Abnormality detection device
AssigneeHitachi, Ltd.
InventorShinji Nakagawa
Life of this patent
Estimating the term…
Abstract
An object of the invention is to provide an abnormality detection device that can detect a processing abnormality of various types of software. The abnormality detection device according to the invention divides an output data series which is output by software into one or more clusters, determines that the output data included in any cluster is normal, and determines that the output data not included in any cluster is abnormal.
Claims12
An abnormality detection device that detects an abnormality of an operation result which is acquired by causing a processor to execute software, the abnormality detection device comprising:
- a divider that acquires a plurality of operation results from a storage device that stores the plurality of operation results and divides the acquired operation results into one or more groups;
- a range setter that sets a normal range of the operation result for each group using a range of data values included in the operation results included in each group;
- and
- an abnormality determiner that acquires a new operation result from the storage device and determines whether the new operation result is normal on the basis of the normal range after the range setter has set the normal range;
- wherein the divider generates vector data having one or more dimensions using the plurality of operation results;
- wherein the divider correlates types of the plurality of operation results with the dimensions of the vector data and then generates the vector data at the time of generating the vector data;
- wherein the range setter sets the normal range for each dimension of the vector data;
- and
- wherein the abnormality determiner determines whether the new operation result is normal on the basis of whether the new operation result is included in the normal range in the dimension of the vector data corresponding to the type of the new operation result,
- wherein the software has a process of calculating at least one of an intake air amount of an internal combustion engine, a fuel injection amount of the internal combustion engine, and an ignition time of the internal combustion engine as a control command value for controlling the internal combustion engine mounted therein, and
- wherein the divider acquires at least one of the intake air amount, the fuel injection amount, and the ignition time which are calculated by causing the processor to execute the software as the plurality of operation results.
An abnormality detection device that detects an abnormality of an operation result which is acquired by causing a processor to execute software, the abnormality detection device comprising:
- a divider that acquires a plurality of operation results from a storage device that stores the plurality of operation results and divides the acquired operation results into one or more groups;
- a range setter that sets a normal range of the operation result for each group using a range of data values included in the operation results included in each group;
- and
- an abnormality determiner that acquires a new operation result from the storage device and determines whether the new operation result is normal on the basis of the normal range after the range setter has set the normal range;
- wherein the divider acquires time-series data which is acquired by causing the processor to execute the software in a time series as the plurality of operation results;
- wherein the abnormality detection device further comprises a transition pattern calculator that calculates a transition pattern in which the operation result at each time in the time-series data transitions between the groups;
- and
- wherein the abnormality determiner determines that the new operation result is abnormal when the new operation result is included in the normal range but departs from the transition pattern;
- wherein the abnormality detection device acquires the new operation result as new time-series data which is acquired by causing the processor to execute the software in a time series,
- wherein the transition pattern calculator calculates an occurrence frequency of a new transition pattern in which the new operation result at each time in the new time-series data transitions between the groups for each transition path between the groups, and
- wherein the abnormality determiner determines that the new operation result is abnormal when the occurrence frequency of the new transition pattern is less than a predetermined threshold value at least one of the times and the new operation result is included in the normal range,
- wherein the software has a process of calculating at least one of an intake air amount of an internal combustion engine, a fuel injection amount of the internal combustion engine, and an ignition time of the internal combustion engine as a control command value for controlling the internal combustion engine mounted therein, and
- wherein the divider acquires at least one of the intake air amount, the fuel injection amount, and the ignition time which are calculated by causing the processor to execute the software as the plurality of operation results.
Dependencies are inferred from the claim text.