Behavioral model generating device and method therefor

A behavioral model generating method, comprising;

• reading data from at least one memory configured to store a plurality of behavioral pattern groups and weighting factors set for the behavioral pattern groups, the behavioral pattern groups each including a plurality of behavioral patterns that indicate consumer actions placed along with time;
• selecting, via a processor and according to the weighting factors, a plurality of behavioral patterns from all the behavioral patterns included in the behavioral pattern groups by duplicate sampling;
• adding, via the processor, modification to each of selected behavioral patterns to generate a plurality of first behavioral patterns;
• generating, via the processor and based on the first behavioral patterns, device operation patterns that indicate device operations placed along with time according to a device operation model that defines correspondence between actions and device operations;
• calculating, via the processor, power consumption data for each of the device operation patterns, which indicate temporal transition of power consumption according to a device power consumption model that defines correspondence between device operations and power consumption;
• reading data from the at least one memory configured to store measured power consumption data that indicates temporal transition of power consumption by one or more devices in a consumer house;
• calculating, via the processor, a likelihood of each calculated power consumption data with respect to measured power consumption data, respectively, and selecting, according to the likelihood, a plurality of behavioral patterns from the first behavioral patterns by duplicate sampling as a first behavioral pattern group;
• selecting, via the processor, a pair of most similar behavioral pattern groups from all the behavioral pattern groups including the first behavioral pattern group;
• selecting, via the processor, a part of all the behavioral patterns included in the pair as a second behavioral pattern group;
• updating, via the processor, the pair of the most similar behavioral pattern groups with the second behavioral pattern group;
• and
• setting, via the processor, a weighting factor of the second behavioral pattern group to a sum of respective weighting factors of the pair of the most similar behavioral pattern groups.

A non-transitory computer-readable medium storing thereon computer-readable instructions which when executed by a computer cause the computer to execute a behavioral model generating method, comprising;

• reading data from at least one memory configured to store a plurality of behavioral pattern groups and weighting factors set for the behavioral pattern groups, the behavioral pattern groups each including a plurality of behavioral patterns that indicate consumer actions placed along with time;
• selecting, according to the weighting factors, a plurality of behavioral patterns from all the behavioral patterns included in the behavioral pattern groups by duplicate sampling;
• adding modification to each of selected behavioral patterns to generate a plurality of first behavioral patterns;
• generating, based on the first behavioral patterns, device operation patterns that indicate device operations placed along with time according to a device operation model that defines correspondence between actions and device operations;
• calculating power consumption data for each of the device operation patterns, which indicate temporal transition of power consumption according to a device power consumption model that defines correspondence between device operations and power consumption;
• reading data from the at least one memory configured to store measured power consumption data that indicates temporal transition of power consumption by one or more devices in a consumer house;
• calculating a likelihood of each calculated power consumption data with respect to measured power consumption data, respectively, and selecting, according to the likelihood, a plurality of behavioral patterns from the first behavioral patterns by duplicate sampling as a first behavioral pattern group;
• selecting a pair of most similar behavioral pattern groups from all the behavioral pattern groups including the first behavioral pattern group;
• selecting a part of all the behavioral patterns included in the pair as a second behavioral pattern group;
• updating the pair of the most similar behavioral pattern groups with the second behavioral pattern group;
• and
• setting a weighting factor of the second behavioral pattern group to a sum of respective weighting factors of the pair of the most similar behavioral pattern groups.