Living activity inference device, and program

Patent metadata

Field	Value
Publication number	US-2016371593-A1
Application number	US-201515122327-A
Country	US
Kind code	A1
Filing date	Feb 12, 2015
Priority date	Mar 11, 2014
Publication date	Dec 22, 2016
Grant date	—

Abstract

An inferrer calculates, for each kind of living activities, an occurrence probability defining a probability that a corresponding living activity occurs, based on an operational state of an electric appliance determined by a detector for each of sections at an inference target time by use of prior probabilities, first conditional probabilities, and second conditional probabilities, and infers that a living activity having a highest occurrence probability has occurred at inference target time. Prior probabilities each define a probability at which a corresponding living activity occurs in one day. First conditional probabilities each define active and inactive probabilities of electric appliance belonging to a corresponding section under a condition where a corresponding living activity occurs. Second conditional probabilities each define probabilities of occurrence of a corresponding living activity within respective time periods divided from one day under a condition where a corresponding living activity occurs.

First claim

1 . A living activity inference device, comprising:
an obtainer configured to obtain, for each of sections of a residence, data on a resource consumption status of one or more appliances belonging to a corresponding one of the sections; an appliance operation detector configured to determine, for each of the sections, an operational state indicative of whether the one or more appliances belonging to a corresponding one of the sections are active or inactive, from the data on the resource consumption status obtained by the obtainer for each of the sections; a storage device configured to preliminarily store
prior probabilities each defining a probability at which a corresponding one of pre-defined multiple kinds of living activities occurs in one day,
first conditional probabilities each defining active and inactive probabilities of the one or more appliances belonging to a corresponding one of the sections under a condition where a corresponding one of the multiple kinds of living activities occurs, and
second conditional probabilities each defining probabilities of occurrence of a corresponding one of the multiple kinds of living activities within respective time periods divided from one day under a condition where the corresponding one of the multiple kinds of living activities occurs; and
an activity inferrer configured to
calculate, for each of the multiple kinds of living activities, an occurrence probability defining a probability that a corresponding one of the multiple kinds of living activities occurs at an inference target time, according to Bayes' theorem based on the operational state of the one or more appliances determined by the appliance operation detector for each of the sections at the inference target time by use of the prior probabilities, the first conditional probabilities, and the second conditional probabilities, and
infer that a living activity having a highest occurrence probability of the multiple kinds of living activities has occurred at the inference target time.

2 . The living activity inference device of claim 1 , wherein
the one or more appliances are one or more electric appliances, a distribution board in the residence is connected to branch lines, each of the branch lines is connected to at least one of the one or more electric appliances belonging to a corresponding one of the sections, and delivers electricity from the distribution board to the at least one of the one or more electric appliances connected thereto, and the obtainer is configured to obtain, for each of the branch lines, power data.

3 . The living activity inference device of claim 1 , wherein
the storage device is configured to preliminarily store transition probabilities each defining probabilities at which the multiple kinds of living activities occur subsequent to a corresponding one of the multiple kinds of living activities, and the activity inferrer is configured to further use the transition probabilities for calculating, for each of the multiple kinds of living activities, the occurrence probability at the inference target time.

4 . The living activity inference device of claim 1 , further comprising:
an inputter configured to receive feedback information indicative of a disagreement between an inference result of the activity inferrer and an actual living activity of a user at the inference target time, and an updater configured to update at least one of the prior probabilities, the first conditional probabilities, and the second conditional probabilities stored in the storage device, based on the feedback information received through the inputter.

5 . The living activity inference device of claim 1 , wherein the multiple kinds of living activities include an at-home state, a go-out state, and a sleep state.

6 . The living activity inference device of claim 1 , wherein
the activity inferrer is configured to
periodically perform a process of detecting change in an inference result with regard to the multiple kinds of living activities as an event,
record a pattern of occurrence of the event in one day, and
determine that there is irregularity in occurrence of the multiple kinds of living activities when an occurrence timing of the event detected after the pattern is recorded does not conform to the pattern.

7 . A program that when executed on a computer, causes the computer to function as the living activity inference device of claim 1 .

Assignees

• Panasonic Ip Man Co Ltd

Inventors

• Nishiyama Takashi
• Shimizu Noriyoshi
• Fujita Tomohiko
• Nakahara Tomoharu

Classifications

• G08B21/0423 Primary

  detecting deviation from an expected pattern of behaviour or schedule · CPC title

• G08B21/0484

  Arrangements monitoring consumption of a utility or use of an appliance which consumes a utility to detect unsafe condition, e.g. metering of water, gas or electricity, use of taps, toilet flush, gas stove or electric kettle · CPC title

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06N5/046 Primary

  Forward inferencing; Production systems · CPC title

• G06Q30/0201

  Market modelling; Market analysis; Collecting market data · CPC title