interval of each abnormal point in the abnormal point list includes: Taking the first point in the abnormal point list as the starting point of the target abnormal jitter interval, and adding the starting point to the first list; Starting from the second point in the abnormal point list, the time interval between the current abnormal point and the previous abnormal point is calculated. If the time interval is greater than the preset time interval, the previous abnormal point is determined to be the end point of the target abnormal jitter interval, and the current abnormal point is determined to be the starting point of the next target abnormal jitter interval. The end point is added to the second list, and the judgment logic is repeatedly executed to obtain multiple abnormal jitter intervals.

10. The method according to claim 9, wherein: The method further comprises: Calculating the effective length of each abnormal jitter interval according to the starting point in the first list and the ending point in the second list; According to the relationship between the effective length of each abnormal jitter interval and the effective length threshold, it is determined whether the abnormal jitter interval among the multiple abnormal jitter intervals is a continuous abnormal jitter interval, and the continuous abnormal jitter interval among the multiple abnormal jitter intervals is determined as a valid abnormal jitter interval; the valid abnormal jitter interval is used to represent the abnormal jitter interval among the multiple abnormal jitter intervals whose effective length is greater than or equal to the effective length threshold.

11. The method according to claim 10, wherein: The method further comprises: Acquire at least one differential sequence corresponding to at least one valid abnormal jitter interval; Performing fast Fourier transform processing on the at least one differential sequence to obtain spectrum data corresponding to the at least one differential sequence; According to the magnitude relationship between the spectrum data corresponding to the at least one differential sequence and the threshold spectrum data, determining whether each valid abnormal jitter interval belongs to a false detection interval; The abnormal jitter intervals whose spectrum data is less than the threshold spectrum data in the valid abnormal jitter intervals are determined as false detection intervals, and the false detection intervals are removed, and the false detection interval judgment logic is repeatedly executed until all valid abnormal jitter intervals are traversed to obtain at least one target abnormal jitter interval.

12. The method according to claim 1, wherein: The method further comprises: The length of each target abnormal jitter interval is used as the corresponding abnormal jitter duration, and the abnormal jitter duration is output.

13. A vehicle abnormal vibration detection device, comprising: An acquisition module is configured to acquire a target time series corresponding to an original time series signal; the original time series signal is a time series signal corresponding to a vehicle state signal, and the target time series is a stationary time series obtained by preprocessing the original time series signal; A detection module is configured to randomly select multiple sample points from the target time series, construct multiple isolated trees, select a data point to be detected from the target time series, traverse an isolated tree in the multiple isolated trees according to the data point to be detected, and determine an abnormality score of the data point to be detected based on the depth of the data point to be detected in each of the isolated trees; A determination module, configured to determine an abnormal point list consisting of detected abnormal points based on the abnormal scores of the data points to be detected; The identification module is configured to identify the abnormal jitter interval according to the time interval of each abnormal point in the abnormal point list.

14. The device according to claim 13, wherein: The acquisition module includes an acquisition unit, a filtering unit and a processing unit; wherein, The acquisition unit is configured to acquire the original time series signal; The filtering unit is configured to perform low-pass filtering on the original time series signal to obtain a first time series; The processing unit is configured to perform differential processing on the first time series to obtain the target time series.

15. According to the device of claim 14, the filtering unit is further configured to perform a global spectrum analysis on the original time series signal to determine a cutoff frequency of a low-pass filter; design a low-pass filter based on the cutoff frequency, and remove glitches and redundant peaks in the original time series signal through the low-pass filter to obtain a first time series.

16. The device according to claim 15, wherein the filtering unit is further configured to design a sampling rate of f according to the cut-off frequency. s , a low-pass filter of order n; calculating the original time series signal according to the differential equation of the low-pass filter to obtain the first time series; wherein, The difference equation is Y(k) represents the sequence value at the kth moment, P(k) is the sequence value at the kth moment in the original time series signal, n is the order of the filter, b0~b n is the input signal coefficient, a0~a n is the output signal coefficient, b0~b n and a0~a n The filter order n, cutoff frequency f c Sure.

17. The device according to claim 14, wherein: The processing unit is further configured to Performing first-order difference processing on the first time series to obtain a first-order difference series; or; Performing second-order difference processing on the first time series to obtain a second-order difference sequence; wherein the target time series includes the first-order difference sequence or the second-order difference sequence.

18. The device according to claim 13, wherein: The detection module is further configured to randomly select multiple sample points from the target time series to generate a sub-sample set, randomly determine a threshold cut point among sample points between the maximum value and the minimum value in the sub-sample set, divide the multiple sample points in the sub-sample set into a left branch and a right branch based on the threshold cut point, and repeat the above steps in the left branch and the right branch respectively until a termination condition is met, thereby constructing multiple isolated trees; A data point to be detected is selected from the target time series, each of the isolated trees is traversed according to the data point to be detected, and the depth of the data point to be detected in each of the isolated trees is calculated; and the abnormal score of the data point to be detected is calculated according to the depth in each of the isolated trees.

19. The device according to claim 13, wherein: The determination module is further configured to calculate the abnormality score of the to-be-detected data point in the following manner: Wherein, x is the data point to be detected, s(x,ψ) is the anomaly score of the data point to be detected, E(h(x)) is the average depth of q isolated trees iTree; c(ψ) is the average path length of a binary search tree constructed for ψ sample points; q is an integer greater than or equal to 2.

20. The device according to claim 13, wherein: The determination module is further configured to determine the abnormal points in the target time series based on the abnormal scores of the data points to be detected; and to arrange the abnormal points in the target time series in ascending order according to the index of the target time series to obtain the abnormal point list.

21. The device according to claim 13, wherein: The identification module is further configured to use the first point in the abnormal point list as the starting point of the target abnormal jitter interval, and add the starting point to the first list; Starting from the second point in the abnormal point list, the time interval between the current abnormal point and the previous abnormal point is calculated. If the time interval is greater than the preset time interval, the previous abnormal point is determined to be the end point of the target abnormal jitter interval, and the current abnormal point is determined to be the starting point of the next target abnormal jitter interval. The end point is added to the second list, and the judgment logic is repeatedly executed to obtain multiple abnormal jitter intervals.

22. The device according to claim 21, wherein The device further includes a filtering module configured to calculate the effective length of each abnormal jitter interval according to the starting point in the first list and the ending point in the second list; determine whether each abnormal jitter interval in the multiple abnormal jitter intervals is a continuous abnormal jitter interval according to the relationship between the effective length of each abnormal jitter interval and the effective length threshold, and determine the continuous abnormal jitter interval in the multiple abnormal jitter intervals as a valid abnormal jitter interval; The effective abnormal jitter interval is used to indicate an abnormal jitter interval whose effective length is greater than or equal to an effective length threshold among the multiple abnormal jitter intervals.

23. The device according to claim 22, wherein: The filtering module is further configured to obtain at least one differential sequence corresponding to at least one valid abnormal jitter interval; perform fast Fourier transform processing on the at least one differential sequence to obtain spectrum data corresponding to the at least one differential sequence; determine whether each valid abnormal jitter interval belongs to a false detection interval based on the size relationship between the spectrum data corresponding to the at least one differential sequence and the threshold spectrum data; determine the abnormal jitter interval whose spectrum data is less than the threshold spectrum data in each valid abnormal jitter interval as a false detection interval, remove the false detection interval, and repeatedly execute the false detection interval judgment logic until all valid abnormal jitter intervals are traversed to obtain at least one target abnormal jitter interval.

24. The device according to claim 13, wherein: The device further includes an output module configured to use the length of each target abnormal jitter interval as the corresponding abnormal jitter duration and output the abnormal jitter duration.

25. An electronic device comprising: one or more processors; a storage device configured to store one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the vehicle abnormal vibration detection method as described in any one of claims 1 to 12.

26. A vehicle comprising: The electronic device as claimed in claim 25.

27. A computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the vehicle abnormal vibration detection method as claimed in any one of claims 1 to 12 is implemented.

28. A computer program, comprising a computer-readable code, wherein when the computer-readable code is run in an electronic device, the processor of the electronic device executes the method for detecting abnormal vehicle vibration as claimed in any one of claims 1 to 12.

29. A computer program product, comprising a computer-readable code, or a non-volatile computer-readable storage medium carrying the computer-readable code, wherein when the computer-readable code runs in a processor of an electronic device, the processor in the electronic device implements the vehicle abnormal vibration detection method as described in any one of claims 1 to 12 when executing.

Images