Dynamic low pressure identification method based on tire pressure data

Abstract

The invention relates to the technical field of pressure identification, and particularly discloses a dynamic low pressure identification method based on tire pressure data, which comprises the following steps: acquiring a tire pressure value, obtaining a low pressure proximity ratio through numerical analysis, constructing a proximity ratio data set, and analyzing the continuity of the low pressure proximity ratio; if the vehicle driving data continuously appear, drawing a time-low voltage approach ratio change curve, determining a sudden change point by using a local abnormal factor algorithm, then obtaining the sudden change point of the vehicle driving data, and constructing a low voltage moment group and a driving moment group; constructing a moment distance matrix by calculating two groups of moment distances, obtaining a normalized moment distance matrix through normalization processing, recognizing a low-voltage state related to vehicle driving data by using a support vector machine algorithm in combination with vehicle historical data, and feeding back the low-voltage state to a vehicle control system; the method can effectively identify the dynamic low pressure, is helpful for timely monitoring the tire low pressure state of the vehicle, and is helpful for guaranteeing the driving safety.

Description

Technical Field

[0001] The present invention relates to the technical field of pressure identification, and in particular to a dynamic low pressure identification method based on tire pressure data. Background Art

[0002] A Chinese patent application with publication number CN103085612B discloses a tire low pressure alarm device and method, comprising: measuring the pressure of each of the above-mentioned multiple tires respectively, and sending a signal corresponding to the above-mentioned pressure; identifying whether the pressure corresponding to the signal is the pressure of the front tire or the pressure of the rear tire, if it is the pressure of the above-mentioned front tire, comparing it with a first reference pressure, and if it is the pressure of the above-mentioned rear tire, comparing it with a second reference pressure, to determine whether the tire corresponding to the above-mentioned signal is in a low pressure state.

[0003] Current technology simply sets different reference pressures based on tire position for comparison and judgment, without considering the dynamic impact of multiple complex factors on tire pressure during vehicle driving. For example, when the vehicle is driving, changes in vehicle speed, passing through bumpy roads, changes in driving conditions, and changes in the tire's own physical state will all cause tire pressure fluctuations. If the vehicle's driving data can be combined, by analyzing the mutation points of the time and pressure change curve, as well as the correlation with the mutation points of time and driving data, dynamic low pressure conditions can be identified more comprehensively and accurately.

[0004] When judging the low-pressure state, current technology lacks in-depth analysis of the continuity and mutation of pressure data. It is difficult to detect the trend and characteristics of abnormal changes in pressure data by relying solely on comparison with the baseline pressure. By constructing a proximity ratio data group, using the run detection method and information entropy to calculate continuous judgment values, judging whether the low-pressure proximity ratio appears continuously, and then combining the local anomaly factor algorithm to accurately determine the mutation point, compared with current technology, it can capture abnormal changes in tire pressure more promptly, which is conducive to timely detection of potential low-pressure risks.

[0005] To this end, the present invention provides a dynamic low pressure identification method based on tire pressure data. Summary of the invention

[0006] The object of the present invention is to provide a dynamic low pressure identification method based on tire pressure data to solve the above-mentioned problems.

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] The dynamic low pressure identification method based on tire pressure data comprises the following steps:

[0009] Step 1, obtaining the air pressure value of the tire, performing numerical analysis on the air pressure value of the tire, and obtaining the low pressure approach ratio;

[0010] Step 2: Obtain the low pressure close ratios at all monitoring moments within the monitoring period, construct a close ratio data group, perform continuity analysis on the low pressure close ratios in the close ratio data group, and determine whether the low pressure close ratios appear continuously;

[0011] Step 3: If the low pressure approach ratio appears continuously in the approach ratio data set, draw a time-low pressure approach ratio change curve, and determine the mutation point of the low pressure approach ratio on the change curve;

[0012] Step 4: based on the mutation point of the low pressure approach ratio on the change curve, obtain the driving data mutation point of the vehicle driving data, and construct a low pressure moment group and a driving moment group;

[0013] Step 5: Perform time distance analysis on the low-pressure time group and the driving time group, construct a normalized time distance matrix, and use machine learning methods to analyze the normalized time distance matrix to identify the low-pressure state related to the vehicle driving data.

[0014] As a further solution of the present invention: the low pressure approach ratio is obtained as follows:

[0015] During the monitoring period, the tire pressure value at each monitoring moment is obtained in real time;

[0016] Obtaining a minimum endpoint value of a preset air pressure range value and a preset low pressure warning value, and using the minimum endpoint value of the preset air pressure range value and the preset low pressure warning value as the low pressure range value;

[0017] If the air pressure value is less than the minimum endpoint value of the low pressure range value, a low tire pressure warning is generated to the vehicle control system;

[0018] If the air pressure value is within the low pressure range, the air pressure value is subtracted from the minimum endpoint value of the low pressure range to obtain the air pressure difference;

[0019] The pressure approach difference is processed by ratioing the length of the pressure range value to obtain the low pressure approach ratio.

[0020] As a further solution of the present invention: the determination method of whether the low pressure approach ratio appears continuously is:

[0021] If a low pressure proximity ratio is generated at the monitoring time during the monitoring period, the low pressure proximity ratio is stored in the proximity ratio data group;

[0022] If no low-pressure proximity ratio is generated at the monitoring time during the monitoring period, a value of "0" is stored in the proximity ratio data group and marked as a null-value proximity ratio;

[0023] Numerical analysis is performed on the approach ratio data set to obtain the probability of low-pressure approach ratio and continuous null value ratio in each subsection of the approach ratio data set;

[0024] Based on the probability of low pressure approach ratio and continuous null value ratio in each sub-section, the continuous judgment value of the approach ratio data group is obtained through the information entropy formula;

[0025] If the continuous determination value is lower than the continuous determination threshold, it is considered that the low-pressure proximity ratio in the proximity ratio data set appears continuously.

[0026] As a further solution of the present invention: the probability of the low pressure close ratio in each subsection of the close ratio data set is obtained in the following manner:

[0027] Based on the proximity ratio data group, the proximity ratio data group is divided into a plurality of sections according to the length of the fixed monitoring time to obtain sub-sections;

[0028] Obtaining the number of low pressure close ratios and the number of null value close ratios in the sub-section, summing the number of low pressure close ratios and the number of null value close ratios to obtain the sum of the section numbers;

[0029] The probability of the low pressure close ratio in each sub-segment is obtained by performing ratio processing on the number of low pressure close ratios and the sum of the number of segments.

[0030] As a further solution of the present invention: the continuous null value ratio is obtained in the following manner:

[0031] Use the run detection method to calculate the maximum value of the number of consecutive null value proximity ratios in each sub-segment to obtain the segment continuous null value;

[0032] Ratio the segment continuous null values ​​to the sum of the segment quantities to obtain the continuous null value ratio.

[0033] As a further solution of the present invention: the mutation point of the low pressure approach ratio on the change curve is obtained as follows:

[0034] If the low pressure approach ratio appears continuously in the approach ratio data group, a time-low pressure approach ratio variation curve is drawn in a two-dimensional rectangular coordinate system with the monitoring time as the X-axis and the low pressure approach ratio as the Y-axis;

[0035] Using the local anomaly factor algorithm, a mutation analysis model is established to determine the mutation point of the low-pressure approach ratio of the time-low-pressure approach ratio change curve.

[0036] As a further solution of the present invention: the mutation analysis model is established in the following manner:

[0037] The low pressure approach ratio corresponding to each monitoring moment in the acquisition time-low pressure approach ratio curve is marked as a data detection point;

[0038] Get the reachable distance d of the data detection point around m data detection points in the two-dimensional rectangular coordinate system k , k is the number of the surrounding data detection point, and the value range of k is [1,m];

[0039] By formula: , obtain the reachable distance from the data detection point to the surrounding data detection points, where d k1 is the Chebyshev distance from the data detection point to the surrounding data detection points, d k2 is the Euclidean distance from the data detection point to the surrounding data detection points;

[0040] By formula: Obtain the abnormal factor Lo of the data detection point;

[0041] When the abnormal factor Lo is higher than the preset factor judgment value, the data detection point is considered to be a mutation point on the time-low pressure approach ratio change curve.

[0042] As a further solution of the present invention: the low-pressure timing group and the driving timing group are constructed as follows:

[0043] Based on the monitoring time corresponding to the mutation point of the low pressure approach ratio on the change curve, the vehicle driving data is obtained from the vehicle driving log;

[0044] The vehicle driving data includes: vehicle speed, vehicle bump value, and tire temperature;

[0045] Based on the vehicle driving data, with time as the X-axis and the vehicle driving data as the Y-axis, three change curves of time-vehicle driving data consisting of vehicle speed, vehicle bump value, and tire temperature are drawn in a two-dimensional rectangular coordinate system;

[0046] Based on the local anomaly factor algorithm, the mutation points of the three change curves of time-vehicle driving data are obtained and marked as driving data mutation points;

[0047] Obtain the monitoring time corresponding to the mutation point of the low-pressure approach ratio within the monitoring period, and construct a low-pressure time group;

[0048] Obtain the monitoring time corresponding to the sudden change point of driving data within the monitoring period and construct a driving time group.

[0049] As a further solution of the present invention: the identification method of the low-pressure state related to the vehicle driving data is:

[0050] Based on the low-voltage time group and the driving time group, the time distances of the low-voltage time group and the driving time group are calculated and used as data elements of the time distance matrix to construct the time distance matrix;

[0051] Normalize each data element in the time distance matrix to construct a normalized time distance matrix;

[0052] Obtain the normalized time distance matrix of the vehicle history from the vehicle's driving log to obtain the historical time distance matrix;

[0053] The historical time distance matrix is ​​combined with the normalized time distance matrix using the support vector machine algorithm to identify the monitoring time in the normalized time distance matrix where the mutation point in the driving time group causes the mutation point in the low-voltage time group.

[0054] The monitoring time at which a mutation point in the low-pressure moment group is generated due to a mutation point in the driving moment group and the corresponding low-pressure approach ratio are obtained to obtain a low-pressure state related to the vehicle driving data.

[0055] As a further solution of the present invention: the construction method of DT in the time distance matrix is:

[0056] By formula: Get the time distance D (t j1 ,s j2 ),in, Represents each monitoring moment in the low-voltage moment group, Represents each monitoring moment in the driving moment group;

[0057] By recursive formula: Fill in the data elements of DT in the moment distance matrix, in They respectively represent the time distance of the cell above, the time distance of the cell to the left, and the time distance of the cell above the left of the time distance matrix DT.

[0058] Beneficial effects of the present invention:

[0059] 1. By acquiring the tire pressure value in real time and calculating the low-pressure approach ratio, and using the run detection method and information entropy to perform continuous analysis on the low-pressure approach ratio, it is helpful to timely monitor the abnormal fluctuation of the tire low-pressure approach ratio. For example, during long-term driving of the vehicle, the tire pressure may slowly drop due to factors such as temperature changes and slight air leakage. Traditional methods may not be able to detect this gradual trend in time. However, the present invention can capture these anomalies in advance through continuous monitoring and analysis of the low-pressure approach ratio, and issue an early warning before the tire pressure problem worsens.

[0060] 2. The mutation point of the low-pressure approach ratio is determined by the local anomaly factor algorithm. At the same time, the low-pressure moment group and the driving moment group are constructed in combination with the vehicle's driving data (vehicle speed, vehicle bump value, tire temperature) for correlation analysis. It can be determined that the mutation of the low-pressure approach ratio is caused by a change in the vehicle's driving state. For example, if the tire low-pressure approach ratio suddenly changes when the vehicle makes a sharp turn, and the mutation moment is highly consistent with the mutation point of the driving data, it is helpful to accurately determine that the low-pressure anomaly is caused by the vehicle's driving operation, providing a basis for subsequent targeted measures. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] The present invention will be further described below in conjunction with the accompanying drawings.

[0062] Figure 1 is a flow chart of air pressure acquisition and continuity analysis of the present invention;

[0063] Figure 2 It is a flow chart of the correlation analysis between air pressure and vehicle driving status in the present invention. DETAILED DESCRIPTION

[0064] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0065] Example 1

[0066] See also Figure 1 As shown, the present invention is a dynamic low pressure identification method based on tire pressure data, comprising the following steps:

[0067] Step 1: Obtain the tire pressure value, perform numerical analysis on the tire pressure value, and obtain the low pressure approach ratio;

[0068] In some embodiments, a pressure sensor is installed in the tire to collect the air pressure value inside the tire in real time;

[0069] During the monitoring period, the tire pressure value at each monitoring moment is obtained in real time;

[0070] It should be noted that there are multiple monitoring moments within the monitoring cycle;

[0071] Obtaining a minimum endpoint value of a preset air pressure range value and a preset low pressure warning value, and using the minimum endpoint value of the preset air pressure range value and the preset low pressure warning value as the low pressure range value;

[0072] It should be noted that the air pressure range value and the low pressure warning value are set by professionals in this field, and the low pressure warning value is lower than the minimum endpoint value of the air pressure range value;

[0073] If the air pressure value is less than the minimum endpoint value of the low pressure range value, a low tire pressure warning is generated to the vehicle control system;

[0074] If the air pressure value is within the low pressure range, the air pressure value is subtracted from the minimum endpoint value of the low pressure range to obtain the air pressure difference;

[0075] The pressure approach difference is processed by ratio with the length of the pressure range value to obtain the low pressure approach ratio;

[0076] Step 2: Obtain the low pressure close ratios at all monitoring moments within the monitoring period, construct a close ratio data group, perform continuity analysis on the low pressure close ratios in the close ratio data group, and determine whether the low pressure close ratios appear continuously;

[0077] During the monitoring period, the low-pressure proximity ratio generated at each monitoring moment is judged to construct a proximity ratio data group;

[0078] Specifically, if a low-pressure proximity ratio is generated at the monitoring moment within the monitoring period, the low-pressure proximity ratio is stored in the proximity ratio data group;

[0079] If no low-pressure proximity ratio is generated at the monitoring time during the monitoring period, a value of "0" is stored in the proximity ratio data group and marked as a null-value proximity ratio;

[0080] Based on the proximity ratio data group, the proximity ratio data group is divided into a plurality of sections according to the length of the fixed monitoring time to obtain sub-sections;

[0081] Obtaining the number of low pressure close ratios and the number of null value close ratios in the sub-section, summing the number of low pressure close ratios and the number of null value close ratios to obtain the sum of the section numbers;

[0082] The number of low-pressure close ratios in the sub-segment is processed by ratio processing with the sum of the number of segments to obtain the probability of low-pressure close ratio in each sub-segment, and the probability of low-pressure close ratio in each sub-segment is marked as q i , i is the number of the sub-segment, the value range of i is [1,n], and n is the total number of sub-segments of the close ratio data group;

[0083] Use the run detection method to calculate the maximum value of the number of consecutive null value proximity ratios in each sub-segment to obtain the segment continuous null value;

[0084] It should be noted that the run detection method scans the low-pressure proximity ratio in each sub-segment. If a null value proximity ratio is scanned, the next data of the sub-segment is scanned to see if it is a null value proximity ratio. If it is a null value proximity ratio, the scan is continued until the data of the low-pressure proximity ratio is scanned to obtain a continuous null value proximity ratio. When the run detection method identifies the last data in the sub-segment, all continuous null value proximity ratios in the sub-segment are compared to obtain the maximum value of the number of continuous null value proximity ratios in the sub-segment, that is, the segment continuous null value.

[0085] For example, if the sub-segment is [0.25, 0, 0, 0.75, 0, 1.1], the segment continuous null value of the sub-segment is 2;

[0086] Ratio the segment continuous null value to the segment number and get the continuous null value ratio, which is marked as l i ;

[0087] By formula: Obtain the continuous determination value H of the proximity ratio data group, where a1 and a2 are 1.25 and 0.86 respectively;

[0088] It should be noted that , Respectively represent the information entropy of low pressure close ratio and continuous null value ratio in the close ratio data group, which is used to determine the continuity of low pressure close ratio in the close ratio data group. When the information entropy of low pressure close ratio is smaller, it means that the distribution of low pressure close ratio in the close ratio data group is more concentrated, and the information entropy of continuous null value ratio is smaller, it means that the continuous null value ratio appears more concentrated.

[0089] Compare the continuity determination value H with a preset continuity determination threshold value to determine whether the low-pressure proximity ratio in the proximity ratio data group appears continuously;

[0090] If the continuous determination value H is lower than the continuous determination threshold, it means that the low pressure approach ratio distribution is relatively concentrated and continuous, and it is considered that the low pressure approach ratio appears continuously in the approach ratio data group;

[0091] If the continuity determination value H is higher than the continuity determination threshold, and the low-pressure proximity ratio distribution is relatively dispersed and discrete, it is considered that the low-pressure proximity ratio in the proximity ratio data group does not appear continuously;

[0092] The technical solution of this embodiment is: obtaining the air pressure value of the tire, performing numerical analysis on the air pressure value of the tire to obtain the low-pressure approach ratio, obtaining the low-pressure approach ratio of all monitoring moments in the monitoring period, constructing a proximity ratio data group, performing continuity analysis on the low-pressure approach ratio in the proximity ratio data group, and determining whether the low-pressure approach ratio appears continuously, which is conducive to timely monitoring of abnormal fluctuations in the tire low-pressure approach ratio and judging the changing trend of the tire low pressure in the time dimension.

[0093] Example 2

[0094] Step 3: If the low pressure approach ratio appears continuously in the approach ratio data set, draw a time-low pressure approach ratio change curve, and determine the mutation point of the low pressure approach ratio on the change curve;

[0095] If the low pressure approach ratio appears continuously in the approach ratio data group, a time-low pressure approach ratio variation curve is drawn in a two-dimensional rectangular coordinate system with the monitoring time as the X-axis and the low pressure approach ratio as the Y-axis;

[0096] Using the local anomaly factor algorithm, a mutation analysis model is constructed to determine the mutation point of the low-pressure approach ratio of the time-low-pressure approach ratio change curve;

[0097] It should be noted that in the time-low pressure approach ratio change curve, the mutation point refers to the point at which the low pressure approach ratio value changes sharply at a certain monitoring time. When the low pressure approach ratio of the tire is at the mutation point, the tire pressure may change due to changes in the physical state of the tire, such as the tire being punctured by a sharp object, causing the gas inside the tire to overflow to the outside;

[0098] If the pressure sensor works abnormally at a certain monitoring moment, such as momentary signal interference or hardware failure, causing the pressure value of the pressure sensor inside the tire to change, or the vehicle's driving state changes, such as the vehicle passing through a bumpy road, sudden braking, sharp turns, etc., causing the tire pressure to change, resulting in a sudden change in the low pressure approach ratio;

[0099] Specifically, the mutation analysis model is established as follows:

[0100] The low pressure approach ratio corresponding to each monitoring moment in the acquisition time-low pressure approach ratio curve is marked as a data detection point;

[0101] Get the reachable distance d of the data detection point around m data detection points in the two-dimensional rectangular coordinate system k , k is the number of the surrounding data detection point, and the value range of k is [1,m];

[0102] Preferably, m=5;

[0103] By formula: , obtain the reachable distance from the data detection point to the surrounding data detection points, where d k1 is the Chebyshev distance from the data detection point to the surrounding data detection points, d k2 is the Euclidean distance from the data detection point to the surrounding data detection points;

[0104] For example, if the coordinates of the data detection point are (x0, y0), and the coordinates of one of the surrounding data detection points are (x1, y1), by the formula: Get the Chebyshev distance using the formula: Get the Euclidean distance;

[0105] It should be noted that the Chebyshev distance can reflect the influence of factors with more significant changes at the monitoring time or when the low pressure approaches the ratio;

[0106] The Euclidean distance comprehensively considers the two-dimensional information of monitoring time and low-pressure proximity, reflects the straight-line distance between two points in two-dimensional space, compares and analyzes the Chebyshev distance and the Euclidean distance, measures the distance relationship between data points from different angles, and more comprehensively captures the differences between data points in different dimensions, which helps to more accurately identify mutation points.

[0107] By formula: Obtain the abnormal factor Lo of the data detection point;

[0108] When the abnormal factor Lo is higher than the preset factor judgment value, the data detection point is considered to be a mutation point on the time-low pressure approach ratio change curve;

[0109] Step 4: based on the mutation point of the low pressure approach ratio on the change curve, obtain the driving data mutation point of the vehicle driving data, and construct a low pressure moment group and a driving moment group;

[0110] Based on the monitoring time corresponding to the mutation point of the low pressure approach ratio on the change curve, the vehicle driving data is obtained from the vehicle driving log;

[0111] The vehicle driving data includes: vehicle speed, vehicle bump value, and tire temperature;

[0112] The vehicle speed and tire temperature are directly measured by the vehicle's speed sensor and temperature sensor, and the vehicle bump value is obtained by the displacement sensor installed on the vehicle suspension by measuring the vertical displacement of the suspension;

[0113] Based on the vehicle driving data, with time as the X-axis and the vehicle driving data as the Y-axis, three change curves of time-vehicle driving data consisting of vehicle speed, vehicle bump value, and tire temperature are drawn in a two-dimensional rectangular coordinate system;

[0114] Based on the local anomaly factor algorithm, the mutation points of the three change curves of time-vehicle driving data are obtained and marked as driving data mutation points;

[0115] Obtain the monitoring time corresponding to the mutation point of the low-pressure approach ratio within the monitoring period, and construct a low-pressure moment group, where each monitoring moment in the low-pressure moment group is marked as t j1 , j1 represents the number of each monitoring moment of the low-voltage moment group, and the value range of j1 is [1, a1];

[0116] Obtain the monitoring time corresponding to the driving data mutation point within the monitoring period, and construct a driving time group, where each monitoring time in the driving time group is marked as s j2 ,j2 represents the number of each monitoring time of the driving time group, and the value range of j2 is [1,a2];

[0117] Step 5: Perform time distance analysis on the low-pressure time group and the driving time group, construct a normalized time distance matrix DTW, and use a machine learning method to analyze the normalized time distance matrix DTW to identify the low-pressure state related to the vehicle driving data;

[0118] Based on the low-pressure time group and the driving time group, the time distance D (t j1 ,sj2 ) and used as the data element of the time distance matrix to construct the time distance matrix DT;

[0119] It should be noted that the low-pressure moment group records the monitoring time of the low-pressure approach ratio mutation point, and the driving moment group records the monitoring time of the driving data mutation point. During vehicle operation, the changes of the low-pressure moment group and the driving moment group may be dislocated in time;

[0120] The time distance matrix DT can quantify the degree of their temporal alignment while allowing the time axis to bend. For example, even if the mutations of the low-pressure approach ratio and the driving data do not occur strictly synchronously, DT can find the most reasonable correspondence through dynamic programming and determine the degree of temporal correlation between the low-pressure moment group and the driving moment group.

[0121] Among them, the dimension of the moment distance matrix DT is a1*a2;

[0122] Specifically, through the formula: Get the time distance D (t j1 ,s j2 ),in, Represents each monitoring moment in the low-voltage moment group, Represents each monitoring moment in the driving moment group;

[0123] By recursive formula: Fill in the data elements of DT in the moment distance matrix, in They represent the time distance of the upper cell, the left cell, and the upper left cell of the time distance matrix DT respectively;

[0124] It should be noted that if the time distance of the cell above, the time distance of the cell to the left, and the time distance of the cell to the upper left of the time distance matrix DT do not exist, then the values ​​of the time distance of the cell above, the time distance of the cell to the left, and the time distance of the cell to the upper left of the time distance matrix DT are 0;

[0125] Normalize each data element in the time distance matrix DT and map each data element in the time distance matrix DT to the range of [0,1];

[0126] By formula: Get the DTW value of each data element after normalization;

[0127] Based on the normalized data element value DTW, a normalized time distance matrix DTW is constructed;

[0128] From the vehicle driving log, the historical driving data of the vehicle is obtained, and the normalized time distance matrix DTW of the vehicle history is constructed to obtain the historical time distance matrix;

[0129] The historical time distance matrix is ​​combined with the normalized time distance matrix DTW using the support vector machine algorithm to identify the monitoring time in the normalized time distance matrix DTW where the mutation point in the driving time group causes the mutation point in the low-voltage time group.

[0130] Obtain the monitoring time when the mutation point in the driving time group causes the mutation point in the low pressure time group, the corresponding low pressure approach ratio, obtain the low pressure state related to the vehicle driving data, and feed it back to the vehicle control system;

[0131] It should be noted that the support vector machine algorithm performs model training on the normalized time distance matrix of the vehicle's history and the current normalized time distance matrix, and selects the SVM kernel function to identify the monitoring moment when the low-pressure moment group produces a mutation point caused by the mutation point in the driving moment group, and the corresponding low-pressure proximity ratio; the vehicle control system refers to the vehicle itself, and the system that assists the driver in driving the car or replaces the driver in automatically driving the car is contained in the vehicle itself.

[0132] The technical solution of this embodiment is: if the low pressure approach ratio appears continuously in the approach ratio data group, draw a time-low pressure approach ratio change curve, determine the mutation point of the low pressure approach ratio on the change curve, obtain the driving data mutation point of the vehicle driving data based on the mutation point of the low pressure approach ratio on the change curve, construct a low pressure moment group and a driving moment group, perform time distance analysis on the low pressure moment group and the driving moment group, construct a normalized time distance matrix DTW, use a machine learning method to analyze the normalized time distance matrix DTW, identify the low pressure state related to the vehicle driving data, which is conducive to accurately determine whether the low pressure anomaly is caused by the vehicle driving operation, and provide a basis for taking targeted measures later.

[0133] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. A dynamic low pressure identification method based on tire pressure data, characterized in that: The steps include: Step 1, obtaining the air pressure value of the tire, performing numerical analysis on the air pressure value of the tire, and obtaining the low pressure approach ratio; Step 2: Obtain the low pressure close ratios at all monitoring moments within the monitoring period, construct a close ratio data group, perform continuity analysis on the low pressure close ratios in the close ratio data group, and determine whether the low pressure close ratios appear continuously; Step 3: If the low pressure approach ratio appears continuously in the approach ratio data set, draw a time-low pressure approach ratio change curve, and determine the mutation point of the low pressure approach ratio on the change curve; Step 4: based on the mutation point of the low pressure approach ratio on the change curve, obtain the driving data mutation point of the vehicle driving data, and construct a low pressure moment group and a driving moment group; Step 5: Perform time distance analysis on the low-pressure time group and the driving time group, construct a normalized time distance matrix, and use machine learning methods to analyze the normalized time distance matrix to identify the low-pressure state related to the vehicle driving data.

2. The dynamic low pressure identification method based on tire pressure data according to claim 1, characterized in that: The low pressure approach ratio is obtained as follows: During the monitoring period, the tire pressure value at each monitoring moment is obtained in real time; Obtaining a minimum endpoint value of a preset air pressure range value and a preset low pressure warning value, and using the minimum endpoint value of the preset air pressure range value and the preset low pressure warning value as the low pressure range value; If the air pressure value is within the low pressure range, the air pressure value is subtracted from the minimum endpoint value of the low pressure range to obtain the air pressure difference; The pressure approach difference is processed by ratioing the length of the pressure range value to obtain the low pressure approach ratio.

3. The dynamic low pressure identification method based on tire pressure data according to claim 1, characterized in that: Whether the low pressure approach ratio appears continuously is determined by: If a low pressure proximity ratio is generated at the monitoring time during the monitoring period, the low pressure proximity ratio is stored in the proximity ratio data group; If no low-pressure proximity ratio is generated at the monitoring time during the monitoring period, a value of "0" is stored in the proximity ratio data group and marked as a null-value proximity ratio; Numerical analysis is performed on the approach ratio data set to obtain the probability of low-pressure approach ratio and continuous null value ratio in each subsection of the approach ratio data set; Based on the probability of low pressure approach ratio and continuous null value ratio in each sub-section, the continuous judgment value of the approach ratio data group is obtained through the information entropy formula; If the continuous determination value is lower than the continuous determination threshold, it is considered that the low-pressure proximity ratio in the proximity ratio data set appears continuously.

4. The dynamic low pressure identification method based on tire pressure data according to claim 3, characterized in that: The probability of the low pressure close ratio in each subsection of the close ratio data set is obtained as follows: Based on the proximity ratio data group, the proximity ratio data group is divided into a plurality of sections according to the length of the fixed monitoring time to obtain sub-sections; Obtaining the number of low pressure close ratios and the number of null value close ratios in the sub-section, summing the number of low pressure close ratios and the number of null value close ratios to obtain the sum of the section numbers; The probability of the low pressure close ratio in each sub-segment is obtained by performing ratio processing on the number of low pressure close ratios and the sum of the number of segments.

5. The dynamic low pressure identification method based on tire pressure data according to claim 4, characterized in that: The continuous null value ratio is obtained as follows: Use the run detection method to calculate the maximum value of the number of consecutive null value proximity ratios in each sub-segment to obtain the segment continuous null value; Ratio the segment continuous null values ​​to the sum of the segment quantities to obtain the continuous null value ratio.

6. The dynamic low pressure identification method based on tire pressure data according to claim 1, characterized in that: The method for obtaining the mutation point of the low pressure approach ratio on the change curve is as follows: If the low pressure approach ratio appears continuously in the approach ratio data group, a time-low pressure approach ratio variation curve is drawn in a two-dimensional rectangular coordinate system with the monitoring time as the X-axis and the low pressure approach ratio as the Y-axis; Using the local anomaly factor algorithm, a mutation analysis model is established to determine the mutation point of the low-pressure approach ratio of the time-low-pressure approach ratio change curve.

7. The dynamic low pressure identification method based on tire pressure data according to claim 6, characterized in that: The mutation analysis model is established as follows: The low pressure approach ratio corresponding to each monitoring moment in the acquisition time-low pressure approach ratio curve is marked as a data detection point; Get the reachable distance d of the data detection point around m data detection points in the two-dimensional rectangular coordinate system k , k is the number of the surrounding data detection point, and the value range of k is [1,m]; By formula: , obtain the reachable distance from the data detection point to the surrounding data detection points, where d k1 is the Chebyshev distance from the data detection point to the surrounding data detection points, d k2 is the Euclidean distance from the data detection point to the surrounding data detection points; By formula: Obtain the abnormal factor Lo of the data detection point; When the abnormal factor Lo is higher than the preset factor judgment value, the data detection point is considered to be a mutation point on the time-low pressure approach ratio change curve.

8. The dynamic low pressure identification method based on tire pressure data according to claim 1, characterized in that: The low-pressure timing group and the driving timing group are constructed as follows: Based on the monitoring time corresponding to the mutation point of the low pressure approach ratio on the change curve, the vehicle driving data is obtained from the vehicle driving log; The vehicle driving data includes: vehicle speed, vehicle bump value, and tire temperature; Based on the vehicle driving data, three change curves of time-vehicle driving data consisting of vehicle speed, vehicle bump value and tire temperature are drawn in a two-dimensional rectangular coordinate system; Based on the local anomaly factor algorithm, the mutation points of the three change curves of time-vehicle driving data are obtained and marked as driving data mutation points; Obtain the monitoring time corresponding to the mutation point of the low-pressure approach ratio within the monitoring period, and construct a low-pressure time group; Obtain the monitoring time corresponding to the sudden change point of driving data within the monitoring period and construct a driving time group.

9. The dynamic low pressure identification method based on tire pressure data according to claim 8, characterized in that: The identification method of the low-pressure state related to the vehicle driving data is: Based on the low-voltage time group and the driving time group, the time distances of the low-voltage time group and the driving time group are calculated and used as data elements of the time distance matrix to construct the time distance matrix; Normalize each data element in the time distance matrix to construct a normalized time distance matrix; From the vehicle's driving log, obtain the normalized time distance matrix of the vehicle's history to obtain the historical time distance matrix; The historical time distance matrix is ​​combined with the normalized time distance matrix using the support vector machine algorithm to identify the monitoring time in the normalized time distance matrix where the mutation point in the driving time group causes the mutation point in the low-voltage time group. The monitoring time at which a mutation point in the low-pressure moment group is generated due to a mutation point in the driving moment group and the corresponding low-pressure approach ratio are obtained to obtain a low-pressure state related to the vehicle driving data.

10. The dynamic low pressure identification method based on tire pressure data according to claim 9, characterized in that: The construction method of DT in the time distance matrix is: By formula: Get the time distance D (t j1 ,s j2 ),in, Represents each monitoring moment in the low-voltage moment group, Represents each monitoring moment in the driving moment group; By recursive formula: Fill in the data elements of DT in the moment distance matrix, in They respectively represent the time distance of the cell above, the time distance of the cell to the left, and the time distance of the cell above the left of the time distance matrix DT.

Images