An Automotive Transmission Condition Monitoring Method with Adaptive Energy Growth Sparsity Measure

Abstract

A method for monitoring the state of an automobile transmission using adaptive energy growth sparsity measurement includes constructing state monitoring indicators for each gear position, constructing indicators for monitoring the overall health of the transmission, and performing state monitoring and fault location for the transmission. The method first obtains a vibration signal of the transmission and constructs state monitoring indicators for each gear position to achieve adaptive energy growth sparsity assessment. Weighted fusion is then performed to obtain monitoring indicators that can reflect the overall health of the transmission. Fault location is then performed using the monitoring indicators and the results of the importance assessment of segmented amplitudes and growth rates during the calculation process. The method implements automatic state monitoring and auxiliary fault location for the automobile transmission, thereby improving the accuracy of automobile transmission state monitoring.

Description

Technical field:

[0001] The present invention belongs to the technical field of automobile transmission state monitoring, and in particular relates to an automobile transmission state monitoring method based on adaptive energy growth sparsity measurement. Background technology:

[0002] The automotive transmission is an essential component, and its reliability impacts the vehicle's overall performance. During operation, transmissions frequently experience gear shifts, causing shock or load changes. Effective condition monitoring of transmissions is difficult due to their structural complexity and variability in operating conditions.

[0003] The gearbox is the main transmission component of the automobile transmission. Common failure modes include tooth surface wear, tooth surface pitting, and gear tooth breakage. The commonly used transmission status monitoring method (Jin Guang, Yuan Zhaodan, Jiang Guanyi, et al. Early fault diagnosis of transmission assembly durability test [J]. Automobile Technology, 2019(06):53-58.) is to observe the gear meshing order and its sideband components in the vibration signal order spectrum to construct monitoring indicators. Due to the complex transmission structure of the transmission and the existence of transmission path changes, the status monitoring of the transmission has the disadvantages of order spectrum differences between different devices, discontinuous monitoring indicators, and difficulty in setting failure thresholds. Therefore, constructing a monitoring indicator that can reflect the degradation of the health status of the transmission, realizing automatic status monitoring of the transmission, and further obtaining the location of the faulty gear is of great significance to the maintenance of automobile transmissions. Summary of the invention:

[0004] In order to overcome the shortcomings of the above-mentioned prior art, the purpose of the present invention is to provide an automobile transmission state monitoring method based on adaptive energy growth sparsity measurement, which realizes automatic monitoring of the automobile transmission state and auxiliary fault location, and improves the accuracy of automobile transmission state monitoring.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0006] A method for monitoring the condition of an automobile transmission using adaptive energy growth sparsity measurement is proposed. The method first acquires the transmission's vibration signal and constructs condition monitoring indicators for each gear to achieve adaptive energy growth sparsity assessment. A weighted fusion is then performed to obtain monitoring indicators that reflect the overall health of the transmission. Fault location is then performed using the monitoring indicators and the importance assessment results of the segmented amplitude and growth rate during the calculation process.

[0007] A method for monitoring the state of an automobile transmission using an adaptive energy growth sparsity metric comprises the following steps:

[0008] 1) Construction of status monitoring indicators under each gear:

[0009] 1.1) Obtaining basic data: Collect the full life vibration signal of the automobile gearbox test bench test, the sampling method is equal angle sampling, and the original order spectrum signal y at time t is obtained by fast Fourier transform t =[y0,y1,…,y N ], and its corresponding order sequence is x=[0,x δ ,2x δ ,…,N·x δ ], the order spectrum is a discrete order domain signal consisting of (N+1) points, with a resolution of x δ ; The value range of time t is [t start ,t end ], that is, the full life of the automobile transmission bench test, the mapping G(t) = {g, c} obtains the gear position g at time t and the number of gear cycles c, that is, time t is the cth cycle to gear position g; basic information of the transmission structure is also required, such as gear position g∈g, g is the set of gear positions g corresponding to all times obtained by mapping G(t) = {g, c}; the engagement order of each gear is Order g =[Order1,Order2,…,Order n ], where Order i ,i∈1,…,n represents the meshing order of the i-th gear pair in gear position g, and there are n meshing orders in total; In addition, it is necessary to set the monitoring indicator alarm threshold based on historical monitoring results. If there is no historical monitoring data or the monitoring data is very different, the default alarm threshold is 0.3;

[0010] 1.2) Remove meshing order components: Remove all meshing orders and their higher harmonic components in the current gear from the order spectrum. That is, set the meshing order components corresponding to the gears involved in the transmission in the order spectrum at each moment in the g gear to 0;

[0011] 1.3) Segmented amplitude sum: Divide the order spectrum after removing the meshing order components into E order segments, then the number of spectral lines in each order segment is The width of each order range is At time t when the gear is in gear position g, the sum of the amplitudes of each segment of the order spectrum is:

[0012]

[0013] 1.4) Calculate the segment amplitude and growth rate: Calculate the average of the segment amplitude and the order spectrum of the gearbox in the first cycle of each gear:

[0014]

[0015] Where N t is the number of moments that satisfy G(t) = {g,1};

[0016] Denote the element-wise division of vectors as μ g,1 is the reference value for calculating the segment energy growth rate. The segment amplitude and growth rate of the order spectrum at time t are:

[0017]

[0018] 1.5) Segmental Amplitude and Growth Rate Importance Assessment: The obtained segmental amplitude and growth rate are assessed for importance using the α-softmax(·) function, so that order segments with larger growth rates are assigned larger weights, while order segments with smaller or even negative growth rates are assigned smaller weights. The specific formula for segmental importance assessment is:

[0019]

[0020] Where α1>0 is a parameter used to control the sparsity of ω;

[0021] 1.6) Segmental Amplitude and Growth Sparsity Metric: Based on the energy growth of different order segments in the order spectrum, the top K order segments with the largest changes are selected. The segmental amplitude and growth sparsity metric SCSI under gear position g is constructed to indicate the health status of the gearbox. Its formula is:

[0022]

[0023] in sort(·) is a descending sort function, K is SCSI t,g The number of order segments with a larger energy growth rate used in the calculation, that is, the number of order segments with a larger energy growth rate ... t,g When using the first K order segments after descending sorting; due to ω t The sum of all elements is 1, so the obtained SCSI t,g There are certain upper and lower bounds, namely SCSI t,g ∈[0,1];

[0024] 2) Construction of indicators for monitoring the overall health of the transmission:

[0025] 2.1) Extend the SCSI indicators in each gear to the full time period: SCSI in gear g t,g Only calculated when {t|G(t)={g,c}}, SCSI under extended gear g t,g To other time points, we need to convert {t|G(t)≠{g,c}} into SCSI t,g The value is taken as the closest SCSI t,g SCSI t′,g,t′=max({t|G(t)={g,c}}), for different gears g∈g, at any time there is a corresponding SCSI t,g ;

[0026] 2.2) Evaluate the importance of SCSI indicators at different gears: t,g The α-softmax(·) function is used to evaluate the importance, so that the indicators with higher data scores are given larger weights. The specific evaluation formula is:

[0027]

[0028] Where α2>0 is a control Sparsity parameters;

[0029] 2.3) Weighted fusion of SCSI indicators at different gears: t,g according to The weights of the two indicators are weighted and fused into the WHI indicator that reflects the overall health status of the gearbox. The specific formula is as follows:

[0030]

[0031] 3) Transmission status monitoring and fault location: The obtained WHI is used to monitor the overall health of the transmission. A smaller WHI value indicates that the current transmission status is closer to the normal state, that is, the healthier the transmission status is. A larger WHI value indicates that the current transmission status deviates further from the normal state, that is, the healthier the transmission status is. A threshold is set based on historical monitoring data. When the WHI exceeds the threshold multiple times in a row, an alarm is issued. After the alarm is issued, the order spectrum of the faulty gear is preliminarily determined based on the ω calculated during the SCSI process, thereby determining the faulty gear meshing pair. Finally, the original order spectrum is used to find the sidebands of the faulty gear meshing pair to determine the faulty gear.

[0032] The beneficial effects of the present invention are:

[0033] The present invention proposes a method for monitoring the state of an automobile transmission using an adaptive energy growth sparsity metric. First, the method constructs a state monitoring indicator for each gear position through adaptive energy growth. Then, weighted fusion is performed to construct a monitoring indicator for the overall health state of the transmission. Finally, the obtained monitoring indicators, segmented amplitudes, and growth rate importance evaluation results are used together with the original order spectrum signal for layer-by-layer analysis to determine the position of the faulty gear in the transmission. This method realizes automatic state monitoring and auxiliary fault location of the automobile transmission, overcoming the shortcomings of traditional methods such as discontinuous monitoring indicators and inconvenient setting of failure thresholds. Description of the drawings:

[0034] Figure 1Flowchart of the present invention.

[0035] Figure 2 This is the automobile gearbox status monitoring result of an embodiment of the present invention. Specific implementation method:

[0036] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0037] Reference Figure 1 , a method for monitoring automobile transmission state based on adaptive energy growth sparsity measurement, comprising the following steps:

[0038] 1) Construction of Condition Monitoring Indicators for Each Gear: The vehicle transmission operating process monitoring data is collected. After obtaining the vibration signal order spectrum, removing the meshing order components, calculating the segmented amplitude sum, calculating the segmented amplitude and growth rate, and evaluating the importance of the segmented amplitude and growth rate, a spectral change sparsity measurement indicator (SCSI) for monitoring the condition of each gear of the transmission is constructed. The specific steps are as follows:

[0039] 1.1) Obtaining basic data: Collect the full life vibration signal of the automobile gearbox test bench test, the sampling method is equal angle sampling, and the original order spectrum signal y at time t is obtained by fast Fourier transform t =[y0,y1,…,y N ], and its corresponding order sequence is x=[0,x δ ,2x δ ,…,N·x δ ], the order spectrum is a discrete order domain signal consisting of (N+1) points, with a resolution of x δ The value range of time t is [t start ,t end ], that is, the full life of the automobile transmission bench test. The mapping G(t) = {g, c} can obtain the gear position g and the number of gear cycles c at time t, that is, time t is the cth cycle to gear position g; basic information of the transmission structure is also required, such as gear position g∈g, where g is the set of gear positions g corresponding to all times obtained by mapping G(t) = {g, c}; the engagement order of each gear is Order g =[Order1,Order2,…,Order n ], where Order i ,i∈1,…,n represents the meshing order of the i-th gear pair in gear position g, and there are n meshing orders in total; In addition, it is necessary to set the monitoring indicator alarm threshold based on historical monitoring results. If there is no historical monitoring data or the monitoring data is significantly different, the default alarm threshold is 0.3;

[0040] 1.2) Removing meshing order components: In the obtained order spectrum, the meshing order and its higher harmonics are order components that exist in both normal and faulty transmission conditions, and their amplitudes are generally much higher than other order components. To prevent the meshing order component from drowning out other order components and highlighting the order component changes caused by transmission faults, the meshing orders and their higher harmonic components in the current gear are removed from the order spectrum. That is, the meshing order components corresponding to the participating gears in the order spectrum at each moment in gear g are set to 0.

[0041] 1.3) Segmented amplitude sum: Divide the order spectrum after removing the meshing order components into E order segments, then the number of spectral lines in each order segment is The width of each order range is At time t when the gear is in gear position g, the sum of the amplitudes of each segment of the order spectrum is:

[0042]

[0043] 1.4) Calculate the segment amplitude and growth rate: Calculate the average of the segment amplitude and the order spectrum of the gearbox in the first cycle of each gear:

[0044]

[0045] Where N t is the number of moments that satisfy G(t) = {g,1};

[0046] Denote the element-wise division of vectors as μ g,1 is the reference value for calculating the segment energy growth rate. The segment amplitude and growth rate of the order spectrum at time t are:

[0047]

[0048] 1.5) Segmental Amplitude and Growth Rate Importance Assessment: The obtained segmental amplitude and growth rate are assessed for importance using the α-softmax(·) function, so that order segments with larger growth rates are assigned larger weights, while order segments with smaller or even negative growth rates are assigned smaller weights. The specific formula for segmental importance assessment is:

[0049]

[0050] Where α1>0 is a parameter used to control the sparsity of ω;

[0051] 1.6) Segmental Amplitude and Growth Sparsity Metric: The segmental energy growth rate gradually becomes sparse during the gear degradation process. The energy growth rate of each order segment tends to 0 in the normal state, and gradually evolves to the fault-related order segment where the energy growth rate increases significantly. Therefore, based on the energy growth of different order segments in the order spectrum, the top K order segments with the largest changes are selected. The segmental amplitude and growth sparsity metric SCSI under gear position g is constructed to indicate the health status of the gearbox. Its formula is:

[0052]

[0053] in sort(·) is a descending sort function, K is SCSI t,g The number of order segments with a larger energy growth rate is used in the calculation; due to ω t The sum of all elements is 1, so the obtained SCSI t,g There are certain upper and lower bounds, namely SCSI t,g ∈[0,1];

[0054] 2) Construction of a monitoring indicator for the overall health status of the transmission: To achieve overall status monitoring of the transmission in different gears, we expanded, evaluated, and weighted the SCSI model to propose a monitoring indicator (WHI) that can reflect the overall health status. The specific steps are as follows:

[0055] 2.1) Extend the SCSI indicators in each gear to the full time period: SCSI in gear g t,g Only calculated when {t|G(t)={g,c}}, SCSI under extended gear g t,g To other time points, we need to convert {t|G(t)≠{g,c}} into SCSI t,g The value is taken as the closest SCSI t,g SCSI t′,g ,t′=max({t|G(t)={g,c}}), for different gears g∈g, at any time there is a corresponding SCSI t,g ;

[0056] 2.2) Evaluate the importance of SCSI indicators at different gears: t,g The α-softmax(·) function is used to evaluate the importance, so that the indicators with higher data scores are given larger weights. The specific evaluation formula is:

[0057]

[0058] Where α2>0 is a control Sparsity parameters;

[0059] 2.3) Weighted fusion of SCSI indicators at different gears: t,g according to The weights of the two indicators are weighted and fused into the WHI indicator that reflects the overall health status of the gearbox. The specific formula is as follows:

[0060]

[0061] 3) Transmission status monitoring and fault location: The obtained WHI can be used to monitor the overall health of the transmission. The smaller the WHI value, the closer the current transmission status is to the normal state, that is, the healthier the transmission status is. The larger the WHI value, the greater the deviation from the normal state, that is, the worse the health status is. A threshold can be set based on historical monitoring data. When the WHI exceeds the threshold multiple times in a row, an alarm is issued. After the alarm, the order spectrum of the faulty gear meshing can be used to preliminarily determine the meshing order of the faulty gear, and thus the faulty gear meshing pair. Finally, the original order spectrum is used to find the sidebands of the faulty gear meshing pair to determine the faulty gear.

[0062] Example: Based on the full life experimental data of automobile gearboxes, the effectiveness of the method of the present invention is verified. The order range of the order spectrum used in this example is [0, 256], the order spectrum data length is N+1=2048, and the resolution is 0.125 orders; the gear cycle of the gearbox experimental data used is 3-4-5-6-7-8-1-2; the parameter combination is selected: the number of order spectrum segments E=128, the number of order segments with a larger rate of change K=16, α1=1, α2=10, and the alarm threshold is set to 0.3; the method of the present invention is used to monitor the health status, and the SCSI and WHI at each moment are calculated in turn. Figure 2 As shown in (a), the WHI indicator exceeds the preset threshold value in the 4th gear of the third cycle and the alarm is triggered. Further observation of the change process of ω in the 4th gear is shown as follows: Figure 2 As shown in (b), it can be seen that the order segments corresponding to orders 4 to 6 change significantly. Figure 2 The order spectrum in (c) corresponds to the meshing order of the gear pair meshing between the intermediate shaft and the output shaft, indicating that this gear pair is faulty. Unpacking and inspection revealed broken teeth and pitting on the gearbox intermediate shaft, demonstrating the effectiveness of the proposed method.

[0063] The method of the present invention is applicable to the health status monitoring of various types of gearboxes. In practical applications, implementers can adjust the parameters and thresholds accordingly based on actual conditions. Afterwards, the method can be used to monitor the health status of different gearboxes, which helps to improve the real-time performance and accuracy of gearbox status monitoring.

Claims

1. A method for monitoring the condition of an automobile transmission using an adaptive energy growth sparsity metric. The method is characterized by first acquiring the transmission's vibration signal and constructing a state monitoring indicator for each gear position to implement an adaptive energy growth sparsity assessment. A weighted fusion is then performed to obtain a monitoring indicator that reflects the overall health of the transmission. Fault location is then performed using the monitoring indicator and the segmented amplitude and growth rate importance assessment results from the calculation process. The automobile transmission state monitoring method based on adaptive energy growth sparsity measurement comprises the following steps: 1) Construction of status monitoring indicators under each gear: 1.1) Obtaining basic data: Collect the full life vibration signal of the automobile gearbox test bench test, the sampling method is equal angle sampling, and the original order spectrum signal y at time t is obtained by fast Fourier transform t =[y0,y1,…,y N ], and its corresponding order sequence is x=[0,x δ ,2x δ ,…,N·x δ ], the order spectrum is a discrete order domain signal consisting of (N+1) points, with a resolution of x δ; The value range of time t is [t start ,t end ], that is, the full life of the automobile transmission bench test, the mapping G(t) = {g, c} obtains the gear position g at time t and the number of gear cycles c, that is, time t is the cth cycle to gear position g; the basic information of the transmission structure is also required, gear position g∈g, g is the set of gear positions g corresponding to all times obtained by mapping G(t) = {g, c}; the engagement order of each gear is Order g =[Order1,Order2,…,Order n ], where Order i ,i∈1,…,n represents the meshing order of the i-th gear pair in gear position g, and there are n meshing orders in total; In addition, it is necessary to set the monitoring indicator alarm threshold based on historical monitoring results. If there is no historical monitoring data or the monitoring data is very different, the default alarm threshold is 0.3; 1.2) Remove meshing order components: Remove all meshing orders and their higher harmonic components in the current gear from the order spectrum. That is, set the meshing order components corresponding to the gears involved in the transmission in the order spectrum at each moment in the g gear to 0; 1.3) Segmented amplitude sum: Divide the order spectrum after removing the meshing order components into E order segments, then the number of spectral lines in each order segment is The width of each order range is At time t when the gear is in gear position g, the sum of the amplitudes of each segment of the order spectrum is: 1.4) Calculate the segment amplitude and growth rate: Calculate the average of the segment amplitude and the order spectrum of the gearbox in the first cycle of each gear: Where N t is the number of moments that satisfy G(t) = {g,1}; Denote the element-wise division of vectors as μ g,1 is the reference value for calculating the segment energy growth rate. The segment amplitude and growth rate of the order spectrum at time t are: 1.5) Segmental Amplitude and Growth Rate Importance Assessment: The obtained segmental amplitude and growth rate are assessed for importance using the α-softmax(·) function, so that order segments with larger growth rates are assigned larger weights, while order segments with smaller or even negative growth rates are assigned smaller weights. The specific formula for segmental importance assessment is: Where α1>0 is a parameter used to control the sparsity of ω; 1.6) Segmental Amplitude and Growth Sparsity Metric: Based on the energy growth of different order segments in the order spectrum, the top K order segments with the largest changes are selected. The segmental amplitude and growth sparsity metric SCSI under gear position g is constructed to indicate the health status of the gearbox. Its formula is: in sort(·) is a descending sort function, K is SCSI t,g The number of order segments with a larger energy growth rate used in the calculation, that is, the number of order segments with a larger energy growth rate ... t,g When using the first K order segments after descending sorting; due to ω t The sum of all elements is 1, so the obtained SCSI t,g There are certain upper and lower bounds, namely SCSI t,g ∈[0,1]; 2) Construction of indicators for monitoring the overall health of the transmission: 2.1) Extend the SCSI indicators in each gear to the full time period: SCSI in gear g t,g Only calculated when {t|G(t)={g,c}}, SCSI under extended gear g t,g To other time points, we need to convert {t|G(t)≠{g,c}} into SCSI t,g The value is taken as the closest SCSI t,g SCSI t',g ,t'=max({t|G(t)={g,c}}), for different gears g∈g, at any time there is a corresponding SCSI t,g ; 2.2) Evaluate the importance of SCSI indicators at different gears: t,g The α-softmax(·) function is used to evaluate the importance, so that the indicators with higher data scores are given larger weights. The specific evaluation formula is: Where α2>0 is a control Sparsity parameters; 2.3) Weighted fusion of SCSI indicators at different gears: t,g according to The weights of the two indicators are weighted and fused into the WHI indicator that reflects the overall health status of the gearbox. The specific formula is as follows: 3) Transmission status monitoring and fault location: The obtained WHI can be used to monitor the overall health of the transmission. The smaller the WHI value, the closer the current transmission status is to the normal state, that is, the healthier the transmission status is. The larger the WHI value, the greater the deviation from the normal state, that is, the worse the health status is. A threshold is set based on historical monitoring data. When the WHI exceeds the threshold multiple times in a row, an alarm is issued. After the alarm, the order spectrum of the faulty gear is preliminarily determined based on the ω calculated during the SCSI process, and thus the faulty gear meshing pair is determined. Finally, the original order spectrum is used to find the sidebands of the faulty gear meshing pair to determine the faulty gear.

Images