Predicting vehicle repair operations
A method of producing vehicles comprises: in a vehicle production process, manufacturing vehicle components of different types, and assembling the vehicle components to form vehicles; creating a set of vehicle records, each being a record of one of the vehicles entering active service; performing vehicle repairs on a subset of the vehicles after they have entered active service; creating a respective record of each of the vehicle repairs, each of which comprises or indicates a vehicle age or usage value, and records a vehicle component fault identified in the vehicle repair; receiving at a data processing stage the vehicle records and vehicle repair records, wherein a predictive algorithm executed at the data processing stage processes the received records so as to, for each type of vehicle component: 1) identify a respective set of the vehicle repair records relating to that type of vehicle component, and 2) use the respective set of vehicle repair records to predict a respective number of or resource value for vehicle component faults of that type for the set of vehicle records based on: a number of vehicles recorded in the set of vehicle records, and a current age or usage of each of the recorded vehicles; comparing the predictions for the different vehicle component types to identify a problem with a particular one of the vehicle component types; and adapting the vehicle production process, so as to remedy the identified problem with the particular vehicle component type for later vehicles produced in the adapted vehicle production process.
1. A method of predicting machine repair operations or machine component faults for a pre-defined variant of a machine variant class, the method comprising, at a processing stage:
selecting, by a predictive algorithm executed at the processing stage, a set of machine repair records for use in making a prediction, each of the machine repair records being a record of a machine repair performed after the machine entered active service, each of which comprises or indicates a historical machine age or usage value, and records a repair operation or machine component fault,
wherein the predictive algorithm uses the selected set of machine repair records to predict a number of, or resource value for, repair operations or machine component faults for a set of machine records, each of the machine records being a record of a machine entering active service, based on: a number of machines recorded in the set of machine records being of the pre-defined variant, and a current age or usage of each of the recorded machines; and
determining a profile for the set of machine repair records based on a number of, or resource value for, repair operations or machine component faults recorded in the set of machine repair records for different historical machine age or usage values, the profile being used to make the prediction,
wherein the step of determining the profile comprises: determining a total number of, or resource value for, repair operations/machine component faults recorded in the filtered set of machine repair records, each resource or count value being calculated as a proportion of the total,
wherein an earnings value is calculated for each of the historical machine age or usage values of the profile based on the corresponding resource or count value of the profile and the number of machines recorded in the set of machine records whose current age or usage matches that historical machine age or usage value of the profile.
2. A method according to claim 1 , wherein the machine is a vehicle, the machine repair records are vehicle repair records, the machine repair is a vehicle repair, the historical machine age or usage value is a historical vehicle age or usage value, the machine component fault is a vehicle component fault and the machine records are vehicle records.
3. A method according to claim 1 , wherein the set of machine repair records is selected by filtering a larger set of available machine repair records based on a particular type of repair operation or a particular type of machine component, such that each repair record of the selected set relates to the particular type of repair operation/machine component, the predicted number or resource value being a predicted number of or resource value for repair operations/machine component faults of the particular type.
4. A method according to claim 1 , wherein the set of machine repair records is selected by filtering a larger set of available machine repair records based on a particular machine attribute or set of machine attributes, such that each repair record of the selected set relates to the particular (set of) machine attribute(s), wherein each of the machine records in the set of machine records relates to a machine having the particular (set of) machine attribute(s) or a similar (set of) machine attribute(s), the predicted number or resource value being a predicted number of or resource value for repair operations/machine component faults for machines having the particular machine attribute.
5. A method according to claim 1 , wherein the profile comprises, for each of a set of historical machine age or usage values, a corresponding resource or count value calculated from the set of repair records.
6. A method according to claim 5 , wherein the resource or count value is a cumulative value calculated as sum of the number of, or resource values for, repair operations/machine component faults recorded in the set of machine repair records up to that historical machine age or usage value.
7. A method according to claim 5 , wherein the prediction is made by performing a non-parametric analysis based on the number of machines recorded in the set of machine records, the current age or usage of each of the machines, and the resource or count values of the profile.
8. A method according to claim 1 , wherein the earnings value for each of the historical machine age or usage values of the profile is calculated by multiplying the corresponding resource or count value of the profile with the number of machines recorded in the set of machine records whose current age or usage matches that historical machine age or usage value of the profile.
9. A method according to claim 1 , comprising determining a total number of machines recorded in the set of machine records and calculating a maturity value for the set of machine records from the earnings values, wherein the maturity value is calculated by calculating a total earnings value from the earnings values as a proportion of the total number of machines.
10. A method according to claim 1 , comprising:
identifying one or more existing machine repair records corresponding to the set of machine records; and
determining a number of, or resource value for, repair operations or machine component faults recorded in the existing machine repair records.
11. A method according to claim 10 , wherein a predicted number of, or resource values for, repair operations or machine component faults for the set of machine records is computed from the number of, or resource value for, repair operations or machine component faults determined for the existing machine repair records based on the number of machines recorded in the set of machine records and the current age or usage of each of the machines.
12. A method according to claim 11 , wherein the profile is used to compute the predicted number.
13. A method according to claim 8 , wherein the method further comprises identifying one or more existing machine repair records corresponding to the set of machine records; and determining a number of, or resource value for, repair operations or machine component faults recorded in the existing machine repair records;
wherein the predicted number of, or resource value for, repair operations or machine component faults for the set of machine records is determined based on the maturity value calculated for the set of machine records and the number of, or resource value for, repair operations/machine component faults recorded in the existing machine repair records.
14. A method according to claim 13 , wherein the predicted number or resource value is determined by dividing, by the maturity, the number of resource value for repair operations/machine component faults recorded in the existing machine repair records.
15. A method according to claim 1 , wherein the predictive algorithm implements the following steps for each of a plurality of repair operation or machine component types:
selecting a respective set of machine repair records relating to that type of repair operation/machine component; and
using the respective set of machine repair records to predict, for the set of machine records, a number of, or resource value for, repair operations or machine component faults of that type based on the number of recorded machines and their current age or usage.
16. A system for predicting machine repair operations or machine component faults for a pre-defined variant of a machine variant class, the system comprising:
electronic storage configured to hold computer readable instructions for executing a predictive algorithm; and
a processing stage coupled to the electronic storage and configured to execute computer readable instructions, the computer readable instructions being configured, when executed, to implement operations comprising:
selecting, by a predictive algorithm executed at the processing stage, a set of machine repair records for use in making a prediction, each of the machine repair records being a record of a machine repair performed after the machine entered active service, each of which comprises or indicates a historical machine age or usage value, and records a repair operation or machine component fault,
wherein the predictive algorithm uses the selected set of machine repair records to predict a number of, or resource value for, repair operations or machine component faults for a set of machine records, each of the machine records being a record of a machine entering active service, based on: a number of machines recorded in the set of machine records being of the pre-defined variant, and a current age or usage of each of the recorded machines; and
determining a profile for the set of machine repair records based on a number of, or resource value for, repair operations or machine component faults recorded in the set of machine repair records for different historical machine age or usage values, the profile being used to make the prediction,
wherein the step of determining the profile comprises: determining a total number of, or resource value for, repair operations/machine component faults recorded in the filtered set of machine repair records, each resource or count value being calculated as a proportion of the total,
wherein an earnings value is calculated for each of the historical machine age or usage values of the profile based on the corresponding resource or count value of the profile and the number of machines recorded in the set of machine records whose current age or usage matches that historical machine age or usage value of the profile.
17. A non-transitory computer readable medium having computer readable instructions configured, when executed, to implement operations comprising:
selecting, by a predictive algorithm executed at the processing stage, a set of machine repair records for use in making a prediction, each of the machine repair records being a record of a machine repair performed after the machine entered active service, each of which comprises or indicates a historical machine age or usage value, and records a repair operation or machine component fault
wherein the predictive algorithm uses the selected set of machine repair records to predict a number of, or resource value for, repair operations or machine component faults for a set of machine records, each of the machine records being a record of a machine entering active service, based on: a number of machines recorded in the set of machine records being of the pre-defined variant, and a current age or usage of each of the recorded machines; and
determining a profile for the set of machine repair records based on a number of, or resource value for, repair operations or machine component faults recorded in the set of machine repair records for different historical machine age or usage values, the profile being used to make the prediction,
wherein the step of determining the profile comprises: determining a total number of, or resource value for, repair operations/machine component faults recorded in the filtered set of machine repair records, each resource or count value being calculated as a proportion of the total,
wherein an earnings value is calculated for each of the historical machine age or usage values of the profile based on the corresponding resource or count value of the profile and the number of machines recorded in the set of machine records whose current age or usage matches that historical machine age or usage value of the profile.