JP2013228891A - Usage prediction method and behavior recommendation method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To recommend an appropriate user behavior relating to current situation improvement, for example, energy saving under the consideration of any situation change other than a relationship between input/output values.SOLUTION: A past period having attributes which are similar to an attribute value of a prediction period is selected, and usage transition of the period is subjected to curve approximation with a first function, and the usage transition determined at the current point of time is synthesized with the first function, and subjected to curve approximation with a second function so that the usage can be predicted. Also, a behavior is recommended on the basis of the evaluation value of each item of possible behavior information on the basis of a relation between the predicted approximation curve and a target value.

Description

  The present invention relates to a system for measuring the usage of resources such as electric power and predicting the future usage from past measurement data, and in particular recommends user actions to reduce the predicted future usage. Regarding technology.

  In recent years, with the progress of global warming, the reduction of forest resources, and the expansion of environmental destruction, the importance of reducing emissions of greenhouse gases such as carbon dioxide has become increasingly important. For this reason, regarding energy conservation (hereinafter abbreviated as energy conservation) measures in Japan, the “Act on Rational Use of Energy” (Amended Energy Conservation Act), which was revised in April 2010, came into effect, and business that includes small and medium-sized retail stores. Energy consumption management has become mandatory. In addition, the environment surrounding energy conservation is undergoing a major revolution, with 15% power saving required for companies due to the Great East Kanto Earthquake in March 2011.

  However, it is not possible to carry out excessive energy-saving activities such as energy saving and firewood that bring about a temperature rise that causes heat stroke by completely stopping cooling, or even turn off the light in the room reading the document. As well as being bad for the body, it leads to a sense of dislike for energy-saving activities and adversely affects continuity.

  On the other hand, letting the user decide what to use for energy conservation activities is attribute of many buildings and rooms, attributes of electrical products used there, influence on environment and operability due to on / off, arrangement of people and things, etc. In the presence of various parameters, it is not easy. Therefore, the system automatically supports activities (actions) balanced with energy saving while minimizing the degradation of comfort for users and the people around them. It is important to make the user feel that.

  For this reason, it is required that the system predicts the usage amount, and further recommends an appropriate action to the user in accordance with a preference value based on the predicted value, a predetermined target value, the surrounding situation of the date and time, and the user's behavior history. Here, in a situation where there are action candidates that affect many environmental factors, it is a problem to evaluate the influence that each action is considered to have and to find an appropriate action.

  As a conventional general prediction method, there is used a method of obtaining a statistical relationship from input / output data in the same case in the past and predicting an output value when an unknown input value is given. For example, in Patent Document 1, the similarity between the prediction target data and the past execution data, and the output data and the variation are obtained from the past execution data. Here, the weight vector A in which each element when calculating the similarity between the factor vectors (such as the passage of time) contributes to the variation is calculated, and the output is predicted based on the weight B corresponding to the number of the plurality of implementation data Has been shown to do.

JP 2010-231447 A

  As described above, the conventional prediction method statistically uses a mathematical model based on a so-called system transfer function or a fixed model based on a spatial approximation between input and output values. The amount of usage is predicted by system estimation. However, it is difficult to consider the relationship between the change of the situation exceeding the relationship between the input and output values, for example, the climate change, the user's behavioral preference, the process and reason until the activity, or the result.

  Furthermore, Patent Document 1 is intended to evaluate the variation in predicted values obtained in this way, that is, to improve accuracy. Therefore, no mention is made of how to deal with the case where the variation of the predicted value exceeds or exceeds the predetermined accuracy. That is, in Patent Document 1, for example, when the deviation from a preset target value is large or when a preset limit value is exceeded, the current deviation or limit is not exceeded, or the difference is indicated to the user. It is difficult to recommend specific actions to make small and improve the situation.

  Therefore, in the prior art, it is a problem to recommend an appropriate user action related to current state improvement, for example, energy saving, in consideration of situational fluctuations other than the relationship between input and output values.

  An object of the present invention is to integrate a climate change, an activity plan, an activity reason, and past results in a system for predicting the usage amount of energy and the like, and to enable a prediction method suitable for human activities to be introduced at a low cost. is there. In addition, by providing a method for recommending actions to reduce usage suitable for the climate and action plan in consideration of the relationship between predicted usage trends and target values, it is possible to identify specific users for situation improvement. In addition to reducing energy consumption, it is intended to provide motivation, strengthening, and maintenance of users' energy-saving activities based on the causal relationship between actions and reduction results.

  According to the usage amount prediction method of the first invention, when the past usage history information and the state attribute information of the period measured in the period and interval specified in advance are acquired and the period for which the prediction is performed is specified The situation attribute information is obtained from the situation attribute information acquisition means or the input means, the prediction status attribute information that is predicted or scheduled attribute information of the period, and the calculation means is the situation attribute that is most similar to the predicted genus information from the situation attribute information of the recording unit A first approximation that selects one piece of information, obtains a usage history from the usage history information for the same period indicated by the selected status attribute information, and approximates the usage with respect to time using the history of the usage history information A curve is obtained, and the value obtained by the first approximate time curve and the usage history information at each time of the history of one or more usage history information belonging to the specified period or immediately before the period for which the prediction is performed Determination of the second approximation curve which gives an approximation between the usage, the prediction curve usage curve obtained by synthesizing the first approximation curve and the second approximation curve.

  The usage amount prediction system according to the second aspect of the present invention is a usage amount for acquiring usage history information including a calculation unit, a display unit, an input unit for user operation, a time acquisition unit, and a usage amount and time. A history information acquisition unit, a status attribute information acquisition unit that acquires status attribute information in a period specified in advance, and a predicted status attribute information acquisition unit that acquires a predicted status attribute from a predicted status attribute generation unit that predicts future attribute changes And a recording unit for recording various information.

  In the usage amount prediction method of the third invention, the situation attribute information and the forecast situation attribute information include date and time, weather information, the activity status of the user, the activity status of the group to which the user belongs, and the user. At least one of an environmental situation of a place to which the user belongs and a situation that can be a factor affecting the usage amount is used.

  In the usage amount prediction method according to the fourth aspect of the invention, the approximate curve is displayed on the display unit together with one or more usage amount history information belonging to immediately before or within the designated period.

  In addition to the first invention, the behavior recommendation method of the fifth invention acquires one or more possible behavior information sets that define user behavior that affects the usage amount, and provides a usage threshold value to be given in advance. Find the prediction excess information including at least one of the time when the usage prediction curve intersects and the extreme value of the prediction curve, and each attribute information of each item in the possible behavior information, the prediction excess information, The evaluation value between the attribute information is obtained by a predetermined method using the predicted situation attribute information, the possible behavior information is selected using the evaluation value for each item of the possible behavior information, and the possible behavior information of the selected possible behavior information is displayed on the display means. Present the contents of the item.

  Further, in the behavior recommendation method of the sixth invention, instead of the possible behavior information in the fifth invention, the behavior history information including the user's behavior that affects the usage amount and the change amount of the usage amount due to the behavior is similarly used. Perform the process.

  Further, in the behavior recommendation method of the seventh invention, in the fifth or sixth invention, a user response to the presented recommendation presentation is acquired from the input means, and the response is included in one of the possible behavior information. If it was, update at least one element from the response content, presentation history, selection history, selection preference, or implementation date and time as possible behavior information or behavior history information elements, Used for evaluation.

  The behavior recommendation system of the eighth invention has the same system configuration as that of the second invention, and the usage amount prediction method of the first invention and at least one of the fifth to seventh inventions Do more than one action recommendation method.

  The behavior recommendation system of the ninth invention has the same system configuration as that of the eighth invention, and includes usage history information consisting of usage and time, status attribute information in a predesignated period, and future attributes. A usage amount prediction method according to the first aspect of the present invention, comprising: a communication unit that acquires predicted status attribute information from a predicted status attribute generation unit that predicts a change via a network; and a recording unit that records various types of information. And at least 1 or more action recommendation method is performed among 5th invention from 7th invention.

  According to the present invention, it is possible to predict the usage of resources such as energy at a low cost, support the user's specific action recall for improving the situation, and motivate the user's energy-saving activities. Can assist in strengthening and maintaining.

It is a block diagram which shows the system configuration | structure which performs the usage-amount prediction and action recommendation of 1st Example of this invention. It is explanatory drawing of the example of the various information hold | maintained at the recording part of 1st Example of this invention. It is explanatory drawing of the example of the usage-amount log information of 1st Example of this invention. It is explanatory drawing of the situation attribute information and prediction situation attribute information of 1st Example of this invention. It is explanatory drawing of the example of the possible action information of 1st Example of this invention. It is explanatory drawing of the example of the target value information of 1st Example of this invention. It is explanatory drawing of the example of the situation attribute evaluation rule of 1st Example of this invention. It is explanatory drawing of the example of the action recommendation rule of 1st Example of this invention. It is explanatory drawing of the example of the situation attribute evaluation result information of 1st Example of this invention. It is explanatory drawing of the example of the situation characteristic information of 1st Example of this invention. It is explanatory drawing of the example of the action recommendation result information of 1st Example of this invention. It is a processing flowchart of the example of the main process of 1st Example of this invention. It is a processing flowchart of the example of the usage-amount prediction process of 1st Example of this invention. It is explanatory drawing of the example of the 1st approximated curve of 1st Example of this invention. It is explanatory drawing of the example of the usage-amount of the day of 1st Example of this invention. It is explanatory drawing of the example of the 2nd approximated curve of 1st Example of this invention. It is explanatory drawing of the example of the prediction curve calculated | required by the synthesis | combination of the 1st approximated curve and 2nd approximated curve of 1st Example of this invention. It is a processing flowchart of the example of the situation characteristic acquisition process of 1st Example of this invention. It is a processing flow of the example of the action recommendation process of 1st Example of this invention. It is explanatory drawing of the example of a display of the prediction curve of 1st Example of this invention. It is explanatory drawing of the example of a display of the action recommendation of 1st Example of this invention. It is a block diagram which shows the system configuration | structure which performs the usage-amount prediction and action recommendation of 2nd Example of this invention. It is a block diagram which shows the system configuration | structure which performs the usage-amount prediction and action recommendation of 3rd Example of this invention. It is a processing flow of the example of the various information acquisition process performed between the server of 3rd Example of this invention, and a terminal. It is explanatory drawing of the example of the display which shows a user the some recommendation action item of 4th Example of this invention. It is a block diagram which shows the system configuration | structure which performs the action recommendation which added the function which controls the external apparatus of the 6th Example of this invention.

  According to a typical embodiment of the present invention, a calculation unit, a display unit, a user (hereinafter referred to as a user) operation input means, a time acquisition unit, a usage amount and a usage amount history including time A usage amount acquisition unit that acquires information, an attribute acquisition unit that acquires attribute information in a period specified in advance, and a prediction status attribute acquisition unit that acquires a prediction status attribute from a prediction status attribute generation unit that predicts future attribute changes The first feature is that a resource usage amount is predicted by a system having a recording unit for recording various information.

  In addition, based on the predicted usage amount and a preset target value, the behavior to reduce the resource usage amount is given to the user by using the possible behavior information listing the possible behaviors given in advance by the user. The second feature is that the recommendation is presented.

  However, application of the present invention is not limited to this embodiment. The power that is the target of the usage amount prediction and behavior recommendation of the present invention is collectively referred to as “resource”.

  Hereinafter, embodiments according to the present invention will be described in detail with reference to the drawings.

  In the following embodiments, using the system in a retail store such as a supermarket, the usage amount of power is predicted, and the user is used when the power usage amount exceeds a preset target value. An example of recommending and presenting an action to reduce the usage fee to a retail store clerk, and updating the information using the action result when the user inputs the action result and using it for the next action recommendation explain.

  FIG. 1 is a configuration diagram of a system according to the first embodiment. The system 100 according to the present embodiment will be described as a device that predicts the amount of power used as a usage amount and recommends a user's action based on the transition of the predicted power amount. The system 100 includes a calculation unit 105, a clock unit 108 for acquiring time, an output unit 106 including various displays for displaying various displays to the user 180, a keyboard and a mouse for inputting user instructions, or touch panel means used with the display. This is realized by a computer having an input unit 107 including a CPU and a memory.

  Also, in this system, a usage sensor 110 that detects power usage is connected to the usage information acquisition unit 101 to acquire the usage. Note that the usage amount information acquisition unit 101 may calculate a total value, a maximum value, a minimum value, an average value, and a statistical value such as a variance of the usage amount in a specific period. Here, in addition to electric power, the usage sensor 110 has various values such as usage of gas and water and sewage, network communication, usage time of machines and electrical equipment, deposits, stock prices, exchange rates, and the like. It may be a means for detecting them.

  On the other hand, a situation attribute sensor 111 that detects weather, precipitation, temperature and humidity, wind direction and intensity, illuminance by indoor lighting, and the like is connected to the situation attribute information acquisition unit 102, and each value is detected and set. To get. Note that the status attribute information acquisition unit 102 may calculate a total value, a maximum value, a minimum value, an average value, and a statistical value such as a variance for a specific period of the status attribute. In addition to the above, the situation information acquired by the situation attribute sensor 111 and the situation attribute information acquisition unit 102 is a sensor device that detects an attribute value of an item that can affect the amount detected by the usage sensor 110 described above, May be a measurement result. For example, it may be a value that can be detected by the above-mentioned usage sensor, or in addition, the number of employees of corporate offices, the attributes of those employees, the number of customers in the retail store, gender, Adult / child distinction, customer flow rate that passes a point of interest within a certain period of time, whether or not the day is a business day, sales form such as sale or event, type and quantity of goods to be displayed, air conditioner and The set temperature and humidity of the refrigerator and heat insulation, the actual temperature, humidity, operating intensity and wind direction, the type and quantity of goods sold, the volume of noise and sound generated by equipment, the type and temperature and ratio of gases such as air, etc. Flow rate and type and flow rate of chemical substances and pollen, magnetic field and electric field strength, radiation type and amount, earthquake intensity and scale, power generation, gas and water supply, infection including cold and influenza Sexual illness Analyzes various types of data such as text, images, video, audio, and sound obtained from information sources including types and trends, trends in articles and trends, the World Wide Web and blogs on the Internet, and network services. It may be a means for detecting, measuring, and setting various attributes such as the type and quantity of information obtained.

  In addition, the system includes a predicted status attribute information generation unit 120 and a predicted status attribute information acquisition unit 104 that predict how the attribute values that can be detected, measured, and set by the status attribute sensor 111 will change in the future. It is assumed that the predicted situation attribute information specified in advance is acquired. The predicted status attribute information acquired by the predicted status attribute information generation means 120 is a future transition prediction or schedule of the above-mentioned various attribute values, for example, attribute information from an information provider related to weather forecast or weather, Schedules for power transmission from power plants and power companies, plans for upcoming events such as corporate offices and retail stores, schedules for employees, trend types and attribute forecasts, expert forecasts for stock prices and exchange rates, and various future simulations The change amount estimation information of the attribute value may be used.

  The system also has a recording unit 103 that records various types of information. FIG. 2 is an explanatory diagram of an example of various information held in the recording unit 103. Along with the above-described usage history information 201, situation attribute information 202, and predicted situation attribute information 203, the recording unit in this embodiment includes possible action information 204, target value information 205, situation attribute evaluation rule 206, action recommendation rule, which will be described later. 207, situation attribute evaluation result information 208, situation feature information 209, and action recommendation result information 210 are recorded. The various types of information described below are acquired from the sensors 110 and 111 and the predicted situation attribute information generation unit 120, the process of acquiring the rule and recording it in the recording unit 103 is the process of the present embodiment performed by the arithmetic unit 105. It may be performed immediately before the process, or may be performed using an acquisition unit (101, 102, 104) or the like in the process of acquiring each information.

(Description of each table)
FIG. 3 is an explanatory diagram of an example of the usage history information 201. It is assumed that the usage amount history information in this embodiment includes the elements of the item number 301, the location 302 where the usage amount is acquired, and the power usage amount 304. In addition, a part of the situation attribute information acquired by the situation attribute information acquisition unit 102 may be included as an element together with the usage history information in association with the date and time, and at the same time as shown in FIG. , Temperature 305, humidity 306, and room temperature 307 to 309 of the room as situation attributes, and room humidity 310. In FIG. 3, the date and time in each item is counted every 30 minutes from 0:00 on December 7, 2011, and the usage amount 304 is the total amount for 30 minutes from that time, and the temperature 306, humidity 307, room temperature 308- 309 and humidity 311 are calculated and held as an average value during the period.

  FIG. 4 is an explanatory diagram of the situation attribute information 202 and the predicted situation attribute information 203. These are collectively described as attribute information. The attribute information in this embodiment includes item number 401, location 402, date 403, day of the week 404, wind direction 405, wind force 406, weather 407, minimum temperature 408, maximum temperature 409, and business day information 410 of the corporate office. Event information 411 and the number of customers 412 are recorded on a daily basis. The row with the item number 9999 at the bottom is the prediction status attribute information 203 on the day of prediction from now on, that is, the prediction of the future attribute information or the schedule, and the other item numbers are past attribute information. 202 elements.

  FIG. 5 is an explanatory diagram of an example of the possible action information 204. The possible behavior information in this embodiment includes an item number 501, a classification 502, a message 503, a presentation period 504 including a start 511 and an end 512, a presentation minimum interval 505, a usage reduction rate 505, a temperature change amount 513, and an upper limit 514. And a lower limit 515, a preference 508, a previous execution date and time 509, and a previous result 510. Here, in this example, the start 511 and the end 512 of the presentation period 504 are date and time, and the conditions described later are illustrated as character strings. However, in each condition of year, month, day, and hour, “*” is not a number. Some parts indicate that all values are met. This enables expression that makes this field value unconditional, such as satisfying the condition if the date matches regardless of the year, or satisfying the condition at any time during the day. Indicates. Further, “*” at the upper temperature limit 514 is also unconditional, and it is possible to make unconditional in other elements as well. On the other hand, an element that has not yet entered a value is indicated by “-”. In other words, it shows elements of items that have not been selected by action recommendation from possible action information described later, or elements that have been selected but have not been set because the user has not acted. The value of the element is updated.

  FIG. 6 is an explanatory diagram of an example of the target value information 205. The target value information in this embodiment has each element of item number 601, date 602, and upper limit value 603. In item No. 0, the element of date 602 is “*”, but this value is selected as the default value so that it matches whenever the date on which the prediction is made does not match the date of another item, ie, the default value. The When the date on which the prediction is made matches the date of another item, the item is selected as a target value, and the upper limit value 603 is used as the target value.

  FIG. 7 is an explanatory diagram of an example of the situation attribute evaluation rule 206. The situation attribute evaluation rule is a rule used in the optimum date selection step S1303 of the usage amount prediction process S1212 (FIG. 13) described later. The situation attribute evaluation rule in this embodiment is composed of the elements of item number 701, name 702, weight 703, and evaluation algorithm 704. The evaluation algorithm 704 sets the value of the predicted status attribute information 203 (the column of the item number “9999” in FIG. 4) to “[element name] 1” and the status attribute information 202 to be compared (other than “9999” in FIG. 4). The item number value of the day of interest in the item number line) is “[element name] 2”, and the value (weight, value) of the item “i” of the situation attribute evaluation rule 206 in the item number “r” of the situation attribute information vri) is an algorithm that simulates the Basic language.

  Specifically, for example, in the “weather” of the item number “1” of the situation attribute evaluation rule 206, it is expressed as “weather 1.main” and “weather 1.sub” using the main weather and its assistance. And That is, when the weather is “clear and cloudy”, “weather 2.main” is “clear” and “weather 2.sub” is “cloudy”. In actual processing, the item number 401 of the weather 407 column with the item number 401 of the prediction status attribute 202 of FIG. 4 is “9999”, “weather 1.main” is set to “rain”, and “weather 1.sub”. Has no value. In addition, “sunny” in the item number “1” is “sunny” as “weather 2.main” and no value as “weather 2.sub”. When these two items are compared, the evaluation value vri is finally 0.0. At this time, if the weather of the item number “1” was “Sunny and rainy”, it would be “Sunny” as “Weather 2.main” and “Rain” as “Weather 2.sub”. , Assuming that “weather 1.main” and “weather 2.sub” match, vri = 0.5.

  The vri value obtained in this manner is multiplied by the weight of each item number, for example, 1.0 in the case of “weather”, to obtain a value relating to the item number. This rule 206 is used in the optimal day selection process in the usage amount prediction process described later. The situation attribute evaluation rule 206 is not only stored in the recording unit 103 in advance, but may be acquired from the outside by reading or communicating information by further including an external storage unit, a network unit, or the like in the system. In this case, it is possible to update to a new situation attribute evaluation rule.

  FIG. 8 is an explanatory diagram of an example of the behavior recommendation rule 207. The behavior recommendation rule is a rule used in behavior recommendation processing S1214 (FIG. 19) described later in behavior recommendation evaluation step S1905. The action recommendation rule in this embodiment is composed of the elements of item number 801, name 802, weight 803, and evaluation algorithm 804. The evaluation algorithm uses the name 1002 and the predicted value 1003 of the situation feature information 209 (FIG. 10) obtained in the usage amount prediction process described later, the classification 502 of each item number of the possible behavior information 204, the element names and their values. The algorithm for obtaining the value vui of the item “i” of the behavior recommendation rule in the item number “u” of the possible behavior information is shown by simulating the Basic language.

Specifically, in the evaluation algorithm of the name “reduction amount” with the item number “1”, the maximum value “100” from the predicted amount “100” of the item “excess amount” with the item number “8” of the situation feature information 209 described later. The absolute value obtained by reducing the product of “2100” and “10%”, which is the reduction rate of possible action information, is divided by the value obtained by adding 1 to the absolute value of the excess amount, and this value is reduced from 1.0. As a result, 1.0− | 100−2100 * 10/100 | / (| 100 | +1.0) = − 0.089 (= −9 / 101), and this value can be the rule item number “1”. The evaluation value vui is applied to the behavior information item number “1”. Expressed by the above general formula, if the predicted amount of excess is x, the maximum value is x _m , and the reduction rate is k, the evaluation value of the reduction amount is 1.0− | x−k · x _m | / ( | X | +1.0), k · x _m is the target value, | x−k · x _m | is the difference between the predicted value and the target value, and “1.0” in the numerator is 0 in the denominator. This is a correction value for not.

  The behavior recommendation rule 207 is not only stored in the recording unit 103 in advance, but may be acquired from the outside by reading or communicating information by further providing the system with an external storage means, a network, etc. It becomes possible to update to a new action recommendation rule.

  FIG. 9 is an explanatory diagram of an example of the situation attribute evaluation result information 208. The situation attribute result information is information that is generated and used when performing an optimum date selection process (S1303 in FIG. 13) in a later-described usage amount prediction process step (S1212 in FIG. 12 and FIG. 13). The situation attribute evaluation result information in this embodiment is composed of elements of rank 901, list item number 902, date 903, and evaluation value 904, and the list item number 902 corresponds to item number 401 of the situation attribute information 202 (FIG. 4). Correspondingly, the date also corresponds to the date 403. The evaluation value is obtained using the above-described situation attribute evaluation rule 206 as described later. In this embodiment, the evaluation values are listed in descending order of the values, and the highest item, that is, the list item number that gives the maximum value, or the date is the most suitable day on the day of the prediction, and the amount of use of that date Used for curve approximation of history information or usage prediction for the day.

  FIG. 10 is an explanatory diagram of an example of the situation feature information 209. The situation feature information is information generated and used in a situation feature acquisition processing step (S1213 in FIG. 12 and FIG. 18) described later. The situation feature information in this embodiment is composed of elements of an item number 1001, a name 1002, and a predicted value 1003, and each item is obtained by a usage amount prediction process (S1212 in FIG. 12 and FIG. 13). Parameters of the first approximate curve, parameters of the second approximate curve, time when the target value is exceeded, maximum predicted value, maximum There are predicted values including the excess of the target value of the value and the time of the maximum value.

  FIG. 11 is an explanatory diagram of an example of the behavior recommendation result information 210. The behavior recommendation result information is information that is generated and used in the behavior recommendation result generation processing (S1906) of the behavior recommendation processing (S1214 in FIG. 12 and FIG. 19) described later. The action recommendation result information in this embodiment is composed of elements of rank 1101, list item number 1102, message 1103, and evaluation value 1104. The list item number 1102 corresponds to the item number 501 of the possible action information (204 and FIG. 5), and the message 1103 similarly corresponds to the message 503. The evaluation value 1104 is obtained using the behavior recommendation rule 207 described above, as will be described later. In this embodiment, the evaluation values are listed in descending order of the values, and the elements such as messages related to the list item number that gives the highest value, that is, the maximum value, are presented to the user 180 using the output unit. Show the action to take.

(Description of each process)
Hereinafter, the process of a present Example performed by a calculating part is shown using such various information.

  FIG. 12 is a process flowchart of an example of the main process of this embodiment. The processing in this embodiment will be described as a general event-driven operation. Here, for the sake of simplicity, a description will be given by sequentially processing a display request event, a timer interrupt event, and an interrupt event from the input unit accompanying the processing. First, the timer is initialized by activation (S1201). This initializes the process S1202 for generating an interrupt event at regular intervals using the clock unit 108 of the system. Specifically, initialization is performed to generate an event every 30 minutes. Next, when an event occurs, event processing is performed (S1200).

  When the event is a display request event (S1203), a screen to be displayed on the output unit 106 is acquired and drawn (S1204). If the event is an input event from the input unit, in step S1221, the event is acquired and input as information for determining an item for the user to select or instruct for the currently operating process (S1222). If this is an instruction to end the operation (S1223), the main process is ended.

  If the event is a timer interrupt event (S1211), a usage amount prediction process S1212, a situation feature acquisition process S1213, and an action recommendation process S1214 are performed.

FIG. 13 is a process flow diagram of an example of the usage amount prediction process S1212. In the usage amount prediction process, the status attribute information 203 (FIG. 4) is first acquired (S1301), and the predicted status attribute information 204 (item number “9999” in FIG. 4) is acquired (S1302). Next, the optimum date is selected (S1303). The optimum date is selected using the above-described situation attribute evaluation rule 206 (FIG. 7) by using the evaluation value α _r for the item number r for each day of the situation attribute using the element value of the noticed prediction attribute information. By using the weight 703 (c _ri ) at the item number i (1 ≦ i ≦ S) and the value (v _ri ) obtained by the algorithm by the evaluation algorithm of each item, for example, the following equation is used.

  That is, the sum obtained by multiplying the weight and the evaluation value is set as the evaluation value of the day, and is associated with each date of the status attribute information 202 as the status attribute evaluation result information 208 (FIG. 9). Is given a ranking 901. The list item number 902 corresponding to the maximum evaluation value obtained in this way, that is, the list item number (r) 902 in which the rank 901 is first is selected as the optimum date. In the case of FIG. 9, the list item number “3” of the status attribute information 202 giving the evaluation value “245.9”, that is, “2011/12/10” is selected as the date.

  Next, usage history information 202 relating to the selected optimal date is acquired (S1304), and a first curve approximation is performed to approximate the change in the power usage 304, which is the usage amount of the day (S1305).

  FIG. 14 is an explanatory diagram of an example of the first approximate curve. From the usage amount history information 202, P data sets (48 in this case) are obtained as the usage amount wt every 30 minutes at time t.

  In FIG. 14, the horizontal axis represents time t, the vertical axis represents usage amount w, and the usage amount data string is represented as a bar graph of wti (1403). For example, a curve approximating a change in wt is expressed as an Nth order function f (t) with respect to t, and the data set is approximated as follows.

Using a so-called general least square method that minimizes the sum of squares of the error of the assumed amount wp = f (t), which is the amount of usage 304 in the usage amount history information 202 at the actual time t. A parameter of f (t), that is, a _i is obtained. For example, when N = 4, that is, when wt is approximated as a quartic function of t, the coefficient a _i (0 ≦ i ≦ 4) can be easily obtained by solving the least square method. Thus, the first approximate curve could be approximated as a curve representing the optimal day data in the usage history information (1404).

  Next, Q pieces of data that have already been acquired up to the current usage amount for the day of prediction are acquired (S1306). Thus, a second approximate curve that approximates the current day usage is obtained from the correlation between the first approximate curve and the obtained current usage (S1307).

  FIG. 15 is an explanatory diagram of an example of usage on the day. The time t1501 is taken on the horizontal axis, the usage w′1502 every 30 minutes is taken on the vertical axis, and the usage w′t1503 of the day at time t is shown by a bar graph.

  FIG. 16 is an explanatory diagram of an example of the second approximate curve. However, in this example, the approximation is a straight line. The horizontal axis indicates the value f (t) of the first approximate curve, the vertical axis indicates the amount of usage w ′ on the current day, and the following data string relating to time t is indicated by a point string 1603.

  These point sequences are assumed as follows with a curve approximating w′t as an M-order function g (f (t)) for f (t).

  The Mth order function g (f (t)) is a correction formula for f (t).

As with the first approximate curve, the parameter b _i (0 ≦ i ≦ M) can be easily obtained using the least square method. For example, when M = 1, that is, when approximating with a straight line as shown in FIG. 16, it can be obtained so as to be expressed by an intercept b ₀ (1606) and a slope b ₁ (1607) indicating a g (f (t)) straight line.

  In this way, a predicted curve to be obtained can be obtained by combining the obtained first approximate curve and the second approximate curve. That is, a second approximate curve is obtained by combining two maps of t → f (t) and y → g (y) (f (t) εy).

FIG. 17 is an explanatory diagram of an example of a prediction curve obtained by synthesizing the first approximate curve and the second approximate curve. The horizontal axis indicates time t (1701), the vertical axis indicates w and w ′ and the predicted usage w′p (1702), and the first approximate curve wp = f (t) is indicated by a dotted line (1704). A curve 1703 of w′p = g (f (t)), which is a combination of the first approximate curve and the second approximate curve, is a curve obtained by enlarging the first approximate curve by a factor of b1 and increasing it by b _0. Can be interpreted. The function amount prediction process S1212 is terminated using the function curve obtained by synthesizing the two approximate curves in this way as the use amount prediction curve at the time t of the day.

  Note that the orders N and M of the higher-order functions of the first and second approximate curves may be other than the exemplified N = 4 and M = 1, and the order is selected on the assumption of data dispersion and trend characteristics. However, it may be changed as appropriate. In addition, it does not necessarily have to be a high-order function as in this example. Spline curve, Bezier curve, circle equation, sine function and cosine with multiple periods and amplitudes It may be expressed by various functions including a (cos) function. Further, the parameters representing these approximate curves may be obtained by methods other than the least square method. For example, generally known parameter estimation and parameter determination methods including Fourier series, discrete Fourier series and discrete cosine series May be used.

  In order to further improve the accuracy of the approximation, when the orders N and M of the higher-order functions of the first and second approximate curves are increased and the number of parameters included in the approximate expression is increased, the parameters obtained so far cannot be used, It is necessary to solve the simultaneous linear equations again, and the error evaluation is also complicated. On the other hand, when the higher-order functions of the first and second approximate curves are expanded by Legendre Polynomials, which are standard orthogonal polynomials, the variables in the product of Legendre polynomials of different degrees are obtained when obtaining each parameter. If the sum is taken to be 0, there is no need to solve simultaneous linear equations for determining the parameters of the approximate expression, and even if the order of the approximate expression is increased to improve accuracy, the parameters are independent of other parameters. Can be obtained sequentially, and the error can be easily evaluated. However, the nth-order Legendre polynomial is a polynomial that includes powers of the nth and lower orders, and is an even function when n is an even number and an odd function when n is an odd number. In addition, when using a Legendre polynomial, it is necessary to convert the variable x in the domain a ≦ x ≦ b to the variable x ′ in the domain-1 ≦ x ′ ≦ 1.

  Returning to FIG. 12, next, the situation feature acquisition process S1213 is performed. FIG. 18 is a flowchart illustrating an example of the situation feature acquisition process S1213. The situation characteristic information is initialized (S1801), and the target value information 205 for the day is acquired from the recording unit 103 (S1802). When the usage amount is predicted (S1803), a prediction curve is acquired (S1804), and the maximum value of the prediction curve is solved analytically or algebraically to obtain the time and the maximum value at which the maximum usage amount is obtained (S1805, S1806). ). Further, the inclination of the current curve is acquired (S1807). In addition, the relationship between the prediction curve and the target value such as the time when the target value is exceeded and the slope at that time is acquired (S1808). Thereafter, the difference between the latest usage amount and the target value is acquired at the current time point when prediction is unnecessary (S1809), and the acquired value is set as the situation feature information 210 and the process ends (S1810).

Next, an action recommendation process is performed (S1214). FIG. 19 is a processing flow of an example of the behavior recommendation processing of this embodiment. First, initialization is performed (S1901), and possible behavior information 204, situation feature information 209, and behavior recommendation rules 207 are acquired from the recording unit 103 (S1902, S1903, S1904). The evaluation value β _u for the item number u of the possible behavior information is used as the weight 803 (the item number i (1 ≦ i ≦ V) of the behavior recommendation rule using the element value of the possible behavior information to be noticed and the situation feature information ( By using d _ui ) and the value (v _ui ) obtained by the algorithm, for example, an action recommendation evaluation value is obtained by the following equation (S 1905).

  That is, the sum of values obtained by multiplying the weight and the evaluation value is used as the evaluation value of the possible action, and the action recommendation result information 208 (FIG. 11) is associated with the message 503 of the possible action information 204 (FIG. 5). Ranking 1101 is assigned in descending order of the value 1104 (S1906). The list item number 501 of the possible behavior information corresponding to the maximum evaluation value obtained in this way, that is, the list item number (u) 1102 in which the rank 1101 is first is selected as the recommended action. In the case of FIG. 11, the list item number “3” of the possible behavior information giving the evaluation value “25.1”, that is, “decrease the set temperature by 1 ° C.” is selected as the message.

  Next, the selected possible action is drawn and displayed using the output unit 106 and presented to the user (S1907). FIG. 20 is an explanatory diagram of a display example of an expected curve, and FIG. 21 is an explanatory diagram of a display example of action recommendation. In this example, first, as shown in FIG. 20, the specific date and time, the target value and the maximum value, the respective times at that time, etc., based on the figure described in FIG. Is an example of drawing and displaying. Furthermore, FIG. 21 shows an example in which recommended actions are displayed together with buttons for allowing the user to select recommended actions. Here, using each item of the item number of the situation feature information 207 and the possible action information 204 selected as the recommended action, for example, the predicted time 2102, the time 2103 expected to be exceeded, the classification 2104, and the message of the recommended action 2105 is displayed. Further, three buttons, that is, an “execute” button 2111, an “impossible” button 2112 and an “other” button 2113 are displayed on the screen. These are used when the user 180 selects whether or not the recommended action has been performed or whether other candidates are recommended.

  Returning to FIG. 19, the result of having the user 180 select one of the three buttons (2111, 2112, 2113) displayed in FIG. 21 is acquired from the input unit 107, and the action of the user is acquired (S <b> 1908). If “other” 2113 is selected (S1911), the next point of the behavior recommendation result information 210 is selected (S1912), and the behavior recommendation display S1907 is performed again. When “execution” 2112 is selected (S1921), the preference level of the action in the possible action information 204 is increased (S1922). For example, an update such as multiplying the value of the preference level by 1.2 in a range where the preference level does not exceed 1 may be used. When “unreasonable” 2113 is selected (S1930), the preference level is decreased (S1931). For example, an update such as dividing the preference level by 1.2 may be used. In addition, each element of the number of times of presentation and the number of implementations, the number of times of refusal, and the number of other selections may be additionally recorded in the possible behavior information, and added according to the presentation and the user's selection, and the preference level may be obtained from these. Specifically, the preference level may be calculated by “the number of times of implementation / the number of times of presentation”. In addition, when multiple users are considered separately, the user who acted according to the recommended behavior is input so that somebody can be identified, and each parameter is recorded corresponding to each parameter, thereby changing the recommended behavior for each specific user. May be. In this way, the result presented as the recommended action and how the user selected and acted on the result are generated as a log and used separately (S1932). By analyzing this log separately, it is possible to make the behavior recommendation more convinced by the user by correcting the parameter for behavior recommendation.

  Finally, when the next prediction process is executed, the previous recommended action is acquired, and the usage amount history information 201 at this time, or the situation feature information 209 obtained in the usage amount prediction process S1212 and the situation feature acquisition process S1213. And the result is written in the previous result (510 in FIG. 5) of the possible behavior information and updated (S1933). For example, when the maximum value of the prediction curve does not exceed the target value, that is, when the excess amount 100 of the situation feature information 209 is 0 or less, the previous result is “◯” as a good action result of the usage amount reduction. Is described. On the other hand, when the situation does not change, “△” is written in the previous result, and when the numerical value increases and the situation gets worse, “×” is written in the previous result to update the possible action information. Of course, it may be an actual improvement amount instead of a symbol, or may be an evaluation method for a result defined separately, and is preferably defined in advance together with an action recommendation rule. When this previous result is evaluated by the action recommendation rule, feedback to the corresponding evaluation value will be applied, and the selection recommendation from the possible action information will change, so it is possible to recommend a more realistic action It becomes.

  In the following embodiment, an example in which the present system is implemented by an apparatus in the form of a general personal computer (PC) will be described.

  FIG. 22 is a block diagram illustrating a system configuration for performing usage amount prediction and behavior recommendation in the second embodiment. In the system 2200 of this embodiment, a CPU 2202, a memory 2201, a secondary storage unit 2203, a communication unit 2207, and a clock 108 are connected through a bus 2206. Further, an external storage unit 2204 is connected and a removable storage medium such as a floppy (registered trademark) disk, an optical medium, or a memory card can be used. Also connected to the bus are an output unit 2209 that enables connection of, for example, a display 2211 for various displays to the user 180, a keyboard 2212 and a mouse 2213 for inputting user instructions, or touch panel means used with the display. An input unit 2208 that enables the connection is connected. The communication unit 2207 is a USB or i. Enables data communication according to one or more standards from various network connections such as wired connection including Link, Ethernet (registered trademark), wireless LAN, WiMAX, Bluetooth, ZigBee (registered trademark), mobile phone network line, etc. Communicate data exchange with external devices and services.

  The communication unit 2207 connects to the usage sensor 110, the situation attribute sensor 111, and the predicted situation attribute generation unit 120 via a network using an appropriate standard, and exchanges data. Here, the network 2221 does not have to be a single network as shown in FIG. 22, and each network 2221 may be connected to another network.

  Using the communication unit, the usage history information 201, the situation attribute information 202, and the predicted situation attribute information 203 are acquired from the sensor and the generation unit connected in this manner, as in the first embodiment. However, unlike the first embodiment in which the recording unit 103 holds these pieces of information, these pieces of information are held unevenly or distributed in the external storage medium 2205 loaded in the memory 2201, the secondary storage unit 2203, and the external storage unit 2204. May be. These acquisition and management can be controlled by a program using a CPU and a bus.

  In such a system, the usage amount prediction method and the behavior recommendation method of this embodiment are loaded into the system as a program. Also, the possible behavior information 204, the target value information 205, the situation attribute evaluation rule 206, and the behavior recommendation rule 207 are loaded in the same manner. Specifically, for example, it may be acquired through the external storage medium 2205 loaded in the external storage unit 2204, or may be acquired in a form stored in advance in the secondary storage unit 2203, loaded into the memory 2201, and executed by the CPU 2202. . Further, the program may be acquired from a program providing service connected to the network using the communication unit 2207 and executed. By acquiring each information in the same manner, the information may be held unevenly or dispersedly in the external storage medium 2205 loaded in the memory 2201, the secondary storage unit 2203, and the external storage unit 2204.

  This program uses various information in the same manner as in the first embodiment, starting with the main process (FIG. 12), the usage amount prediction process S1212 (FIG. 13), the situation acquisition process S1213 (FIG. 18), and the action recommendation process S1214 (FIG. 19) etc., but the explanation is omitted.

  In the following embodiment, an example will be described in which the present system is implemented as two devices in the form of a general personal computer (PC), each as a server and a terminal used by a user.

  FIG. 23 is a block diagram illustrating a system configuration for performing usage amount prediction and behavior recommendation in the second embodiment. The system according to the present exemplary embodiment includes two servers 2300 and a terminal 2350 used by the user 180.

  In the server 2300, a CPU 2302, a memory 2301, a secondary storage unit 2303, a first communication unit 2307, a second communication unit 2310, and the clock 108 are connected through a bus 2306. In addition, an external storage unit 2304 is connected, and an externally removable storage medium such as a floppy disk, an optical medium, or a memory card can be used. The first communication unit is a USB or i. Enables data communication according to one or more standards from various network connections such as wired connection including Link, Ethernet, wireless LAN, WiMAX, Bluetooth, ZigBee, mobile phone network line, etc., and exchanges data with external devices and services Can communicate.

  A second communication unit 2310 is further connected to the bus 2306 of the server 2300 to enable data communication according to one or more standards as described above, and to connect to the communication unit of the terminal 2350 to perform data communication with the terminal.

  The first communication unit 2307 connects to the usage sensor 110, the situation attribute sensor 111, and the predicted situation attribute generation unit 120 via a network using an appropriate standard, and exchanges data. Here, the network 2221 may not be a single one as shown in FIG. 23, and each may be connected to another network.

  A terminal 2350 is connected to a CPU 2352, a memory 2351, a secondary storage unit 2353, a communication unit 2357, and a clock 2368 through a bus 2356. Further, a storage medium that can be taken out externally, such as a floppy disk, an optical medium, or a memory card, to which the external storage unit 2354 is connected may be used. For example, a display 2361 for connecting various displays to the user 180 can be connected to the terminal bus, for example, a keyboard 2362 and a mouse 2363 for inputting user instructions, or a touch panel used with the display. An input unit 2360 including means is connected. The communication unit 2357 is connected to the communication unit of the server 2300 according to at least one of the standards as described above, and performs data communication with the server.

  In the server 2300, the usage history information 201, the situation attribute information 202, and the predicted situation attribute information 203 are acquired from the sensor and the generation unit connected in this manner, using the first communication unit, as in the first embodiment. . However, unlike the first embodiment in which the information is stored in the recording unit 103, the information may be unevenly distributed or distributed in the external storage medium 2305 loaded in the memory 2301, the secondary storage unit 2303, and the external storage unit 2304. . These acquisition and management can be controlled by a program using a CPU and a bus.

  In such a system comprising the server 2300 and the terminal 2350, the usage amount prediction method and the behavior recommendation method of this embodiment are loaded as a program on the server or terminal. Also, the possible behavior information 204, the target value information 205, the situation attribute evaluation rule 206, and the behavior recommendation rule 207 are loaded in the same manner. Specifically, when loading these into a server, for example, they are acquired through the external storage medium 2305 loaded in the external storage unit 2304 or acquired in a form stored in advance in the secondary storage unit 2303 and loaded into the memory 2301. It may be executed by the CPU 2302. Further, it may be acquired from a program providing service connected to the network using the communication unit 2307 and executed. By acquiring each information in the same manner, the information may be held unevenly or distributed in the external storage medium 2205 loaded in the memory 2301, the secondary storage unit 2303, and the external storage unit 2304. The same applies when loading to a terminal.

  By using a program operating on the server 2300 and the terminal 2350 operating in the present system, various information is used in the entire system in the same manner as in the first embodiment, starting with the main processing (FIG. 12) and the usage amount prediction processing S1212 (FIG. 13). And processing steps such as status acquisition processing S1213 (FIG. 18) and behavior recommendation processing S1214 (FIG. 19) are performed. At this time, the processing performed on the server and the processing performed on the terminal may be unevenly distributed or shared, and it is only necessary to obtain the same processing order and processing results as in the first embodiment. I will omit the explanation.

  However, the terminal 2350 does not have at least the usage amount history information 201, the situation attribute information 202, and the predicted situation attribute information 203. Also, the transmission of results etc. is not defined. Therefore, the program on the terminal additionally executes processing acquired by the data communication from the server 2300 through the communication unit 2357. At that time, the server side program also performs processing corresponding to this.

  FIG. 24 is an example of a processing flow of various information acquisition processing executed between the server 2300 and the terminal 2350 in the present embodiment. This processing may be executed in advance at the initial stage of both programs, or may be executed in an information acquisition step in the processing. Here, an example is described in which various types of information are exchanged assuming a request (request) based on an XML (eXtensible Markup Language) sentence using HTTP (Hyper Text Transport Protocol) communication using a general web technology and data communication based on a response. To do.

  First, on the terminal side, a related information database is initialized (S245), and a data acquisition request statement for the server is generated as an XML statement (S2452). In this XML sentence, the information type to be acquired may be added, and additional information including a time acquisition range, necessary items, elements, and the like may be added. A session with the server is created according to the HTTP protocol (S2401, S2453), and a request statement is sent from the terminal to the server (S2454). When the server receives the request statement (S2403), it analyzes it (S2403), and when the request is acquisition of DB, that is, various information (S2404), it acquires any of various information held by the server (S2405). The information acquired in this way is converted into XML with a predefined grammar and structure and generated as a result reply XML part (S2406). A reply XML sentence is sent from the server to the terminal (S2408). After the terminal receives this (S2455), the HTTP session is terminated (S2408, S2456). The terminal analyzes the received reply sentence XML (S2457), extracts the original information (S2457), and updates the DB in the terminal (S2458).

  As described above, by acquiring various kinds of information also in the terminal, it is possible to share the processing part of the entire system also on the terminal 2350 and realize the usage amount prediction method and the behavior recommendation method. In addition, after all the processes are executed on the server 2300, only the behavior recommendation result information 210 may be sent to the terminal. In any case, as the presentation of the recommended action to the user 180 and the action acquisition from the user, in the process after the step S1907 of the action recommendation process S1214 (FIG. 19), most of various information is output on the terminal side. Since it is necessary to use in the unit 2359 and the input unit 2358, it is desirable to perform processing on the terminal side.

  In the following embodiment, a modification example in which the content of action recommendation evaluated by the present system is presented to the user and selected is shown. Similar to the first embodiment, each processing step up to behavior recommendation is performed, and the behavior recommendation result information in FIG. In the first embodiment, the action recommendation users are presented one by one from the top, but in this embodiment, a plurality of actions are recommended simultaneously.

  FIG. 25 is an explanatory diagram of an example of display for presenting a plurality of recommended action items to the user. The action recommendation result (210, FIG. 11) is obtained by performing the above-described evaluation on each item of the plurality of possible action information (204, FIG. 5), and arranging the evaluation values in descending order (however, in ascending order depending on the evaluation method). As a result, the number designated in advance from the top of the plurality of possible behavior information, three in FIG. 25, is selected and displayed. In the example shown in FIG. 25, the following three items are selected: “lower the air-conditioning temperature by 1 ° C.” (2502), “set the refrigerator temperature to“ medium ”” (2503), and “set the lighting to 80%” (2504). Then, the recommended action is presented to the user 180 from the output unit 106. The user selects one of these and presses one of the execution buttons (2512, 2513, 2514) corresponding to each. The corresponding execution button is pressed to generate a log of the action selected in the execution (S1921) of the user action acquisition (S1908) of the action recommendation process (FIG. 19) and update the preference level of the possible action information.

  On the other hand, when the user determines that the recommended action item is not appropriate among the three recommended action items and the implementation is impossible, the impossible button (2112) can be pressed. In addition, when the user wants to show other action recommendation, the other button (2113) can be pressed. In these cases as well, processing is performed corresponding to other determinations (S1911) or impossible (S1930) determinations in the behavior recommendation processing (FIG. 19).

  In the following embodiment, when the content of the action recommendation evaluated by this system is presented to the user, the selection method of the next recommended action in the first embodiment and the selection of the plurality of items or the recommended action of the second point in the fourth embodiment A modification of the selection method is shown. Similar to the first embodiment, the processing steps up to the behavior recommendation are performed, and the description of the behavior recommendation result information in FIG.

  The action recommendation result (210, FIG. 11) is obtained by performing the above-described evaluation on each item of the plurality of possible action information (204, FIG. 5), and arranging the evaluation values in descending order (however, in ascending order depending on the evaluation method). As a result, in the example of FIG. 11, there are a number of items for lowering the set temperature at the top, and for the user, even if the next point is selected, the same content is displayed, which may be inconvenient. is there. In order to avoid this, when acquiring the classification 502 relating to the item number of the possible action information (FIG. 5) corresponding to the list item number 1102, recording the already recommended action classification, and selecting the next point These recorded classifications may be changed and the next point may be selected and presented.

  For example, in the presentation example of FIG. 4 (FIG. 25), the classification of “reducing the air conditioning temperature by 1 ° C.” (2502) is “air conditioning”, and the classification of “set the refrigerator temperature to“ medium ”” (2503) is “refrigerator”. ”,“ Make lighting 80% ”(2504), such as“ Lighting ”, the classifications are different. Therefore, a plurality of possible actions are reclassified and evaluated in the possible action items for each classification. The values (1104 in FIG. 11) may be selected and presented in descending order.

  Further, the recommended action candidate to be the next point is displayed by selecting the other button (2113) to be pressed when the user does not desire the recommended action presented in the first and fourth embodiments, and the action is selected from among them. In doing so, it is also possible to select an action belonging to a category that has not been recommended so far, as described above. That is, when another button is pressed, not only the next point candidate but also an action belonging to another category can be selected.

  In addition, as a result of performing the recommended action as this classification, the impact on the user or system installation situation, for example, the degree of influence on the degree of attracting customers at a supermarket etc. due to temperature change, or work due to less lighting By classifying the actions according to the degree of influence such as a decrease in place or a reduction in work efficiency, it is also possible to give changes to recommended items and order of actions.

  In the following embodiment, a modified example of the system for executing one of recommended actions according to the user's selection or automatically executing the action recommended at the top regardless of the user's selection will be shown.

  FIG. 26 is a block diagram showing a system configuration for recommending an action to which a function for controlling an external device corresponding to the action is added. In this example, an external device control unit 2602 is further provided in the system of FIG. 1 to control the external device 2601 and acquire the status of the external device. At this time, the control and status acquisition of the external device is compliant with commonly used control standards such as ECHONET, Zigbee, LONWORKS, DMX512, DALI, BACnet, H-Link, etc. Connection schemes and protocols may be used.

  In the action recommendation display S1907 of the action recommendation process (S1214, FIG. 19) performed by the calculation unit 105, the user selects an action to be presented to the user, for example, “decrease the air conditioning by 1 ° C.” as shown in FIG. When 2111 is pressed, an external device can be actually controlled as a user action by performing a step of controlling the external device via the external device control unit 2602 after the action recommendation process execution determination S1921. Become. Further, based on this implementation result, it is possible to update the system internal state associated with actual behavior implementation and the status where the system is installed by performing each step of subsequent preference increase S1922 and behavior log generation S1932. it can.

  On the other hand, the system can automatically perform a predetermined action without performing a series of processing steps in which a recommended action is presented to the user and the user selects an action. Specifically, according to the behavior recommendation result (210, FIG. 11) generated as a result of the behavior recommendation result generation S1906 of the behavior recommendation process (S1214, FIG. 19), the external device is used by using the highest recommended behavior of the result. An action may be automatically performed by controlling an external device via the control unit 2602. Further, based on this implementation result, it is possible to update the system internal state associated with actual behavior implementation and the status where the system is installed by performing each step of subsequent preference increase S1922 and behavior log generation S1932. it can.

DESCRIPTION OF SYMBOLS 100: Usage amount prediction and action recommendation system, 101: Usage amount acquisition part, 102: Situation attribute acquisition part, 103: Recording part, 104: Prediction situation attribute acquisition part, 105: Calculation part, 106: Output part, 107: Input 110: Usage sensor 111: Situation attribute sensor 120: Predicted situation attribute generation means 180: User

Claims (14)

A resource usage prediction method using a computer having a storage unit and an input unit,
Obtain past usage history information measured in a period and interval specified in advance and status attribute information of the period and store it in the storage unit,
When a period for performing the prediction is specified, the status attribute information acquisition unit connected to the computer and the prediction status attribute information that is predicted or planned attribute information of the period is acquired from either of the input unit,
From the recording unit, select one situation attribute information that is most similar to the predicted genus information from the situation attribute information,
Obtaining a usage history from the usage history information for the same period indicated by the selected status attribute information;
Using the history of the usage history information to obtain a first approximate curve that gives an approximation of the usage with respect to time;
Between the value obtained from the first approximate curve and the usage amount of the history of the usage history information at each time of the history of at least one usage history information belonging to the period immediately before or within the period in which the prediction is performed Find a second approximation curve that gives an approximation of
A usage amount prediction method, wherein a curve obtained by synthesizing the first approximate curve and the second approximate curve is generated as a usage amount prediction curve. A usage prediction system,
A calculation unit, a display unit, an input unit for user operation, a time acquisition unit,
A usage history information acquisition unit that acquires usage history information consisting of usage and time;
A situation attribute information acquisition unit for acquiring situation attribute information in a period specified in advance;
A prediction status attribute information acquisition unit that acquires a prediction status attribute from a prediction status attribute generation unit that predicts future attribute changes;
A recording unit for recording various information,
The calculation unit selects a past period having an attribute similar to the attribute value of the prediction period, approximates the usage amount transition of the period with a first approximate curve, and determines the usage amount transition determined at the present time as the first A usage amount prediction system that predicts a usage amount by approximating a second approximation curve obtained by synthesizing values obtained from one approximation curve. The situation attribute information and the predicted situation attribute information are:
It includes at least one of the date and time, weather information, the activity status of the user, the activity status of the group to which the user belongs, the environmental status of the place to which the user belongs, and the status that can be a factor affecting the usage amount. The usage amount prediction method according to claim 1. The first and second approximate curves are
The usage amount prediction method according to claim 1, wherein the usage amount prediction method is displayed on the display unit together with one or more usage amount history information belonging immediately before or within the period in which the prediction is performed. A behavior recommendation method for use of resources using a computer having a storage unit, an input unit, and a display unit,
Select a historical time period with attributes similar to the forecast period attribute value,
Approximating the usage amount transition of the period with a first approximate curve,
A usage amount prediction curve is generated by approximating the usage amount transition determined at the present time with a second approximate curve obtained by combining values obtained from the first approximate curve,
Obtaining a possible behavior information set including at least one possible behavior information defined by user behavior that affects the usage amount;
Finding the prediction excess information including at least one of the time when the target threshold value of the usage amount given in advance and the usage amount prediction curve intersect, and the extreme value of the prediction curve,
Obtaining an evaluation value between the attribute information in a predetermined method using each attribute information that each item in the possible behavior information has, the prediction excess information, and the prediction status attribute information,
Select the possible behavior information using the evaluation value for each of the possible behavior information items,
An action recommendation method comprising presenting the contents of the item of the selected possible action information on the display unit.   6. The behavior recommendation method according to claim 5, wherein, instead of the possible behavior information, behavior history information including a user behavior that affects the usage amount and a change amount of the usage amount due to the behavior is used. Action recommendation method. In the behavior recommendation method,
Obtaining the user's response to the presented recommendation from the input unit,
When the response is included in one of the possible behavior information, at least one element from the response content, the presentation count history, the selection count history, the selection preference level, or the implementation date, The behavior recommendation method according to claim 5, wherein the behavior recommendation method is updated as possible behavior information or an element of the behavior history information and used for the evaluation. A calculation unit, a display unit, an input unit for user operation, a time acquisition unit,
A usage amount acquisition unit that acquires usage history information including usage amount and time;
A situation attribute acquisition unit for acquiring situation attribute information in a period specified in advance;
A prediction status attribute acquisition unit that acquires a prediction status attribute from a prediction status attribute generation unit that predicts future attribute changes;
A recording unit for recording various information,
The computing unit is
Select a historical time period with attributes similar to the forecast period attribute value,
Approximating the usage amount transition of the period with a first approximate curve,
A usage amount prediction curve is generated by approximating the usage amount transition determined at the present time with a second approximate curve obtained by combining values obtained from the first approximate curve,
Obtaining a possible behavior information set including at least one possible behavior information defined by user behavior that affects the usage amount;
Finding the prediction excess information including at least one of the time when the target threshold value of the usage amount given in advance and the usage amount prediction curve intersect, and the extreme value of the prediction curve,
Obtaining an evaluation value between the attribute information in a predetermined method using each attribute information that each item in the possible behavior information has, the prediction excess information, and the prediction status attribute information,
Select the possible behavior information using the evaluation value for each of the possible behavior information items,
Presenting the content of the selected item of possible action information on the display unit,
An action recommendation system characterized by this. In the behavior recommendation system,
The usage history information consisting of the usage amount and time, the status attribute information in a period designated in advance, and the predicted status attribute information from the predicted status attribute generation means for predicting future attribute changes, respectively via the network The behavior recommendation system according to claim 8, further comprising a communication unit to acquire.   A storage medium readable by a computer, which stores a program for executing the behavior recommendation method according to claim 5. In the behavior recommendation method,
When at least one of the recommended action items is displayed, the evaluation value is displayed in descending order or ascending order, and a log of the selected action is generated and a preference level of the possible action information is updated. The action recommendation method according to claim 5. In the behavior recommendation method,
When displaying at least one of the recommended action items, the item number of the possible action information is classified and recorded, and when the next item is selected, the recorded classification is changed and the next point is further changed. 6. The method according to claim 5, wherein the item is selected and displayed. In the behavior recommendation system,
Having an external device control unit for controlling external devices in response to actions,
9. The behavior recommendation according to claim 8, wherein the external device control unit controls the external device using the highest recommended behavior of the result according to a behavior recommendation result which is the selected possible behavior information. system. In the behavior recommendation system,
14. The behavior recommendation system according to claim 13, wherein after the control of the external device by the external device control unit, the log generation of the selected behavior and the preference level of the possible behavior information are updated.

Images