Cloud-based vehicle fault diagnosis method, apparatus, and system
A cloud-based vehicle fault diagnosis method includes receiving monitoring data uploaded by a vehicle; extracting eigenvectors of the monitoring data from the monitoring data; storing the eigenvectors of the monitoring data based on classification using, as a label, a part or a functional system of the vehicle from which the monitoring data comes; and performing, based on a support vector machine algorithm, fault diagnosis in parallel on the eigenvectors stored based on the classification.
1. A cloud-based vehicle fault diagnosis method, comprising:
receiving monitoring data of a vehicle, wherein the monitoring data comprises monitored data of a working status of either a part of the vehicle or a functional system of the vehicle;
extracting first eigenvectors from the monitoring data, wherein the first eigenvectors are a set of numbers representing the monitoring data;
classifying the first eigenvectors as classified eigenvectors using a first label, wherein the first label indicates either the part of the vehicle or the functional system of the vehicle;
storing the classified eigenvectors as stored eigenvectors;
performing, based on a support vector machine algorithm, a fault diagnosis in parallel on the stored eigenvectors to obtain a fault, wherein the fault affects a safety state of the vehicle, and wherein performing the fault diagnosis in parallel comprises:
constructing a fault classifier based on a decision directed acyclic graph (DDAG); and
calculating, according to the fault classifier based on the DDAG, a diagnosis result of the stored eigenvectors in parallel based on the classified eigenvectors, wherein the diagnosis result comprises at least a position at which the fault occurs; and
sending, after obtaining related data of the fault, a danger warning signal to the vehicle, wherein the danger warning signal instructs the vehicle to take a fault countermeasure.
2. The cloud-based vehicle fault diagnosis method of claim 1 , wherein before extracting the first eigenvectors from the monitoring data, the cloud-based vehicle fault diagnosis method further comprises:
parsing the monitoring data to obtain parsed monitoring data;
classifying the parsed monitoring data as classified parsed monitoring data using a second label, wherein the second label indicates either the part of the vehicle or the functional system of the vehicle, and wherein the second label corresponds to the first label; and
storing the classified parsed monitoring data, and
extracting second eigenvectors of the parsed monitoring data from the parsed monitoring data.
3. The cloud-based vehicle fault diagnosis method of claim 1 , further comprising further extracting the first eigenvectors from the monitoring data using wavelet packet decomposition.
4. The cloud-based vehicle fault diagnosis method of claim 3 , wherein before storing the classified eigenvectors as the stored eigenvectors, the cloud-based vehicle fault diagnosis method further comprises reducing a dimension of the classified eigenvectors through kernel principal component analysis to obtain dimension-reduced eigenvectors, wherein storing the classified eigenvectors comprises:
classifying the dimension-reduced eigenvectors as classified dimension-reduced eigenvectors using a third label, wherein the third label indicates either the part of the vehicle or the functional system of the vehicle; and
storing the classified dimension-reduced eigenvectors, and
wherein performing the fault diagnosis on the vehicle using the stored eigenvectors comprises performing, based on the support vector machine algorithm, fault diagnosis on the classified dimension-reduced eigenvectors.
5. The cloud-based vehicle fault diagnosis method of claim 1 , wherein the fault corresponds to a severity level and wherein the severity level comprises a severe fault, a moderate fault, or a general fault.
6. The cloud-based vehicle fault diagnosis method of claim 5 , wherein the severity level is the severe fault, and wherein the danger warning signal notifies the vehicle that occurrence of the severe fault endangers normal driving of the vehicle.
7. A cloud-based vehicle fault diagnosis apparatus comprising:
a microcomputer configured to store instructions, which when executed by the microcomputer, cause the microcomputer to be configured to:
receive monitoring data of a vehicle, wherein the monitoring data comprises monitored data of a working status of either a part of the vehicle or a functional system of the vehicle;
extract first eigenvectors from the monitoring data, wherein the first eigenvectors are a set of numbers representing the monitoring data;
classify the first eigenvectors as classified eigenvectors using a first label, wherein the first label indicates either the part of the vehicle or the functional system of the vehicle; and
a feature database coupled to the microcomputer and configured to store, based on classification of the first eigenvectors, the classified eigenvectors as stored eigenvectors, and
wherein the microcomputer is further configured to:
perform, based on a support vector machine algorithm, a fault diagnosis in parallel on the stored eigenvectors to obtain a fault, wherein the fault affects a safety state of the vehicle, and wherein performing the fault diagnosis in parallel comprises:
constructing a fault classifier based on a decision directed acyclic graph DDAG; and
calculating, according to the fault classifier based on the DDAG, a diagnosis result of the stored eigenvectors in parallel based on the classified eigenvectors, wherein the diagnosis result comprises at least a position at which a fault occurs; and
send after obtaining related data of the fault, a danger warning signal to the vehicle, wherein the danger warning signal instructs the vehicle to take a fault countermeasure.
8. The cloud-based vehicle fault diagnosis apparatus of claim 7 , wherein the cloud-based vehicle fault diagnosis apparatus further comprises a central database coupled to the microcomputer and the feature database, and wherein the central database is configured to:
parse the monitoring data to obtain parsed monitoring data;
classify the parsed monitoring data as classified parsed monitoring data using a second label, wherein the second label indicates either the part of the vehicle or the functional system of the vehicle, and wherein the second label corresponds to the first label; and
store the classified parsed monitoring data, and wherein the microcomputer is further configured to extract second eigenvectors of the parsed monitoring data from the parsed monitoring data.
9. The cloud-based vehicle fault diagnosis apparatus of claim 7 , wherein the microcomputer is configured to extract the first eigenvectors from the monitoring data using wavelet packet decomposition.
10. The cloud-based vehicle fault diagnosis apparatus of claim 9 , wherein the instructions further cause the microcomputer to be configured to:
reduce a dimension of the classified eigenvectors through kernel principal component analysis to obtain dimension-reduced eigenvectors; and
perform, based on the support vector machine algorithm, fault diagnosis on the dimension-reduced eigenvectors, and
wherein the feature database is further configured to:
classify the dimension-reduced eigenvectors as classified dimension-reduced eigenvectors using a third label, wherein the third label indicates either the part of the vehicle or the functional system of the vehicle; and
store the classified dimension-reduced eigenvectors.
11. The cloud-based vehicle fault diagnosis apparatus of claim 7 , wherein the fault corresponds to a severity level, and wherein the severity level comprises a severe fault, a moderate fault, and a general fault.
12. The cloud-based vehicle fault diagnosis apparatus of claim 11 , wherein the instructions further cause the microcomputer to be configured to further determine the severity level is the severe fault, and wherein the danger warning signal is configured to notify the vehicle that occurrence of the severe fault during operation of the vehicle endangers normal driving of the vehicle.
13. A cloud-based vehicle fault diagnosis system, comprising:
a vehicle; and
a cloud-based vehicle fault diagnosis apparatus coupled to the vehicle and configured to:
receive monitoring data of the vehicle, wherein the monitoring data comprises monitored data of a working status of either a part of the vehicle or a functional system of the vehicle;
extract first eigenvectors from the monitoring data, wherein the first eigenvectors are a set of numbers representing the monitoring data;
classify the first eigenvectors as classified eigenvectors using a first label, wherein the first label indicates either the part of the vehicle or the functional system of the vehicle; and
store, based on classification of the first eigenvectors, the classified eigenvectors as stored eigenvectors;
perform, based on a support vector machine algorithm, fault diagnosis in parallel on the stored eigenvectors of the vehicle to obtain a fault, wherein the fault affects a safety state of the vehicle, and wherein performing the fault diagnosis in parallel comprises:
constructing a fault classifier based on a decision directed acyclic graph DDAG; and
calculating, according to the fault classifier based on the DDAG, a diagnosis result of the stored eigenvectors in parallel based on the classified eigenvectors, wherein the diagnosis result comprises at least a position at which a fault occurs; and
send after obtaining related data of the fault affecting safety state of the vehicle, a danger warning signal to the vehicle, wherein the danger warning signal instructs the vehicle to take a fault countermeasure.
14. The cloud-based vehicle fault diagnosis system of claim 13 , wherein the cloud-based vehicle fault diagnosis system is further configured to send the danger warning signal to the vehicle when the fault is diagnosed as a severe fault, and wherein the danger warning signal notifies the vehicle of an occurrence of the severe fault during real-time operation of the vehicle endangers normal driving of the vehicle.
15. The cloud-based vehicle fault diagnosis system of claim 13 , wherein the cloud-based vehicle fault diagnosis system is configured to extract the first eigenvectors from the monitoring data using wavelet packet decomposition.
16. The cloud-based vehicle fault diagnosis system of claim 13 , wherein the fault corresponds to a severity level.
17. The cloud-based vehicle fault diagnosis system of claim 16 , wherein the cloud-based vehicle fault diagnosis system is configured to:
reduce a dimension of the classified eigenvectors through kernel principal component analysis to obtain dimension-reduced eigenvectors; and
perform, based on the support vector machine algorithm, the fault diagnosis on the dimension-reduced eigenvectors;
classify the dimension-reduced eigenvectors as classified dimension-reduced eigenvectors using a third label, wherein the third label indicates either the part of the vehicle or the functional system of the vehicle; and
store the classified dimension-reduced eigenvectors.
18. The cloud-based vehicle fault diagnosis system of claim 16 , wherein the severity level comprises a severe fault, a moderate fault, or a general fault.
19. The cloud-based vehicle fault diagnosis system of claim 13 , wherein the cloud-based vehicle fault diagnosis apparatus is further configured to:
parse the monitoring data to obtain parsed monitoring data;
classify the parsed monitoring data as classified parsed monitoring data using a second label, wherein the second label indicates either the part of the vehicle or the functional system of the vehicle, and wherein the second label corresponds to the first label; and
store the classified parsed monitoring data; and
extract second eigenvectors of the parsed monitoring data from the parsed monitoring data.