Slow leak detection system for vehicle tires

Abstract

A system for detecting leaks in tires of a vehicle is provided, the system including sensors that measure pressure and temperature of the tires. The system has a processor that obtains data from the sensors and detects a leak in one of the tires by using a natural leak model and / or a damage comparison model. The natural leak model uses temperature measurements of the tire to detect a leak in the tire, which may be a slow leak of the tire. The damage comparison model detects a leak in the tire by comparing the tire with other tires on the vehicle.

Description

Technical Field

[0001] The present subject matter relates to a pressure loss detection system for a vehicle that can alert a user that one or more tires are experiencing a pressure loss that should be addressed. More specifically, the present application relates to a pressure loss detection system for a vehicle that uses a comparison of the tire to other tires on the vehicle and a natural pressure loss model to detect leaks in the tire. Background Art

[0002] Non-pneumatic tires on a vehicle need to be monitored to ensure that they are properly inflated and not damaged. In addition to manually checking the tires, it is known to provide a vehicle with a tire pressure monitoring system that can automatically monitor the tires and provide this information to the operator of the vehicle. A tire pressure monitoring system utilizes a tire pressure sensor, which is a small electronic device located on the rim, valve stem, or other portion of the wheel and tire that can detect the air pressure within the tire. This information can be wirelessly transmitted to a receiver in the vehicle and processed by an onboard computer, which then alerts the operator if one of the tires has low air pressure. The alert can be sent to a dashboard on the dashboard of the vehicle. Such tire pressure monitoring systems provide tire pressure information to the operator without requiring the operator to manually check the air pressure of the tires, and allow tire pressure information to be displayed to the driver while the vehicle is in operation. Operating a vehicle with tires that have low inflation pressure results in poor fuel economy, accelerated tire wear, and can cause tire blowouts and vehicle accidents.

[0003] A slow leak in a pneumatic tire is a gradual loss of air from the tire over time, which may be difficult for the operator to detect, unlike a fast leak that the driver can feel while operating the vehicle or that the driver can see when approaching the vehicle. Among other reasons, a slow leak may be caused by a nail piercing the tire, but still within the hole it created and partially sealing the hole, causing a gradual loss of air over time. This slow leak may have occurred for some time before the air pressure in the tire is low enough to alert the tire pressure monitoring system. This delay shortens the amount of time the operator will have to repair the slow leak. The temperature of the tire will also cause changes in tire pressure, and the tire pressure monitoring system may not be able to distinguish between pressure changes caused by leaks and pressure changes caused by temperature. In addition, the tire pressure monitoring system may not be able to properly evaluate other events, such as the intentional removal of air pressure or the reinflation of the tire. Although various types of tire pressure monitoring systems are known to detect air loss in tires, there is still room for change and improvement in the art. Summary of the invention

[0004] The present invention is directed to a system for detecting leakage, comprising:

[0005] a first sensor that obtains pressure and temperature data from a first tire of the vehicle;

[0006] a second sensor that obtains pressure and temperature data from a second tire of the vehicle;

[0007] a third sensor that obtains pressure and temperature data from a third tire of the vehicle;

[0008] a fourth sensor that obtains pressure and temperature data from a fourth tire of the vehicle;

[0009] A processor obtains data from the first sensor, the second sensor, the third sensor, and the fourth sensor and determines whether the first tire is leaking using a natural leakage model and / or a damage comparison model. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] A complete and enabling disclosure of the present invention, including the best mode thereof, to one of ordinary skill in the art is set forth in the specification with reference to the accompanying drawings, wherein:

[0011] Figure 1 is a schematic diagram of a system for detecting tire leaks.

[0012] Figure 2 is a schematic diagram of a system for detecting tire leaks.

[0013] Figure 3 is a graph of the original pressure data and the normalized pressure data.

[0014] Figure 4 Is with Figure 3 Different graphs of data showing the original data and the normalized data with outliers removed.

[0015] Figure 5 is a graph of the tire's daily ratio over time.

[0016] Figure 6 is with Figure 5 A graph of the pressure measured over the same period of time, showing the pressure loss at a specific point in time.

[0017] Figure 7 is a graph of the tire's daily ratio over time.

[0018] Figure 8 is with Figure 7 A graph of the pressures measured over the same time period showing the normalized pressure increase upon a re-inflation event.

[0019] Fig. 9 is a graph showing normalized pressure and filtered pressure of a tire over time, with re-inflation events and rapid leaks evident in the data.

[0020] Fig.10 is a graph of a segmented data set showing normalized pressure over time and predicted natural loss pressure over time.

[0021] Fig.11 is a graph of one data segment showing normalized pressure, natural pressure loss, and prediction interval in the data segment.

[0022] Fig.12 is a graph showing the pressure of a first tire with no leaks over time.

[0023] Fig.13 yes Fig.12 The curve diagram of the cumulative damage model of the first tire, in which the damage rates of the compared tires are not greater than 1, indicating that Fig.12 The first tire was not leaking.

[0024] Fig.14 is a graph of the pressure of the first tire over time, which illustrates leakage of the first tire.

[0025] Fig.15 yes Fig.14 A graph of a cumulative damage model of a first tire, wherein damage rates of two or more of the compared tires are greater than 1, indicating leakage of the first tire.

[0026] The use of the same or similar reference numbers in different drawings indicates the same or similar features. DETAILED DESCRIPTION

[0027] Reference will now be made in detail to embodiments of the present invention, one or more examples of which are illustrated in the accompanying drawings. Each example is provided to illustrate the present invention and is not intended to be limiting of the present invention. For example, a feature shown or described as part of one embodiment may be used with another embodiment to yield a third embodiment. The present invention is intended to include these and other modifications and variations.

[0028] The present invention provides a leak detection system that studies the tires 12, 16, 20, 24 of a vehicle 10 and notifies a user that one or more of the tires 12, 16, 20, 24 is experiencing a leak. The leak detected may be a slow leak, and if the leak is detected at an early stage, the user may be allowed additional time to repair the leak. Therefore, the system attempts to notify the driver early enough in the process of a slow leak so that the driver has enough time to repair the tire 12, 16, 20, 24 at his convenience, rather than being stranded on the road without knowing that the tire 12, 16, 20, 24 is leaking, which is too late. The system utilizes a processor 30 that obtains pressure and temperature data of the tires 12, 16, 20, 24 and processes the information using a natural leak loss model that utilizes the temperature data obtained on the tires 12, 16, 20, 24. The system may additionally or alternatively cause the processor 30 to perform a cumulative damage model for each tire 12, 16, 20, 24 compared to the other tires 12, 16, 20, 24 on the vehicle 10, and based on the modeling, notify the user that the tire 12, 16, 20, 24 is experiencing a leak. Still further, the system may use both the natural leakage model with temperature and the cumulative damage model, and signal a leak only when both models indicate a leak in the tire 12, 16, 20, 24 at a particular time.

[0029] The tire system for vehicle 10 attempts to detect an abnormal rate of pressure loss in any of tires 12, 16, 20, 24 and provide a warning before the pressure reaches a critical level requiring immediate intervention. In addition to this output, the system may also provide the driver with an estimate of the time before the tire 12, 16, 20, 24 needs to be re-inflated, whether due to natural pressure loss of the tire over time or due to an abnormal pressure loss event such as a puncture that may be a slow leak. The system may operate on two time scales, with the first time scale having a long time frame in which small pressure deviations are detected outside of an expected range. The shorter time frame in which the tire system operates is suitable for the detection of rapid pressure events.

[0030] Figure 1An embodiment of a tire system is shown in which a first tire 12 is measured by a first sensor 14 mounted in a rim carrying the first tire 12. The first sensor 14 measures the pressure and temperature in the cavity between the rim and the first tire 12. The information obtained by the first sensor 14 can be transmitted by it to a receiver 28 that can be carried on the vehicle 10. The transmission can be performed by wireless transmission, or in some cases it can be hard-wired transmission. The receiver 28 can then wirelessly transmit its data to a processor 30 and / or a data storage device 36. These elements 30, 36 can be located in the cloud or other areas away from the vehicle and do not need to be carried on the vehicle. The processor 30 can process the information received from the receiver 28 and can use the information to determine whether the tire 12 is leaking. The information from the first sensor 14 can be stored on the data storage device 36, and the processor 30 can use the data in the data storage device 36 to make a leak determination, which data may include not only data from the first sensor 14 but also data from other sensors. Once the processor 30 processes the data, it can return the information to the user in various ways. In one case, the output from the processor 30 can be sent to a vehicle display 32 located in the vehicle 10 so that the operator can receive the information. Additionally or alternatively, the processor 30 can send information to a handheld display 34 such as a cellular phone so that the user can obtain the output from the processor 30.

[0031] In some embodiments, the tire measurement system requires both the air pressure of the tire 12, 16, 20, 24 and the air temperature of the tire 12, 16, 20, 24 to determine if a leak exists. Figure 2A vehicle 10 is shown with a first tire 12, a second tire 16, a third tire 20, and a fourth tire 24, each having their own sensors, labeled first sensor 14, second sensor 18, third sensor 22, and fourth sensor 26. Data from these sensors 14, 18, 22, 26 (the data may be the pressure and temperature of the tires 12, 16, 20, 24) is communicated to a receiver 28, which is located on the vehicle 10 and within range and is capable of transmitting the received data to a data storage device 36 and / or a processor 30. The processor 30 is a computer processor and may use the received data to determine whether one of the tires 12, 16, 20, 24 is leaking. The processor 30 may extract data from the data storage device 36 and / or directly from the receiver 28 for use in making this determination. The processor 30 may have its own internal data storage that allows the processor to know what program to execute in order to process the data. The processor 30 may also obtain data or programs from the data storage device 36 to execute. Once the processor 30 processes the data and determines leak information for the tires 12, 16, 20, 24, the processor may transmit the output to a vehicle display 32 that may be located in the vehicle 10 for display to a user. The data may also be transmitted by a data storage device 36 mounted on the vehicle 10 to a data storage device 40 remote from the vehicle 10 via wireless transmission. The data storage device 40 may simply store data from the tires 12, 16, 20, 24, or may include a processor that may also analyze the data to determine if a leak is occurring. The data storage devices 36, 40 may be computers with a processor and memory and the ability to store information, or may simply be an element that stores information, like a hard drive or a storage device on the cloud.

[0032] Data from the sensors 14, 18, 22, 26 is stored in the data storage device 36 and is obtained by the processor 30, or is sent directly to the processor 30, or manages to reach the processor 30 through a combination of the two. The processor 30 can be a processor onboard the vehicle 10 or can be remote from the vehicle 10. The processor 30 is capable of processing the received data to determine if there is a slow leak in one of the tires 12, 16, 20, 24. The processor 30 can be a computer with long-term and short-term memory and data storage capabilities. Therefore, it should be understood that as used herein, the processor 30 is not only capable of executing instructions according to a program that processes data, but also capable of storing data in short-term memory or long-term storage, such as via a hard drive or other data storage component.

[0033] The data may be pre-processed to clean the data for subsequent processing steps. The first thing the processor 30 may do is to resample the data at a fixed sampling rate. This may be accomplished by setting the sampling rate of the data from the various sensors 14, 18, 22, 26 to the same time interval and using the median of the data within that time variable as the specified reading for that particular sensor 14, 18, 22, 26. This cleanup may be used when the sensors 14, 18, 22, 26 return different numbers for the same time period between them. For example, over the course of one minute, the first sensor 14 may return pressure readings of 2.1 bar, 2.2 bar, 2.4 bar, 2.5 bar, and 2.7 bar. The processor 30 may determine the median of the data set to be 2.4 bar and set the reading of the first sensor 14 to 2.4 bar / minute. The temperature reading of the first sensor 14 may be processed in a similar manner to produce a data set of the first sensor 14 temperature / minute. The pressure and temperature data measured from the other sensors 18, 22, 26 may be resampled and set to the median of the data within each minute so that all sensors 14, 18, 22, 26 produce data sets within the same amount of time. Resampling the data in this manner improves the efficiency of data analysis. Figure 3 is a graph showing raw pressure data from one of the sensors over time and normalized pressure data after such normalization.

[0034] Most of the daily variation in pressure within the tires 12, 16, 20, 24 is due to temperature fluctuations from day to night and from when the vehicle 10 is moving to when the vehicle is stationary. The temperature readings of the sensors 14, 18, 22, 26 are used in the following formula, where T cold At 17 degrees Celsius:

[0035]

[0036] For example, if the data returned is a measured temperature of 25 degrees Celsius (T measured ) and the measured pressure returned is 2250 mbar (P measured ), then P cold will be calculated as 2250 mbar*((17+273.15) / (25+273.15))=2250 mbar*(290.15 / 298.15)=2250 mbar*0.973=2189.6 mbar. The benefit of normalizing the temperature based on the ideal gas law is that it does not require a priori knowledge about the measurements. The disadvantage of this temperature normalization is that it ignores the moisture content in the air and changes in the position of the sensors 14, 18, 22, 26 in the tires 12, 16, 20, 24, which may change the apparent sensitivity to temperature. The resampled pressure in the previous step may be used for P measuredThe pressure, and in some cases, the processor 30 can also resample the temperature in the same manner as the pressure resample, and the resampled temperature can be used in the formula T measured .

[0037] The processor 30 may then perform a processing step in which outliers are discarded. Outliers may be discarded in a number of ways, and one such way may involve removing data that is more than 0.1 bar away from a data point on either side to achieve a confidence interval of approximately 99%. If such an outlier is the first or most recent data point in the series, it is not discarded. Figure 4 shows a similar Figure 3 The data shown in the data set are different from the data set, but in Figure 4 The results of the foregoing preprocessing steps performed by the processor 30 provide a data set for each sensor 14, 18, 22, 26 that is cleaner than the raw data and has a standard deviation of approximately half the standard deviation of the raw data on any given day. In some embodiments, the standard deviation may drop from about 40 millibars for the unpreprocessed data set to about 20 millibars for the filtered data set.

[0038] The data set provided by the sensors 14, 18, 22, 26 may then be processed to determine if one of the tires 12, 16, 20, 24 is experiencing a fast leak and / or a slow leak. However, the processor 30 must be able to ignore certain situations within the data set that may mask or obscure a fast and / or slow leak. For example, the tires 12, 16, 20, 24 may experience a natural leak of air from the inside through the inner liner, which is slow in the winter and faster in the summer. The tires experiencing natural air leaks are consistent for each of the tires 12, 16, 20, 24. In addition, re-inflation of the tires 12, 16, 20, 24 may be performed by the owner and may be displayed in the data. Re-inflation of the tires may result in an increase in pressure from day to day, and the re-inflation is generally consistent in trend, although not in magnitude, for each wheel position. In addition, the pressure data set may have deflation of the tires 12, 16, 20, 24 therein, which may be due to a pressure check of the tires 12, 16, 20, 24 or by manual tire deflation. Deflation typically (although not always) affects all tires 12, 16, 20, 24 on the vehicle 10 at the same time and may result in false positives of leak detection. Finally, the pressure data set may show leaks of the tires 12, 16, 20, 24, which are abnormal pressure changes but are not reflected in all wheel positions. The shape of the leak in the graph is typically linear, exponentially decreasing, or exponentially increasing. The processor 30 needs to identify the leak and needs to ignore the data in the data set due to natural leaks, re-inflation, or deflation.

[0039] The processor 30 may then process the data set to determine if one of the tires 12, 16, 20, 24 has a rapid leak. The processor 30 may calculate a daily pressure ratio comparing the currently measured pressure to the median pressure of the previous day by using the following formula:

[0040]

[0041] Here, P(t) is the pressure measured at time t, Δt is the change in time, and in one example is a time of 1 day. [P(t-Δt); P(t)] are all pressures measured between time t-Δt and t. For example, in the last day, if the first sensor 14 measures pressure readings of 2.40 bar, 2.42 bar, 2.40 bar, 2.5 bar, 2.5 bar and 2.5 bar, the median will be 2.45 bar. If the P(t) measured at the time point considered is 2.5 bar, then r=2.5 bar / 2.45 bar=1.02. The daily ratio r of tires with little or no leakage is between 0.95 and 1.05. Thus, when the daily ratio of two consecutive data points drops below the detection limit, a warning is triggered for a rapid leak. In some embodiments, the processor 30 may return a rapid leak warning when the r value is lower than 0.90 for two consecutive data points. Figure 5 The calculated r-values ​​for a sample data set of the first sensor 14 over the course of several months are shown. In July 2019, the r-value dropped below 0.90 for two consecutive data points, indicating that the first tire 12 was leaking rapidly. Figure 6 The pressure sensed by the first tire 12 during this same time period is shown, which verifies that a pressure loss is occurring. Figure 5 and Figure 6 It is for the first sensor 14 and the first tire 12 , and data for the other sensors 18 , 22 , 26 and tires 16 , 20 , 24 are not shown.

[0042] In addition to detecting fast leaks of tires 12, 16, 20, 24, the processor 30 can also analyze the data set to determine whether the tire is experiencing a slow leak. In order to detect whether the tire 12, 16, 20, 24 is experiencing a slow leak, the processor utilizes two different schemes and combines them to achieve robust detection. The first method is to compare the measured pressure with the pressure predicted by the physical model of natural pressure loss over time. The pressure loss rate is a function of the current pressure and temperature in the tire 12, 16, 20, 24. The second method is to compare the pressure loss rate of each tire, such as tire 12, to identify when the pressure loss of the tire 12 becomes significantly different from the other tires 16, 20, 24. The second method can also be performed on the other tires 16, 20, 24 compared to the other three tires on the vehicle 10. The processor 30 can use these two methods to determine that the tire 12, 16, 20, 24 is indeed leaking.

[0043] In order to perform a determination of the measured pressure versus the pressure predicted by the physical model of natural pressure loss over time, the processor 30 may first remove any fast leak data as previously calculated from the data set. The data set may be segmented, where if the delta time between two data points is greater than 10 days, the data becomes a new segment group. Also, if reinflation of the tire 12, 16, 20, 24 is detected, the data becomes a new segment group. As previously discussed, reinflation occurs when someone fills the tire with air or nitrogen. In this step, the processor 30 again looks at the previously calculated day-by-day ratio r to detect reinflation. When the continuous value rises above the inflation limit, the tire 12, 16, 20, 24 is reinflated, which in one embodiment may be set to 1.065. If the data set is provided so that there is no data from the previous day, the processor 30 may utilize additional rules to detect reinflation. Here, if the pressure difference between two consecutive data points is higher than 1.25 bar, the processor 30 detects reinflation. Figure 7 The calculated daily ratio r over time is shown, and a horizontal re-inflation line of 1.065 bar is shown in the graph. As shown, the daily ratio r over time exceeds the re-inflation line of 1.065 bar three times on the graph, indicating that re-inflation is occurring in the tire 12 at these three times. Figure 8 Shows Figure 7 The normalized pressure in millibars of the tire 12 over time allows the pressure readings to be viewed to verify that re-inflation is actually occurring at these three times. As shown, the normalized pressure is Figure 7 There is an increase at the three time points marked on the graph, indicating that re-inflation has occurred at these moments.

[0044] Next, with fast leak data removed and the data set segmented into groups based on refill, processor 30 may process the normalized pressure in each of these segments. Fig. 9 New data that is different from previous data is included to illustrate this next step in the process. Fig. 9 The processor takes the normalized pressure and calculates a rolling average, then Fig. 9 This average for each day in is designated as the filtered pressure. Taking a rolling average of the normalized pressure reduces the noise in the data set and produces the following results when comparing the normalized pressure to the filtered pressure: Fig. 9 . To calculate a rolling average to obtain the filtration pressure, the processor 30 may take all the normalized pressures for the previous 24 hour period for any one day data point and average them together, and that average becomes the filtration pressure for that one day data point. For example, if the processor 30 is trying to determine the filtration pressure for 1 / 11 / 2018, the processor takes the normalized pressure readings from the previous 24 hour period, which may be 1.65 mbar, 1.65 mbar, 1.70 mbar, 1.71 mbar, 1.66 mbar, and 1.68 mbar, and the average of those six readings is 1.67 mbar. On the chart, 1.67 mbar is then designated as the filtration pressure for 1 / 11 / 2018. Fig. 9 4 segmented data sets are indicated in FIG. 1 , and there is a rapid leak between data sets 2 and 3. The rapid leak data is removed from the process. Reinflation of the tire 12 occurs between data sets 1 and 2, and also occurs between data sets 3 and 4.

[0045] Now that the processor 30 has established a cleaned and segmented data set, the processor 30 can apply a natural leakage rate model to estimate how the tire 12 is experiencing natural leakage changes under environmental conditions. The tire 12 will naturally lose air pressure over time due to the normal occurrence of air permeating through the tire 12 and expelling. This natural loss of air over time is not to be confused with a slow leak of the tire 12, but instead is air loss that occurs through the inner liner of the tire 12. In some cases, the natural air loss rate may be a rate of 1 to 5 millibars of pressure per day. This natural air loss rate depends on the temperature and pressure of the tire 12. To account for this dependence on temperature and pressure, the processor 30 utilizes a functional exponential model that takes into account the thermomechanical functions of the tire 12 and the physicochemical mechanism of diffusion-absorption of air components of the enclosure cavity. Using this natural leakage model, the processor 30 can ascertain the pressure at any given time taking into account the natural pressure loss. The natural leakage model uses the following formula:

[0046] Ppredicted(t)P(t-△ +)*(1-α(t)*△ t )

[0047] in

[0048] Here, P predicted (t) is the pressure at a given time. P(t-Δ t ) is the pressure at the previous data point, where Δ t is the delta time between the new data point and the previous data point. Normalized Pressure is the pressure value used. T(t) is the temperature at a given time. E and τ0 are constants that depend on the compositional properties of the wall of the casing of the tire 12, and in other words, are constants that depend on the geometry of the tire 12 and the permeability of the material that constitutes the tire 12. These parameters are tire 12 dependent, but the values ​​of E=5800K (E must be between 5000K and 6000K to approximate the activation energy of the inner liner of the tire 12) and τ0=1 / 3 are good approximations for a wide range of standard tire designs.

[0049] Fig.10 A series of segmented data sets are shown, where normalized pressure is displayed as a function of time. Fig.10 The natural loss pressure (P predicted (t)) and leads to a much tighter fit than the normalized pressure. P predicted The calculation of (t) can be illustrated by example. The first sensor 14 measuring the first tire 12 provides both temperature and pressure readings at each time. A segmented data set in the segmented data set, such as Figure 1 Segmented data set #1 is shown, and its data points can be converted from normalized pressure to natural loss pressure as shown. At a given time point in the first data set, the first sensor 14 provides T(t) for that particular given time point, and in this example, T(t) can be 32 degrees Celsius. The parameters E = 5800K (equivalent to 5526.8°C) and τ0 = 1 / 3 are set as described above. Substituting these numbers into the formula for the pressure loss rate α(t) above, the formula is written as: α(t) = 1 / (1 / 3)*e -(5526.85 / (32+273.15)) =3*e -18.11 =3*1.36435×10 -8 =4.093×10 -8 .

[0050] When the pressure loss rate is calculated, Δ t is 1 minute, which is the time between the readings of the sensor 14. The pressure one minute before the pressure point being calculated is the previous pressure, and it is used as P(t-Δ t ) is included in the formula and is 2220 mbar in the example. Solving for P predicted(t) = 2220 mbar * (1-4.093 x 10-8 * 1) = 2220 mbar * (0.9999999591) = 2219 mbar. The natural loss pressure of all data points can be calculated for all data segments, and this is Fig.10 is shown for each of these segments. Fig.10 The data in is the normalized pressure and the adjusted natural loss pressure of one of the tires 12 , and these variables are not shown for each of the tires 12 , 16 , 20 , 24 of the vehicle 10 .

[0051] In calculating P predicted (t), the natural pressure loss of each segment can be looked at to determine if the tire 12 is experiencing a slow leak, and the leak detection is a comparison of the first sensor 14 data with the model. A 99% prediction interval is established around the pressure predicted by the natural pressure loss model. The 99% interval is established by the upper and lower bounds using the following formula:

[0052] IP 99％ (t) = [P predicted (t)-2.576*σ;P predicted (t)+2.576*σ]

[0053] In order to eliminate noise outside of this interval, the standard deviation σ may be overestimated, so the processor 30 may set the standard deviation σ to be equal to 40 millibars. In other embodiments, the standard deviation σ may be different. This is the standard deviation of the uncorrected pressure measured by the sensor 14. The prediction interval allows the model to be readjusted. In this regard, if 20 consecutive data points are outside the top of the interval, the processor 30 will restart the model to shift the prediction interval back to the actual inflation pressure. The second point that the prediction model allows is for the processor 30 to predict whether the tire is losing pressure faster than expected. In this regard, if 20 consecutive data points are outside the bottom of the interval, the processor 30 detects an area of ​​potential leakage at that point in the algorithm. Although 20 consecutive data points are given as an example, any number may be used in other embodiments. If P calculated by a particular sensor 14 for a particular time predicted (t) is 2219 mbar, then IP 99％ (t) = [2219 - (2.576 * 40); 2219 + (2.576 * 40)] = [2219 - 103.04; 2219 + 103.04] = 2115.96 mbar; i.e. 2322.40 mbar. The bottom of the interval at this time point will be 2115.96 mbar, and the top of the interval at this time point will be 2322.40 mbar. Fig.11 is a graph of a data segment as previously discussed, but with new data not previously shown. Fig.11 For a segment of measured data, pressure is on the Y-axis and time is on the X-axis. Normalized pressure, for example relative to Fig. 9 The discussion calculates those pressure readings and the rolling average, plotted on a graph over time. predicted (t) Natural loss pressure is also plotted on the graph over time. IP with upper and lower bounds 99％ (t) The forecast interval is also plotted on the graph over time as shown. The normalized pressure is within the forecast interval until about March 2019, when it drops below the lower limit of the forecast interval and is out of range. The normalized pressure below the lower limit of the forecast interval is Fig.11 If the normalized pressure is lower than the lower limit, the processor 30 detects a leak in the tire 12. Fig.11 The graph of is for only one sensor 14 of the vehicle 10, and is only one data segment for that sensor 14 of the vehicle 10. The other tires 16, 20, 24 and their data segments may be evaluated in a similar manner to determine if they are also experiencing a slow leak.

[0054] The above actions performed by the processor 30 result in the use of a natural leak model to predict slow leaks from the tires 12, 16, 20, 24. However, if the model has a poor starting point, the use of the slow leak model may cause the processor 30 to erroneously detect leaks. In order to avoid false alarms from the vehicle 10 monitoring system, the natural leak model should be combined with a comparison of adjacent tires 12, 16, 20, 24 on the vehicle 10. In this regard, the first tire 12 will be compared to the other three tires 26, 20, 24 on the vehicle 10. The second tire 16 will be compared to tires 12, 20, 24, and the third tire 20 will be compared to tires 12, 16, and 24. Moreover, the fourth tire 24 will be compared to tires 12, 16, 20.

[0055] In order to compare tires 12, 16, 20, 24, the velocity of any of tires 12, 16, 20, 24 at any time should be calculated. To calculate this velocity, the pressure loss is assumed to be linear and the following formula is used:

[0056]

[0057] rate(t) is the rate at any given time, and P(t) is the pressure at a given time. thresh is the pressure threshold and is a constant. In the case of passenger car tires, P thresh Can be set to 1500 mbar. Variable N days (t) is the time until the tire reaches P threshEstimated number of days.

[0058] To solve for rate(t), the previously determined day-to-day ratio r can be used, and the term 1-r(t) expresses the percentage change in pressure over the course of a day. The term P(t) is the pressure at a given time. P(t) is the actual measured pressure, and r(t) is the day-to-day ratio r as calculated previously. In this way, rate(t) can be expressed as follows:

[0059] rate(t)=P(t)*(1-r(t))

[0060] In one example, if the pressure at time t is 1800 mbar, and the previously calculated daily ratio r is 0.99, then rate(t) can be calculated as 1800 mbar*(1-0.99)=1800 mbar*(0.01)=18. Solving for rate(t) and substituting P thresh Setting this to a constant allows processor 30 to use the previous formula to solve for N as follows days (t):

[0061]

[0062] In this regard, processor 30 may then calculate the time required to reach P at any given time. thresh The remaining number of days required (N days (t)). For example, if rate(t) is 18, and if P thresh is 1500 mbar, and if the pressure at time t is 1800 mbar, N days (t)=(1800-1500) / 18=300 / 18=16.66.

[0063] Thus, the processor 30 is able to solve for the rate(t) of each of the tires 12, 16, 20, 24 at various times, typically once per minute when the vehicle 10 is operating. No measurements may be taken when the vehicle 10 is not operating. The processor 30 may then calculate the rate difference between each tire 12, 16, 20, 24 by using the following formula:

[0064] diff i (t) = rate est (t)-rate i (t)

[0065] In this calculation, rate est (t) is the rate (t) of the tire under consideration, such as the first tire 12, and rate i(t) will be the rate(t) of the second tire 16 to which the first tire 12 is compared. These rates are those calculated as previously. For example, if the rate of the first tire 12 at a particular time t is 18, and the rate of the second tire 16 at the same time t is 17.5, then the rate difference between the first tire 12 and the second tire 16 at time t is diff i (t) = 18 - 17.5 = 0.5. This velocity difference is calculated for all measurements made in the last 5 days for the first tire 12 versus the second tire 16. The processor 30 will also perform this same velocity difference calculation for each measurement made in the last 5 days for the first tire 12 versus the third tire 20, and will perform this same velocity difference calculation for each measurement made in the last 5 days for the first tire 12 versus the fourth tire 24. Then, only the diff of the first tire 12 versus the second tire 16 is used. i (t) data, the processor 30 calculates the cumulative damage model for these rate differences in the last 5 days using the following formula:

[0066]

[0067] In this formula, c is equal to 1 / 10 and is a constant, and a is also a constant and is equal to 0.4. i (t) is one of the previously calculated velocity differences of the first tire 12 versus the second tire 16, and N days (t) is the previously calculated number of days remaining to reach the time threshold at time t. For example, if diffi(t) is 0.5 and Ndays(t) is 243 days, the calculation would be (1 / 10)*(0.5 / 243) 0.4 =(1 / 10)*(0.00205761) 0.4 =(1 / 10)*(0.08420642)=0.00842064. The processor 30 will then move to the next time, which is usually 1 minute, and use the diff of the next time i (t) and N days (t) value Perform the same calculation again to get another number. Calculate all times in the last 5 days and their corresponding values ​​diff i (t) and N days(t). All of these numbers are added together to get D for the first tire 12 compared to the second tire 16. The system then compares the cumulative damage D to a cumulative rate limit to see if D exceeds the cumulative rate limit. In some embodiments, the cumulative rate limit is 1. If D is greater than 1, then there is a significant difference between the first tire 12 and the second tire 16. Although described as being summed over the last 5 days, in other embodiments, the summation may be performed over at least the last day, at least the last 2 days, at least the last 3 days, at least the last 4 days, at least the last 5 days, or over the last 5-10 days, according to different exemplary embodiments.

[0068] The processor 30 may then run the same calculation to determine the D of the first tire 12 relative to the third tire 20. In this regard, the diff used i (t) is the velocity difference between the first tire 12 and the third tire 20 at time t, and N days (t) is the number of days that the first tire 12 reaches the threshold at time t. Enter these values ​​into the above formula and get a number. Use the diff at the next time t i (t) and N days (t) The calculation is repeated for this time t, and likewise again until all measurements made within the last 5 days are tabulated. All of these numbers are added together to get D for the first tire 12 relative to the third tire 20. This number D is then compared to the cumulative rate limit set for the first tire 12 relative to the third tire 20. In some cases, the cumulative rate limit may be 1. If D is greater than 1, a significant difference between the first tire 12 and the third tire 20 is noted.

[0069] The processor 30 may repeat this process using the values ​​for the first tire 12 versus the fourth tire 24 to obtain D for the first tire 12 relative to the fourth tire 24. Again, if this particular D is greater than the cumulative rate limit for the first tire 12 and the fourth tire 24 (which may be 1 in some cases), a significant difference between the first tire 12 and the fourth tire 24 is noted.

[0070] The previously mentioned calculation D is a cumulative damage model that determines the damage inflicted on a material. As physical events follow one another, the ability of a physical system to resist physical events decreases and the damage incurred increases. In this regard, the cumulative damage model defines damage as the number of repetitions of an event that damages an object over a certain period of time. In the described system, the event damage is the ratio r difference between the tire under consideration and its neighbor. If the cumulative rate difference damage D crosses a value of 1, the processor 30 deems that sufficient event damage exists to justify a significant difference between the two tires.

[0071] The cumulative rate differential damage D is calculated as previously described at each time t for the first tire 12 versus the second tire 16. The same process allows the cumulative rate differential damage D to be calculated for the first tire 12 versus the third tire 20 and for the first tire 12 versus the fourth tire 24. As the processor 30 moves through the adjacent tire comparisons, if the cumulative rate differential damage D for at least two of the adjacent tires 16, 20, 24 is greater than 1, then the processor 30 will detect a leak in the first tire 12. In this regard, if D for the first tire 12 versus the second tire 16 is 1.1, and D for the first tire 12 versus the third tire 20 is 0.03, and D for the first tire 12 versus the fourth tire 24 is 0.05, then only one of the cumulative rate differential damages D is greater than 1, and thus the processor 30 does not detect a leak in the first tire 12. However, if the same numbers were unchanged except for the fact that the cumulative rate differential damage D for the first tire 12 versus the fourth tire 24 was 1.2, it would be the case that 2 of the D values ​​were greater than 1, and in this case the processor 30 would indicate that the first tire 12 was leaking under this cumulative damage model.

[0072] Another example of the processor 30 detecting a leak via a cumulative damage model is presented below, where the previous process is performed for the first tire 12 relative to the second tire 16 , the third tire 20 , and the fourth tire 24 . Fig.12 An example of filtered pressure versus time is shown for the first tire 12, where the pressure has not dropped below 1500 millibars, and therefore the first tire 12 is not leaking. The pressure does tend to trend downward over time, which may indicate a natural loss of pressure in the first tire 12 or may indicate a slow leak in the tire 12. However, using the cumulative rate differential damage D of the first tire 12 relative to the second tire 16, the third tire 20, and the fourth tire 24, no such leak is detected. Fig.13 Shown in Fig.12 The cumulative rate differential damage D for the three tires 16 , 20 , 24 is the same amount of time, and none of the D values ​​exceeds 1. Since none of the D values ​​exceeds 1, the processor 30 indicates that the first tire 12 is not leaking.

[0073] Fig.14 Another example is shown in FIG. 1 , where filter pressure versus time for a first tire 12 is again presented, only this time there is a leak in the first tire 12 . Fig.14 According to the data presented in FIG. 1 , around March 2019, the first tire 12 started to lose air pressure, and in April 2019, the pressure dropped below the threshold of 1500 mbar. Fig.15, the cumulative damage D of the first tire 12 compared to the adjacent tires 16, 20, 24 is shown. The cumulative rate difference damage D of the fourth tire 24 reaches a value of 1 in about October 2018, but the damage D of the second tire 16 and the third tire 20 do not reach this value, so the processor 30 does not indicate a leak in the first tire 12 because only one tire, not two, of the adjacent tires reaches a value of 1. Turning to the data from about March 2019, it can be seen that two or even three of the damage D values ​​exceed 1, which means that the processor 30 detects a leak in the first tire 12 at this time. Fig.14 The pressure of the first tire 12 in FIG. 1 is not lower than the 1500 mbar threshold in March 2019, so the detected leak may be a slow leak of the first tire 12. In April 2019, the cumulative damage D of the first tire 12 compared with all three other tires 16, 20, 24 is greater than 1, which also means that the processor 30 has detected a leak in the first tire 12 under the adjacent tire comparison model. Fig.14 The filter pressure of the first tire 12 is shown to be below the 1500 mbar threshold in April 2019, verifying that the first tire 12 is in fact leaking.

[0074] The processor 30 may also repeat this same analysis for the second tire 16, where the processor may perform an adjacent tire damage comparison for the second tire 16 relative to the first tire 12, the third tire 20, and the fourth tire 24 to see if the second tire 16 is leaking under this model. In addition, the processor 30 may analyze the third tire 20 relative to the adjacent first tire 12, the second tire 16, and the fourth tire 24 to determine if the third tire 20 is leaking under this particular damage model. Finally, the fourth tire 24 may be looked at to determine if it is leaking by comparing it to the first tire 12, the second tire 16, and the third tire 20 under the cumulative damage model discussed previously. Thus, the processor 30 may be able to check all four tires 12, 18, 20, 24 to determine if the tire is leaking relative to the adjacent tires of any of them based on this adjacent tire comparison analysis. Although described as requiring at least two of the tires 18, 20, 24 to cross the cumulative rate difference line of 1 in order to detect a leak, in other embodiments, only at least one of the tires 18, 20, 24 needs to cross the cumulative rate difference line 1 in order for the processor 30 to signal a leak in the first tire 18, and in other embodiments, all three tires 18, 20, 24 must cross the cumulative rate difference line 1 in order for the processor 30 to signal a leak in the first tire 18.

[0075] The system of the present invention can alert the operator of the vehicle whether there is a tire leak by comparing the tire to other tires on the vehicle through a damage model, or through the discussed natural leakage model using temperature data, or through a combination of these two methods. In this regard, it is possible that these two methods can be combined in order to minimize the false alarm rate. The processor 30 can only identify the leak when the processor 30 determines that one of the tires 12, 18, 20, 24 is leaking under the damage model tire comparison and by the natural leakage model using temperature data at the same time t. If a leak is detected by one of these methods but not both methods at the same time t, the processor 30 does not signal that a leak is occurring. However, there are other embodiments in which the processor 30 can signal a leak when a leak is detected by the natural leakage model using temperature data or by the damage model tire comparison. When the processor 30 determines that a leak is occurring, the information can be sent to the vehicle display 32 and / or the handheld display 34 to alert the user of the vehicle 10 that one or more of the tires 12, 16, 20, 24 has a leak.

[0076] Although the subject matter of the present invention has been described in detail with respect to specific embodiments and methods thereof, it should be understood that, after understanding the foregoing, those skilled in the art may easily conceive of modifications, variations, and equivalents of such embodiments. Therefore, the scope of the present disclosure is by way of example rather than by way of limitation, and the present disclosure does not exclude the inclusion of such obvious modifications, variations, and / or additions to the subject matter.

Claims

1. A system for detecting a leak, comprising: a first sensor that obtains pressure and temperature data from a first tire of the vehicle; a second sensor that obtains pressure and temperature data from a second tire of the vehicle; a third sensor that obtains pressure and temperature data from a third tire of the vehicle; a fourth sensor that obtains pressure and temperature data from a fourth tire of the vehicle; a processor that obtains data from the first sensor, the second sensor, the third sensor, and the fourth sensor and uses a natural leakage model and / or a damage comparison model to determine whether the first tire is leaking, wherein: The natural leakage model includes: wherein if not all pressure measurements of the first sensor are ready at a fixed sampling rate, the processor resamples the pressure measurements at the fixed sampling rate, and wherein the processor calculates a set of normalized pressure readings at different times; wherein the processor calculates a predicted pressure of the first tire at the specific time by using the normalized pressure at a previous time, a time difference between a specific time and the previous time, and a temperature measurement of the first sensor at the specific time; wherein the processor calculates a prediction interval using the predicted pressure, and if the normalized pressure at the specific time drops below the prediction interval at the specific time, the processor detects a leak in the first tire under the natural leakage model, even if a sufficient number of future measured normalized pressures exceed the normalized pressure at the specific time, the processor detects a leak in the first tire under the natural leakage model; and The damage comparison model includes: wherein the processor calculates a daily pressure ratio (r(t)) of the first tire at different times, a daily pressure ratio of the second tire at different times, a daily pressure ratio of the third tire at different times, and a daily pressure ratio of the fourth tire at different times; wherein the processor calculates rate(t) of the first tire by the formula rate(t)=P(t)*(1-r(t)), wherein P(t) is the pressure of the first tire at time t, and r(t) is the day-to-day ratio r of the pressure of the first tire at the time t; wherein the processor calculates rate(t) of the second tire by the formula rate(t)=P(t)*(1-r(t)), wherein P(t) is the pressure of the second tire at time t, and r(t) is the day-to-day ratio r of the pressure of the second tire at the time t; wherein the processor calculates rate(t) of the third tire by the formula rate(t)=P(t)*(1-r(t)), wherein P(t) is the pressure of the third tire at time t, and r(t) is the daily ratio r of the pressure of the third tire at the time t; wherein the processor calculates rate(t) of the fourth tire by the formula rate(t)=P(t)*(1-r(t)), wherein P(t) is the pressure of the fourth tire at time t, and r(t) is the daily ratio r of the pressure of the fourth tire at the time t; The processor passes the formula N days (t)=(P(t)–P thresh ) / rate(t) to calculate the estimated number of days (N) from the first tire at different times until the first tire reaches the pressure threshold. days (t)), where Pthresh is a constant, P(t) is the pressure of the first tire at time t, and rate(t) is the previously calculated rate(t) of the first tire; The processor uses the formula diff i (t) = rate est (t)–rate i (t) calculating the speed difference between the first tire and the second tire at the different times, where diff i (t) is the velocity difference between the first tire and the second tire at time t, rate est (t) is the previously calculated rate(t) of the first tire at time t, and rate i (t) is the rate(t) of the second tire calculated previously; The processor uses the formula diff i (t) = rate est (t)–rate i (t) calculating the speed difference between the first tire and the third tire at the different times, where diff i (t) is the velocity difference between the first tire and the third tire at time t, rate est (t) is the previously calculated rate(t) of the first tire at time t, and rate i (t) is the rate(t) of the third tire calculated previously; The processor uses the formula diff i (t) = rate est (t)–rate i (t) calculating the speed difference between the first tire and the fourth tire at the different times, where diff i (t) is the speed difference between the first tire and the fourth tire at time t, rate est (t) is the previously calculated rate(t) of the first tire at time t, and rate i (t) is the rate(t) of the fourth tire calculated previously; The processor calculates the cumulative damage of the first tire relative to the second tire by the following formula: where diff i (t) is the previously calculated diff between the first tire and the second tire at time t i (t) Rate difference, N days (t) is the previously calculated estimated number of days (N) for the first tire at time t days (t)), where c and a are constants, and where the summation is performed over a period of at least the most recent day; The processor calculates the cumulative damage of the first tire relative to the third tire by the following formula: where diff i (t) is the previously calculated diff between the first tire and the third tire at time t i (t) Rate difference, N days (t) is the previously calculated estimated number of days (N) for the first tire at time t days (t)), where c and a are constants, and where the summation is performed over a period of at least the most recent day; The processor calculates the cumulative damage of the first tire relative to the fourth tire by the following formula: where diff i (t) is the previously calculated diff between the first tire and the fourth tire at time t i (t) Rate difference, N days (t) is the previously calculated estimated number of days (N) for the first tire at time t days (t)), where c and a are constants, and The summation is performed over at least the most recent day; wherein the processor compares the cumulative damage D of the first tire relative to the second tire, the cumulative damage D of the first tire relative to the third tire, and the cumulative damage D of the first tire relative to the fourth tire with a cumulative rate limit to detect leakage of the first tire under the damage comparison model; and A display outputs leaks of the first tire detected by the processor.

2. The system according to claim 1, wherein the processor outputs the detected leakage of the first tire only when the processor detects the leakage of the first tire at the specific time under the natural leakage model and also detects the leakage of the first tire at the same specific time under the damage comparison model.

3. The system of claim 1, wherein calculating the sum of the cumulative damage of the first tire relative to the second tire is performed over a period of the last 5 days; wherein calculating the sum of the cumulative damage of the first tire relative to the third tire is performed within the last 5 days; The calculation of the sum of the cumulative damage of the first tire relative to the fourth tire is performed within the last 5 days.

4. The system of claim 1, wherein the processor calculates the cumulative damage of the first tire relative to the second tire by setting the constant c to 1 / 10 and by setting the constant a to 0.4; wherein the processor calculates the cumulative damage of the first tire relative to the third tire by setting the constant c to 1 / 10 and by setting the constant a to 0.4; and wherein the processor calculates the cumulative damage of the first tire relative to the fourth tire by setting the constant c to 1 / 10 and by setting the constant a to 0.

4.

5. The system of claim 1 , wherein the processor detects leakage of the first tire under the damage comparison model if at least two of the cumulative damage D of the first tire relative to the second tire, the cumulative damage D of the first tire relative to the third tire, and the cumulative damage D of the first tire relative to the fourth tire are greater than 1.

6. The system of claim 1 , wherein the pressure measurements of the first sensor are not at a fixed sampling rate, and the processor resamples the pressure measurements of the first sensor by calculating a median of the pressure measurements of the first sensor within one minute time increments.

7. The system of claim 1 , wherein in the natural leak model, the processor calculates the set of normalized pressure readings by the following formula: Where T measured is the temperature measured by the first sensor at the specific time, where P measured is the pressure measured by the first sensor at the particular time, and where T cold It's 17 degrees Celsius.

8. The system of claim 7, wherein the processor removes outliers from the set of normalized pressure readings by discarding a normalized pressure reading if the normalized pressure reading is greater than 0.1 bar in magnitude than a previous or subsequent normalized pressure reading.

9. The system of claim 1 , wherein in the natural leak model, the processor segments the normalized pressure readings from the first sensor by removing normalized pressure readings indicating a rapid air pressure leak of the first tire and starting a new segment when the normalized pressure readings indicate re-inflation of the first tire.

10. The system of claim 9, wherein the processor filters the normalized pressure readings in the segment from the first sensor by calculating a rolling average over a prior period and designating the rolling average as the filtered pressure at the particular time.

11. The system of claim 10, wherein the rolling average is an average of all of the normalized pressures from the first sensor over a 24 hour period prior to the particular time.

12. The system of claim 1, wherein the processor calculates the predicted pressure by the following formula: Ppredicted(t)=P(t-Δ t )*(1-α(t)*Δ t ) in Where T(t) is the temperature of the first tire, and P(t-Δ t ) is the pressure of the first tire at the previous data point, where Δ t is the time between the readings of the first sensor, where E and τ0 are constants.

13. The system of claim 12, wherein E is 5800 Kelvin, and wherein τ0 is 1 / 3.

14. The system of claim 1, wherein in the natural leakage model, the processor calculates the prediction interval using the following formula: IP 99％ (t)=[P predicted (t)-2.576*σ;P redicted (t)+2.576*σ] where σ is 40 mbar and where if for P predicted At least 20 consecutive data points, P predicted Drop below IP 99％ (t) is less than the lower limit, the processor detects the leakage of the first tire.

15. The system of claim 1, wherein the display is built into the vehicle, and wherein the processor is remote from and not engaged with the vehicle.

Images