JP2011257149A - Air temperature prediction system, air temperature prediction method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To allow air temperature prediction with high accuracy.SOLUTION: Reference air temperature differences are stored in a reference air temperature difference database 133 in association with dates, and weather pattern information obtained by patterning general weather conditions is stored in a weather pattern database 134. An air temperature prediction part 116 generates forecast weather pattern information obtained by patterning a weather forecast, reads out a date (date of similar weather) of weather pattern information matching the generated forecast weather pattern, reads out a reference air temperature difference associated with the read date, and adds the read reference air temperature difference to a reference air temperature calculated on the basis of an air temperature model to calculate an air temperature predictive value.

Description

  The present invention relates to an air temperature prediction system, an air temperature prediction method, and a program.

  There is a great need for temperature prediction because the amount of river inflow into the reservoir due to electricity demand and snow melting is closely related to temperature. The Japan Meteorological Agency and others have announced temperature predictions, but since only the highest and lowest temperatures are released, it is difficult to know the temperature at each time. Therefore, a technique for predicting the temperature at each time using a forecast has been proposed (see Patent Document 1).

Japanese Patent No. 3462933

  However, since the method described in Patent Document 1 predicts the temperature at each time from the predicted temperature value of the next day according to the tendency of the change in temperature, the accuracy of the prediction is the predicted forecast It depends on the precision of the value.

  The present invention has been made in view of such a background, and an object thereof is to provide an air temperature prediction system, an air temperature prediction method, and a program capable of performing highly accurate temperature prediction.

  A main invention of the present invention for solving the above-described problems is a system for predicting air temperature, which is associated with a past date, calculated based on a predetermined air temperature model, and the reference air temperature. To store a reference temperature difference database that stores a reference temperature difference that is a difference from the actual temperature value on the date, and a first pattern that includes a weather word indicating the weather on the date in association with a past date. A pattern database, a weather forecast acquisition unit for acquiring a weather forecast including a weather word indicating weather on a future date, and a pattern for extracting the weather word from the weather forecast and generating a second pattern including the weather word From the generation unit and the pattern database, the date corresponding to the first pattern that is the same month as the future date and matches the second pattern The reference temperature difference corresponding to the weather similarity date is read from the reference temperature difference database, the reference temperature is calculated based on the temperature model, and the reference temperature difference is added to the reference temperature And a temperature prediction unit that calculates a predicted value of the temperature.

  In the temperature prediction system of the present invention, when the first and second weathers occur on a past date, the first pattern includes a first weather word indicating the first weather, A second weather word indicating the second weather, a time word indicating the ratio of the second weather to the day or the order of the weather are included, and the pattern generation unit includes a plurality of weather words in the weather forecast. If included, the weather word and the time word may be extracted from the weather forecast, and the second pattern including the weather word and the time word may be generated.

  In the temperature prediction system of the present invention, the time word may be “temporary”, “sometimes”, or “after”.

  In the temperature prediction system of the present invention, the pattern generation unit divides a character string obtained by deleting words other than the weather word and the time word from the weather forecast immediately after the weather word, so that only the weather word is obtained. Alternatively, a weather chunk that is a character string composed of the time word and the weather word may be generated, and a combination of the weather chunks may be generated as the pattern.

  Further, in the temperature prediction system of the present invention, the temperature model is a model for calculating a reference temperature using year as an explanatory variable, and the temperature prediction system stores a temperature actual value storing a temperature actual value on a past date. For each month and day of the past date, an estimation unit that calculates a regression coefficient and a constant for the explanatory variable based on the actual value and the temperature model of each year, and the month and day in association with the month and day A coefficient storage unit that stores a regression coefficient and the constant, and for each of the past dates, the actual value corresponding to the past date is read from the temperature actual result database, and the date corresponding to the month and day of the past date The regression coefficient and the constant are read from the coefficient storage unit, and the regression coefficient, the constant, and the year of the past date are applied to the temperature model. A reference temperature difference registration unit that calculates the reference temperature and registers the difference between the reference temperature and the actual value as the reference temperature difference in the reference temperature difference database in association with the past date. You may do it.

  In the temperature prediction system of the present invention, when there are a plurality of weather similar dates, the temperature prediction unit calculates a temperature obtained by adding the reference temperature difference to the reference temperature for each of the weather similar dates, It is also possible to sort the collected temperatures and use a predetermined percentile value as the predicted temperature value from the sorted temperatures.

  Another aspect of the present invention is a method for predicting air temperature, which is associated with a past date, calculated based on a predetermined air temperature model, and the air temperature on the date from the reference air temperature. A reference temperature difference database that stores a reference temperature difference that is a difference from a past actual value, and a pattern database that stores a first pattern including a weather word indicating weather on the date in association with a past date. A computer having a weather forecast including a weather word indicating weather at a future date, extracting the weather word from the weather forecast, generating a second pattern including the weather word, from the pattern database, Read out a weather similar date that is the same month as the future date and that corresponds to the first pattern that matches the second pattern, and that is similar to the weather Reading the reference temperature difference corresponding to the date from the reference temperature difference database, calculating the reference temperature based on the temperature model, and adding the reference temperature difference to the reference temperature to calculate a predicted temperature value And

  According to another aspect of the present invention, there is provided a program for predicting an air temperature, which is associated with a past date and calculated based on a predetermined air temperature model, and the date from the reference air temperature. A reference temperature difference database that stores a reference temperature difference that is a difference from an actual temperature value in the past, and a pattern database that stores a first pattern that includes a weather word indicating the weather on the date in association with a past date; Obtaining a weather forecast including a weather word indicating the weather on a future date in a computer, and extracting the weather word from the weather forecast to generate a second pattern including the weather word; From the pattern database, the same month as the future date and the date corresponding to the first pattern that matches the second pattern Reads a similar air date, reads the reference temperature difference corresponding to the weather similar date from the reference temperature difference database, calculates the reference temperature based on the temperature model, and adds the reference temperature difference to the reference temperature And calculating the predicted temperature value.

  Other problems and solutions to be disclosed by the present application will be made clear by the embodiments of the present invention and the drawings.

  According to the present invention, highly accurate temperature prediction can be performed.

It is a figure which shows the hardware constitutions of the temperature prediction system. It is a figure which shows the software structure of the temperature prediction system. It is a figure which shows the structural example of the weather general condition information memorize | stored in the weather general condition database. It is a figure which shows the structural example of the temperature trend coefficient database 132. FIG. It is a figure which shows the structural example of the reference | standard temperature difference database. It is a figure which shows the structural example of the weather pattern information memorize | stored in the weather pattern database. It is a figure which shows the flow of the calculation process of the temperature trend coefficient by the temperature trend coefficient determination part 112. FIG. It is a figure which shows an example of the table | surface which rearranged average temperature in order of the date and the year. Table 32 coefficients _{a d} and _{b d,} is a diagram for explaining a state where the table 33 of the temperature trend coefficient MAa _d and MAb _d are calculated. Is a graph of the reference temperature was calculated and the graph of the reference temperature was calculated by the moving average prior to the temperature model coefficients were applied a _d and b _d, the moving average after temperature trend factor MAa _d and temperature models to which the MAb _d of . It is a figure which shows the flow of a registration process of a reference | standard temperature difference. It is a figure which shows the flow of the registration process of weather pattern information. It is a figure which shows the structural example of the pattern table 161 used for the registration process of weather pattern information. It is a figure for demonstrating the specific example of the registration process of weather pattern information. It is a figure which shows the flow of the prediction process of temperature. It is a figure which shows an example of the weekly forecast information 51 concerning the weather forecast information for several days. It is a figure which shows the flow of the creation process of the weather pattern information based on a weather forecast. It is a figure which shows an example of the screen which displays the predicted value calculated | required by the prediction process of temperature.

  Hereinafter, the temperature prediction system 10 according to an embodiment of the present invention will be described. The temperature prediction system 10 of the present embodiment predicts the temperature based on a theoretical value (hereinafter referred to as “reference temperature”) determined by a statistical model (hereinafter referred to as “temperature model”) for determining the temperature. The temperature prediction system 10 of the present embodiment estimates a coefficient included in the temperature model from past actual temperature values. The temperature prediction system patterns past weather, stores the difference between the reference temperature and the past actual temperature value (hereinafter referred to as “reference temperature difference”) for each pattern, and patterns the weather forecast. The temperature is predicted by adding the reference temperature difference corresponding to the weather forecast pattern to the reference temperature.

  FIG. 1 is a diagram illustrating a hardware configuration of the temperature prediction system 10. The temperature prediction system 10 includes a CPU 101, a memory 102, a storage device 103, a communication interface 104, an input device 105, and an output device 106. The storage device 103 stores various programs and data, for example, a hard disk drive, a flash memory, a CD-ROM drive, or the like. The CPU 101 implements various functions by reading a program stored in the storage device 103 into the memory 102 and executing it. The communication interface 104 is an interface for connecting to the communication network 40. The communication interface 104 is constructed by, for example, Ethernet (registered trademark), a public telephone line network, or a wireless communication network. The input device 105 is, for example, a keyboard, a mouse, a touch panel, or a microphone that receives data input from a user. The output device 106 is, for example, a display, a printer, or a speaker that outputs data.

  FIG. 2 is a diagram illustrating a software configuration of the temperature prediction system 10. The temperature prediction system 10 includes a weather overview acquisition unit 111, a temperature trend coefficient determination unit 112, a reference temperature difference registration unit 113, a weather pattern registration unit 114, a weather forecast acquisition unit 115, a temperature prediction unit 116, a weather overview database 131, and a temperature trend. A coefficient database 132, a reference temperature difference database 133, and a weather pattern database 134 are provided. Note that the weather overview acquisition unit 111, the temperature trend coefficient determination unit 112, the reference temperature difference registration unit 113, the weather pattern registration unit 114, the weather forecast acquisition unit 115, and the temperature prediction unit 116 are stored in the storage device of the CPU 101 included in the temperature prediction system 10. This is realized by reading the program stored in 103 into the memory 102 and executing it. The weather overview database 131, the temperature trend coefficient database 132, the reference temperature difference database 133, and the weather pattern database 134 are realized as storage areas provided by the memory 102 and the storage device 103 provided in the temperature prediction system.

  The weather overview database 131 stores information (hereinafter referred to as “weather overview information”) indicating the weather and temperature results for past dates. FIG. 3 is a diagram illustrating a configuration example of weather overview information stored in the weather overview database 131. As shown in the figure, the weather overview information includes a character string representing the weather in the date, the average daily temperature, and the daytime time zone (06: 0 to 18:00) (hereinafter referred to as “weather overview”). The weather overview in the night time zone (18: 00 to the next day 06:00) is included. The weather overview information is generally provided by the Japan Meteorological Agency or weather companies. In addition to the above items, the weather overview information may include, for example, the highest temperature, the lowest temperature, the amount of precipitation, and the like.

  The weather overview acquisition unit 111 acquires weather overview information. The weather overview acquisition unit 111 can, for example, access a server operated by the Japan Meteorological Agency or a private weather company to acquire weather overview information. Moreover, you may make it the weather general condition acquisition part 111 receive the input of weather general condition information from a user. The weather overview acquisition unit 111 registers the acquired weather overview information in the weather overview database 131.

The temperature trend coefficient database 132 (corresponding to the “coefficient storage unit” of the present invention) is a coefficient (hereinafter referred to as “temperature trend coefficient”) included in a model for calculating a reference temperature (hereinafter referred to as “temperature model”). Is stored in association with the month and day. In this embodiment, the temperature model is expressed by the following equation.
Reference temperature _d = MAa _d + MAb _d × year (d: month and day)

FIG. 4 is a diagram illustrating a configuration example of the temperature trend coefficient database 132. As shown in the figure, temperature trends coefficient database 132 in association with the date, stores two temperature trend coefficient MAa _d and MAb _d.

The temperature trend coefficient determination unit 112 (corresponding to the “estimation unit” of the present invention) calculates the temperature trend coefficient. In the present embodiment, the temperature trend coefficient determination unit 112 determines the coefficient a _d for each month and day d based on the average temperature included in the weather overview information registered in the weather overview database 131 and the temperature model. estimated the b _d, it calculates the moving average for estimation the coefficient _{a d} and _{b d} a temperature trend factor MAa _d and MAb _d. Details of the temperature trend coefficient calculation process will be described later.

  The reference temperature difference database 133 stores a reference temperature difference for each date. FIG. 5 is a diagram illustrating a configuration example of the reference temperature difference database 133. As shown in the figure, the reference temperature difference database 133 stores reference temperature differences in association with dates.

  The reference temperature difference registration unit 113 calculates the reference temperature difference and registers it in the reference temperature difference database 133. The reference temperature difference registration unit 113 calculates the reference temperature by substituting the date of the weather overview information into the temperature model to which the temperature trend coefficient is applied, and calculates the difference between the reference temperature and the average temperature of the weather overview information. The reference temperature difference is calculated, and the calculated reference temperature difference is registered in the reference temperature difference database 133 in association with the date of the weather overview information.

  The weather pattern database 134 (corresponding to the “pattern database” of the present invention) is information (hereinafter referred to as “weather pattern information”) in which weather conditions are patterned for each date. Corresponding). FIG. 6 is a diagram illustrating a configuration example of weather pattern information stored in the weather pattern database 134. The weather pattern information includes a term indicating the weather (hereinafter referred to as “weather word”) in association with the date, and a term indicating the time occupied by the weather in one day or the occurrence of the weather (hereinafter referred to as “time word”). ”) For each combination (hereinafter referred to as“ weather pattern ”). In this embodiment, the weather word is one of “sunny”, “cloudy”, “snow”, “rain”, and “heavy rain”, and the time word is “sometimes”, “temporary”, or “after”. Shall. The time word includes a null character (indicating that no time is specified). In this embodiment, the weather pattern is “sunny” “cloudy” “snow” “rain” “rainy” “sometimes heavy rain” “temporarily heavy rain” “after heavy rain” “sometimes rain” “temporary rain” “after rain” “sometimes Snow, “temporary snow”, “after snow”, “sometimes cloudy”, “temporarily cloudy”, “after cloudy”, “sometimes sunny”, “temporarily sunny” and “aftergoing”

  The weather pattern registration unit 114 generates weather pattern information based on the weather overview information and registers it in the weather pattern database 134. The details of the weather pattern information registration process will be described later.

  The weather forecast acquisition unit 115 acquires information describing a weather forecast (hereinafter referred to as “weather forecast information”). The weather forecast acquisition unit 115 may acquire weather forecast information from, for example, the Japan Meteorological Agency or a weather company, or may receive input of weather forecast information from the user. For example, the weather forecast acquisition unit 115 may periodically acquire weather forecast information or may acquire a weather forecast in response to an input from the user. The weather forecast information includes a date, a predicted value of temperature on that date (hereinafter referred to as “forecast temperature”), and a weather forecast on that date. Unlike the weather overview information, the weather forecast describes the weather on a daily basis. Further, in the weather forecast, the weather and a time word as necessary are described as in the weather overview.

  The temperature predicting unit 116 (corresponding to the “pattern generating unit” and “temperature predicting unit” of the present invention) corrects the forecast value of the temperature based on the weather forecast according to the reference temperature difference on a similar weather day. Predict the temperature. Details of the temperature prediction process by the temperature prediction unit 116 will be described later.

Hereinafter, the process performed by the temperature prediction system 10 of this embodiment is demonstrated.
First, the temperature trend coefficient registration process, the reference temperature difference registration process, and the weather pattern information registration process will be described.

== Registration process of temperature trend coefficient ==
FIG. 7 is a diagram showing a flow of temperature trend coefficient calculation processing by the temperature trend coefficient determination unit 112. Note that the temperature trend coefficient calculation process may be performed only once after the weather overview information for the past predetermined period is registered in the weather overview database 131, or the weather overview information is registered in the weather overview database 131. It may be performed every time. Moreover, you may make it perform regularly.

The temperature trend coefficient determination unit 112 reads the weather overview information registered in the weather overview database 131, removes the weather overview information that was read on February 29 (S201), and calculates the average temperature of the weather overview information. Are rearranged in order of month and day, for example, as shown in FIG. 8 (S202). In the example of Table 31 shown in FIG. 8, the average temperatures for each year are arranged with the date as the row. The temperature trend coefficient determination unit 112 estimates coefficients a _d and b _d for each month and day using the average temperature of each year and the temperature model (S203). In the table 32 shown in FIG. 8, a _d and b _d corresponding to each month and day are displayed. Then the temperature trend coefficient determining unit 112, the coefficient _{a d} and _{b d,} the following equation to calculate the result of moving averaged as temperature trend factor MAa _d and MAb _d (S204). In this embodiment, N = 7.

ｄ＝２／２９の場合、

In the case of d = 1/1-2/28, 3/1-12/31,

If d = 2/29,

In this way, as shown in FIG. 9, Table 32 coefficients _{a d} and _{b d,} Table 33 of the temperature trend coefficient MAa _d and MAb _d are calculated. Figure 10 is a reference temperature calculated by the moving average prior to the coefficients a _d and the graph of the reference air temperature b _d is calculated by the applied temperature model, temperature model according to the temperature trend coefficient MAa _d and MAb _d after the moving average It is a graph of. As shown in FIG. 10, by standardizing the temperature trend coefficient with a moving average, the reference temperature calculated from the temperature model to which the coefficient is applied is also leveled.

As described above, when the temperature trend factor MAa _d and MAb _d is calculated, temperature trend coefficient determining unit 112 registers the temperature trend coefficient MAa _d and MAb _d in temperature trend coefficient database 132 (S205).

As described above, the temperature trend factor MAa _d and MAb _d for each month is registered in the temperature trend coefficient database 132.

== Registration process of reference temperature difference ==
FIG. 11 is a diagram illustrating a flow of registration processing of the reference temperature difference. It is assumed that the reference temperature difference registration unit 113 executes the process of FIG. 11 every time the weather overview information is registered in the weather overview database 131.

Reference temperature difference registration unit 113 reads the temperature trend coefficient MAa _d and MAb _d corresponding to the date of the date contained in the weather overview information from the temperature trend coefficient database 132 (S221), the read temperature trend factor and Weather The standard temperature is calculated by applying the year of the date included in the overview information to the above temperature model (S222), and subtracting the calculated reference temperature from the average temperature included in the weather overview information (S223), and the calculated reference temperature difference is registered in the reference temperature difference database 133 in association with the date included in the weather overview information (S224).

== Weather pattern information registration process ==
FIG. 12 is a diagram showing a flow of weather pattern information registration processing, and FIG. 13 is a diagram showing a configuration of a table (hereinafter referred to as “pattern table 161”) used in the weather pattern information registration processing. .

As shown in FIG. 13, the pattern table 161 is a table with weather as a row and time words as a column. Further, the matrix of the pattern table 161 is arranged so that items having a high degree of temperature are on the upper and left sides. In the example of FIG. 13, the rows are arranged in the order of “clear”, “cloudy”, “rain”, “heavy rain”, and “snow”, and the columns are arranged in the order of “sometimes” “temporary” “after”.
The registration process shown in FIG. 12 is executed for the acquired weather overview information every time the weather overview acquisition unit 111 acquires the weather overview information.

  The weather pattern registration unit 114 generates a pattern table 161 in which “0” is set in each cell (S401). The weather pattern registration unit 114 performs the following processing for each of the daytime weather overview and the nighttime weather overview included in the weather overview information.

  The weather pattern registration unit 114 deletes words other than weather words and time words from the weather overview (S402), divides the weather overview immediately after the term indicating weather, and sets the divided words as weather chunks (S403). . The weather pattern registration unit 114 sets the value of the pattern table 161 corresponding to the time word and the weather included in the weather chunk to “1” for each divided weather chunk (S404).

  The weather pattern registration unit 114 performs the above processing for each of the weather conditions in the daytime period and the weather conditions in the nighttime period, and then sets “1” for each row of the pattern table 161. The leftmost cell is searched, and “0” is set in all the cells to the right of the searched cell (S405). For each column of the pattern table 161, the weather pattern registration unit 114 searches for the uppermost cell in which “1” is set, and sets “0” to all the cells below the searched cell. Is set (S406).

  The weather pattern registration unit 114 creates weather pattern information including a combination of a time word and a weather word and a date from the pattern table 161 created as described above, and registers the weather pattern information in the weather pattern database 134 (S407). .

  FIG. 14 is a diagram for explaining a specific example of the registration process. In the example of FIG. 14, the weather overview 1711 for the daytime period included in the weather overview information 171 is “cloudy after a cloudy rain”, and the weather overview 1712 for a nighttime zone is “temporarily cloudy after the sky”. . The weather pattern registration unit 114 divides the daytime weather overview 1711 immediately after the weather word, that is, after the “cloudy”, “rainy”, and “sunny”, and displays “cloudy”, “temporarily raining”, and “afternoon”. A weather chunk is created (S403). Next, the weather pattern registration unit 114 corresponds to a weather word and a time word included in each weather chunk, that is, a cell 1721 corresponding to “cloudy” and an empty character, and a cell corresponding to “rain” and “temporary”. 1722, and “1” is set in the cell 1723 corresponding to “clear” and “after” (S404). Similarly, the weather pattern registration unit 114 also divides the weather overview 1712 for the night time zone into weather chunks of “sunny” and “temporarily cloudy after” (S403), and cells corresponding to “sunny” and sky characters. 1731, and “1” is set in the cells 1732 and 1733 corresponding to “cloudy”, “temporary”, and “after” (S404). The weather pattern registration unit 114 sets “0” in other cells, leaving only the leftmost “1” in each row of the pattern table 161 (S405). In the row 174, the leftmost cell 1741 among the cells in which “1” is set is left, and “0” is set in the other cells 1742 and 1743. The weather pattern registration unit 114 sets “0” in other cells, leaving only the top “1” in each column of the pattern table 161 (S406). In the column 175, the cell 1751 at the top of the cells set with “1” is left, and “0” is set in the other cells 1752. When the above processing is performed, in the pattern table 161 in FIG. 14, “1” is set only in the cell 1751 corresponding to “clear” and blank, and the cell 1753 corresponding to “rain” and “temporary”. It becomes. Based on the pattern table 161, weather pattern information 177 in which “1” is set only for “cloudy” and “temporary rain” is generated and registered in the weather pattern database 134.

== Prediction processing of temperature ==
Next, temperature prediction processing will be described. FIG. 15 is a diagram showing the flow of temperature prediction processing. The process of FIG. 15 is executed every time the weather forecast acquisition unit 114 acquires weather forecast information. FIG. 16 is a diagram illustrating an example of information related to a weekly forecast (hereinafter referred to as “weekly forecast information 51”), which is an example of weather forecast information for a plurality of days. In the example of FIG. 16, the temperature prediction unit 116 performs the processing described below for the weather forecast information for each date from June 16, 2009 to June 22, 2009, which is included in the weekly forecast information 51. Do.

  The temperature prediction unit 116 acquires a temperature trend coefficient corresponding to a date (hereinafter referred to as “forecast date”) included in the weather forecast information from the temperature trend coefficient database 132 (S421), and the acquired temperature trend coefficient and The reference temperature is calculated by applying the year of the forecast date to the temperature model (S422). Next, the temperature prediction unit 116 creates information obtained by patterning the weather forecast included in the weather forecast information (hereinafter referred to as “forecast weather pattern information”) (S423). FIG. 17 is a diagram showing the flow of the process of creating weather pattern information based on the weather forecast.

  The temperature prediction unit 116 creates the pattern table 161 of FIG. 13 and sets “0” in all the cells (S441). The temperature prediction unit 116 deletes words other than weather words and time words from the weather forecast (S442), divides the weather forecast immediately after each weather word, and sets the divided character strings as weather chunks (S443). . For each weather chunk, the temperature predicting unit 116 sets “1” in the cell of the pattern table 161 corresponding to the time word and the weather word included in the weather chunk (S444). The temperature prediction unit 116 selects, for each row of the pattern table 161, the leftmost cell among the cells in which “1” is set, and sets “0” to all the cells to the right of the selected cell. (S445). In addition, the temperature prediction unit 116 selects, for each column of the pattern table 161, the uppermost cell among the cells in which “1” is set, and sets “0” to all the cells below the selected cell. "Is set (S446). The temperature prediction unit 116 extracts the cell value corresponding to each combination of the time designation word and the weather word from the pattern table 161, and creates forecast weather pattern information in which the extracted value is set (S447). ).

  Returning to FIG. 15, the temperature predicting unit 116 includes the date included in the weather forecast information among the dates included in the weather pattern information that matches the weather pattern included in the generated forecast weather pattern information from the weather pattern database 134. Are searched for the same month (S424). In the present embodiment, a search is made for a pattern in which the weather pattern included in the forecast weather pattern information matches the weather pattern included in the weather pattern information. The temperature prediction unit 116 acquires a reference temperature difference corresponding to the searched date from the reference temperature difference database 133 (S425). The temperature prediction unit 116 sets the number of searched dates to m (S426). The temperature prediction unit 116 calculates the temperature for each searched date by adding the reference temperature difference corresponding to the date to the reference temperature (S427). The temperature prediction unit 116 sets the calculated average temperature as the predicted average temperature (S428). The temperature predicting unit 116 sorts the calculated temperatures in descending order (S429), and sets a value obtained by multiplying m by a predetermined percentile setting value (converted to an integer value) as n (S430). The percentile setting value may be designated from the user or may be a predetermined value. Further, n may be calculated for a plurality of percentile values. The temperature prediction unit 116 sets the nth temperature from the top of the sorted temperatures as the predicted temperature (S431). Note that the temperature prediction unit 116 may use the predicted average temperature as the predicted temperature, for example.

  As described above, the predicted temperature is calculated based on the weather forecast information. In the example of FIG. 18, it is assumed that the weekly forecast information 51 of FIG. 16 is acquired, and five percentile setting values of 10%, 30%, 50%, 70%, and 90% are set. . The number of similar data is m obtained in step S426, and is the number of dates corresponding to “June” and the weather pattern in which “sunny” of the weather forecast included in the weather forecast information on June 16 is patterned. is there. The average value is the predicted average temperature obtained in step S428. The 10% value, the 30% value,... 90% value are the 10th percentile value when the temperatures calculated in step S427 are arranged in descending order, 30 Percentile value: 90th percentile value, which is the predicted temperature obtained in step S431 when the percentile setting value is set to 0.1, 0.3, 0.5, 0.7, 0.9.

  As described above, in the temperature prediction system 10 of this embodiment, past weather conditions are patterned for each date, and a reference temperature difference that is a difference between the reference temperature calculated by the temperature model and the actual temperature value is calculated. By storing each date and patterning future weather forecasts, adding the reference temperature difference for past dates in the same pattern to the reference temperature, the reference temperature obtained by the temperature model is It can be adjusted by the difference. Since the correlation between the temperature and the weather is considered to be high, it is possible to predict the temperature with high accuracy by adjusting the reference temperature according to the weather.

  Moreover, in this embodiment, while matching a pattern, the reference temperature is adjusted using the reference temperature difference corresponding to the date of the same month. Since it is considered that the temperature is close if it is the same weather in the same month, it is expected that a closer temperature is calculated according to the temperature prediction system 10 of the present embodiment.

  In the present embodiment, the character strings included in the weather summary are patterned and stored, and the weather forecasts are also patterned and compared in the same manner. Can be extracted easily.

  In the temperature prediction system 10 according to the present embodiment, the date matching the forecast weather pattern information is searched for the same month as the date included in the weather forecast information. , Including typhoon information indicating whether or not a typhoon has occurred in the forecast area, rainy season information indicating whether or not a rainy season has occurred, and El Nino information indicating whether or not an El Nino has occurred Of the dates that match the forecast weather pattern information and the typhoon information, rainy season information, El Nino information, etc., may be searched for the same month as the forecast date. If the typhoon is not occurring or not, if the rainy season is not occurring or not Elniño is not occurring, the temperature may be different, so in addition to the weather pattern, the typhoon or rainy season, It is expected that the prediction accuracy of the temperature calculated using the corresponding reference temperature difference is increased by performing matching using an event that affects the temperature other than the weather such as El Nino. In this case, for example, when the weather overview information or the weather forecast information includes a flag value indicating whether or not a typhoon is occurring, a flag value indicating whether or not El Nino is occurring, and the like, These flag values can be added to the weather pattern information and the forecast weather pattern information, and matching can be performed including these flag values.

  In this embodiment, the weather is assumed to be one of “sunny”, “cloudy”, “snow”, “rain”, and “heavy rain”, but may include, for example, “sleet” or “hail”. . In the present embodiment, the time word is assumed to be “sometimes”, “temporary”, or “after”, but may include, for example, “gradually”, “beginning”, “depending on location”, and the like.

  In the temperature prediction system 10 of the present embodiment, the weather overview information includes the weather overview of the day time zone and the night time zone, but not limited to this, only the weather overview of the day. May be included, or weather conditions for three or more time zones such as morning, afternoon, and night may be included.

  In the present embodiment, a weather pattern is created for past weather summary information, but past weather forecast information may be used instead of past weather summary information. A database for storing weather forecast information may be provided.

  In this embodiment, the weather summary information includes the average temperature. However, when only the maximum temperature and the minimum temperature are included, the average value is calculated as the average temperature. It may be.

  Further, in this embodiment, it is determined whether or not the weather pattern related to the past weather overview matches the weather pattern related to the future weather forecast, but the comparison is made by various pattern matching. Also good. Further, based on the weather pattern, a decision tree having a reference temperature difference as a leaf with a weather word or a time word as a node may be created, and the reference temperature difference may be obtained by tracing the decision tree.

  In the present embodiment, weather words and time words are arranged in descending order of the degree of temperature in the matrix of the pattern table 161. For example, the degree related to the temperature in association with the weather word and the time word. You may make it provide the priority memory | storage part which memorize | stores (priority). In this case, the weather pattern registration unit 114 generates a list of weather chunks in step S403 of FIG. 12 without using the pattern table 161, and the degree corresponding to the included weather word for each of the weather chunks included in the list The degree corresponding to the included time word is acquired from the priority storage unit, and the total value of the acquired degrees is set as the priority of the weather chunk. The weather pattern registration unit 114 identifies, for each weather word, a weather chunk that includes the weather word, leaves the highest priority among the identified weather chunks, and deletes the others from the list. The weather pattern registration unit 114 can create weather pattern information and register it in the weather pattern database 134 based on the weather chunks remaining in the list.

  In the present embodiment, the weather pattern database 134 registers a flag value corresponding to a date and a weather chunk. However, the present invention is not limited to this. For example, a weather chunk list is registered in association with a date. You may make it do. The weather pattern database 134 may also store the average temperature included in the weather summary information in association with the date.

  In this embodiment, in step S427 in FIG. 15, the predicted temperature value is calculated by adding the reference temperature difference to the reference temperature calculated by the temperature model. However, the average temperature is included in the weather forecast information. If so, the reference temperature difference may be added to the average temperature included in the weather forecast information. Further, when the weather forecast information includes the maximum temperature and the minimum temperature, the reference temperature difference can be added to the average value of the maximum temperature and the minimum temperature.

  Further, in step S424 in FIG. 15, all the dates of the weather pattern information that match the forecast weather pattern information may be searched.

  In the present embodiment, the average temperature is predicted, but the maximum temperature and the minimum temperature can be predicted instead.

  Although the present embodiment has been described above, the above embodiment is intended to facilitate understanding of the present invention and is not intended to limit the present invention. The present invention can be changed and improved without departing from the gist thereof, and the present invention includes equivalents thereof.

10 Temperature prediction system 101 CPU
DESCRIPTION OF SYMBOLS 102 Memory 103 Memory | storage device 104 Communication interface 105 Input device 106 Output device 111 Weather general condition acquisition part 112 Temperature trend coefficient determination part 113 Reference | standard temperature difference registration part 114 Weather pattern registration part 115 Weather forecast acquisition part 116 Temperature prediction part 131 Weather general condition database 132 Temperature trend coefficient database 133 Standard temperature difference database 134 Weather pattern database

Claims (8)

A system for predicting temperature,
A reference temperature difference database that stores a reference temperature difference that is a difference between a reference temperature calculated based on a predetermined temperature model in association with a past date and an actual value of the temperature on the date from the reference temperature;
A pattern database for storing a first pattern including a weather word indicating weather on the date in association with a past date;
A weather forecast acquisition unit for acquiring a weather forecast including a weather word indicating the weather on a future date;
A pattern generation unit that extracts the weather word from the weather forecast and generates a second pattern including the weather word;
Read out from the pattern database a weather similar date that is the same month as the future date and corresponds to the first pattern that matches the second pattern, and the reference corresponding to the weather similar date A temperature prediction unit that reads a temperature difference from the reference temperature difference database, calculates the reference temperature based on the temperature model, and calculates a predicted value of the temperature by adding the reference temperature difference to the reference temperature;
A temperature prediction system characterized by comprising: The temperature prediction system according to claim 1,
When the first and second weathers occur on a past date, the first pattern includes a first weather word indicating the first weather and a second weather indicating the second weather. Word, a time word indicating the percentage of the second weather in the day or the order of the weather,
When the weather forecast includes a plurality of the weather words, the pattern generation unit extracts the weather words and the time words from the weather forecast, and includes the weather words and the time words. Generating a pattern of
A temperature prediction system characterized by The temperature prediction system according to claim 2,
The time word is one of “temporary”, “sometimes” and “after”;
A temperature prediction system characterized by The temperature prediction system according to claim 2,
The pattern generation unit divides a character string obtained by deleting words other than the weather word and the time word from the weather forecast immediately after the weather word, and only the weather word or the time word and the weather word Generating a weather chunk that is a character string composed of: and generating a combination of the weather chunks as the pattern;
A temperature prediction system characterized by The temperature prediction system according to claim 1,
The temperature model is a model for calculating a reference temperature using year as an explanatory variable,
The temperature prediction system is
A temperature record database that stores the actual temperature values for past dates;
For each month and day of the past date, an estimation unit that calculates a regression coefficient and a constant for the explanatory variable based on the actual value and the temperature model of each year;
A coefficient storage unit for storing the regression coefficient and the constant in association with a date;
For each of the past dates, the actual value corresponding to the past date is read from the temperature actual result database, and the regression coefficient and the constant corresponding to the month and day of the past date are read from the coefficient storage unit, The reference temperature is calculated by applying the regression coefficient, the constant, and the year of the past date to the temperature model, and the difference between the reference temperature and the actual value is used as the reference temperature difference to correspond to the past date. And a reference temperature difference registration unit for registering in the reference temperature difference database,
A temperature prediction system characterized by comprising: The temperature prediction system according to claim 1,
When there are a plurality of weather similar dates, the temperature prediction unit calculates a temperature obtained by adding the reference temperature difference to the reference temperature for each of the weather similar dates, sorts the calculated temperatures, and sorts the sorted temperatures A predetermined percentile value as a predicted value of the temperature,
A temperature prediction system characterized by A method for predicting temperature,
A reference temperature difference database that stores a reference temperature difference that is a difference between a reference temperature calculated based on a predetermined temperature model in association with a past date and an actual value of the temperature on the date from the reference temperature; A computer having a pattern database that stores a first pattern including a weather word indicating weather on the date in association with a past date,
Get a weather forecast that includes a weather word that indicates the weather for a future date,
Extracting the weather word from the weather forecast, generating a second pattern including the weather word;
Read out from the pattern database a weather similar date that is the same month as the future date and corresponds to the first pattern that matches the second pattern, and the reference corresponding to the weather similar date Reading a temperature difference from the reference temperature difference database, calculating the reference temperature based on the temperature model, and adding the reference temperature difference to the reference temperature to calculate a predicted value of the temperature;
A temperature prediction method characterized by A program for predicting temperature,
A reference temperature difference database that stores a reference temperature difference that is a difference between a reference temperature calculated based on a predetermined temperature model in association with a past date and an actual value of the temperature on the date from the reference temperature; In a computer having a pattern database that stores a first pattern including a weather word indicating weather on the date in association with a past date,
Obtaining a weather forecast including a weather word indicating weather at a future date;
Extracting the weather word from the weather forecast and generating a second pattern including the weather word;
Read out from the pattern database a weather similar date that is the same month as the future date and corresponds to the first pattern that matches the second pattern, and the reference corresponding to the weather similar date Reading a temperature difference from the reference temperature difference database, calculating the reference temperature based on the temperature model, and adding the reference temperature difference to the reference temperature to calculate a predicted value of the temperature;
A program for running

Images