Prediction data generation device and vehicle control device

A vehicle control device comprising a prediction data generation device that generates prediction data in which a node of a directed graph is defined based on data groups acquired at a same time among data groups, each of which includes a plurality of types of data including data related to a device in a vehicle, and in which a link of the directed graph is defined by a transition of the node, the prediction data being used as an input of a transition prediction process which predicts the transition of the node, the prediction data generation device comprising:

• a storage device;
• and
• a processor configured to
• define, for each of a predetermined plurality of categories, a node per category as the node of the directed graph by a data subgroup of the data group, the data subgroup including a plurality of types of data belonging to the category,
• define a link per category as the link of the directed graph by a transition from one of a pair of different nodes to another,
• use the node per category, the link per category, and at least one of the number of observations of the node per category or the number of observations of the link per category as the prediction data, and
• execute a storage update process that updates the prediction data in order and stores the updated prediction data in the storage device in accordance with an update of the data group along with elapse of time,
• wherein the storage update process includes at least one of a process of increasing the number of observations of the node on a condition that the node per category is previously known, or a process of increasing the number of observations of the link on a condition that the link per category is previously known,
• wherein the processor is configured to execute the transition prediction process and a command generation process that generates a command for controlling the device in the vehicle based on prediction performed by the transition prediction process;
• and
• the transition prediction process sets a node corresponding to a current node of the vehicle as a start point node among the nodes stored in the storage device and sets a specific node from candidate nodes reached by one or a plurality of transitions defined by the link stored in the storage device as a predicted node specifying a subsequent state of the vehicle, and includes at least one of a process of selecting the candidate node having a relatively large number of observations as the predicted node preferentially over the candidate node having a relatively small number of observations or a process of selecting the candidate node having a relatively large number of observations of each link connecting the start point node and the candidate node as the predicted node preferentially over the candidate node having a relatively small number of observations of each link connecting the start point node and the candidate node.