JP2018156468A - Deviation determination device, day type discrimination device, prediction device, program, and deviation determination method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a deviation determination device which can allow a small degree of deviation from a standard reference pattern concerning time series load data, and the like.SOLUTION: A deviation determination device of one embodiment includes a determination object data storage part, a reference data storage part, an allowable data storage part and a deviation determination part. The determination object data storage part stores determination object data being time series data of a load value of a determination object. The reference data storage part stores reference data representing a reference related to a variation pattern of a load. The allowable data storage part stores allowable data representing an allowance when the variation pattern of the load deviates from the reference. The deviation determination part determines whether or not time series of the load value represented by the determination object data deviates from the reference represented by the reference data depending on whether or not it exceeds an allowable range represented by the allowable data.SELECTED DRAWING: Figure 1

Description

  Embodiments described herein relate generally to a deviation determination device, a day type determination device, a prediction device, a program, and a deviation determination method.

  It is desirable to appropriately use resources such as energy in buildings and various facilities or equipment. In order to operate these resources optimally, it is necessary to improve the accuracy of a prediction device that predicts the load. Prediction in the prediction device is performed based on actual load data to be managed. This performance data is data which has a load numerical value for every day and every time slot | zone. The pattern of changes in load performance during the day varies from day to day. However, by appropriately setting the day type, a change pattern according to the day type can be predicted. In order to accurately predict fluctuations in the load record data, it is necessary to reflect the latest load record data in the prediction model by using it as learning data or statistical processing target data.

  The pattern of load change in a building or the like varies depending on the operation status of the equipment that becomes the load, the activity status of the person who uses the device, the operation status of the air conditioning equipment according to the weather condition, and the like. Among these, the factor that greatly affects the operation of equipment and the activities of personnel is the day type. The day type is a type such as weekday or holiday (Saturday, Sunday, holiday, etc.). Depending on the case and purpose, it may be divided into more detailed day types (weekdays after holidays, premium Fridays, Saturdays, etc.). When performing load prediction, a different prediction model is often used for each day type.

  In the prior art, a basic day type is set manually by a human. In addition, in the related art, if there is an error in the setting of the day type, the prediction result may not match the actual load pattern. Further, in the prior art, attempts have been made to automatically correct or exclude data that causes disturbance from the actual load data.

  Whether or not a specific day falls into a specific day type in terms of fluctuations in load results, the degree of agreement between the load result data for that specific day and the standard fluctuation pattern data for that day type Automatically determined based on However, if this degree of matching is judged too strictly, there is a problem that the possibility of being classified as “others” or being classified as a peculiar day increases due to being out of the preset day type. Similarly, when the pattern of load change during the day shifts in the time direction, there is a problem that the number of classifications increases. When the number of classifications increases, there may arise a problem that the management load on the administrator increases. Moreover, if the number of classifications increases too much, there may be a problem that it takes too many days to accumulate load actual data for each classification (for each day type). Moreover, if the amount of load record data for each classification (for each day type) is not sufficiently accumulated, there is a possibility that the accuracy of the load prediction is lowered.

  In other words, in determining whether the time-series load data deviates from the standard change pattern, there is a possibility that a problem may occur due to the strict application of the pattern.

JP 2017-021497 A

  The problem to be solved by the present invention is a deviation determination device, a day type determination device, a prediction device, a program, which can allow a small amount of deviation from a standard reference pattern with respect to time-series load data, And providing a method for judging deviation.

  The deviation determination apparatus according to the embodiment includes a determination target data storage unit, a reference data storage unit, an allowable data storage unit, and a deviation determination unit. The determination target data storage unit stores determination target data that is time-series data of load values to be determined. The reference data storage unit stores reference data representing a reference related to a load variation pattern. The permissible data storage unit stores permissible data representing a degree that is permitted when a load variation pattern deviates from the reference. The deviation determination unit determines whether or not a time series of load values represented by the determination target data deviates from a reference represented by the reference data based on whether or not the allowable range represented by the allowable data is exceeded.

The block diagram which shows schematic function structure of the deviation judgment apparatus of 1st Embodiment. Schematic which shows an example of the structure of the judgment object data which the judgment object data storage part of 1st Embodiment memorize | stores. Schematic which shows an example (1st aspect) of a structure of the reference | standard data which the reference | standard data storage part of 1st Embodiment memorize | stores. Schematic which shows an example (2nd aspect) of a structure of the reference | standard data which the reference | standard data storage part of 1st Embodiment memorize | stores. In 1st Embodiment, the graph which shows the example of the range used as the reference | standard of load actual value, and the example of the time transition of load actual value. The block diagram which shows schematic function structure of the deviation judgment apparatus of 2nd Embodiment. Schematic which shows an example of a structure of the weight data which the weight data storage part of 2nd Embodiment memorize | stores. The block diagram which shows schematic function structure of the deviation judgment apparatus of 3rd Embodiment. Schematic which shows an example of a structure of the time shift data which the time shift data memory | storage part of 3rd Embodiment memorize | stores. The block diagram which shows schematic function structure of the deviation judgment apparatus of 4th Embodiment. The block diagram which shows schematic function structure of the day classification judgment apparatus of 5th Embodiment. The flowchart which shows the process sequence for judging a day classification in the day classification judgment apparatus of 5th Embodiment. The block diagram which shows the schematic function structure of the prediction apparatus of 6th Embodiment, and the energy crisis optimal operation system comprised using the prediction apparatus. The graph which shows the example of the load performance data regarding a day type determination in case the tolerance by weight data is not applied. The graph which shows the example of the load performance data regarding a day type determination in the case of applying tolerance by weight data in 5th Embodiment, 6th Embodiment, etc. FIG. The graph which shows the example of the load performance data in the case of performing the judgment by day type by performing the allowance by the time shift in the fifth embodiment and the sixth embodiment. The graph which shows the example of the load performance data in the case of performing the judgment by day type by performing the allowance by the time shift in the fifth embodiment and the sixth embodiment. The table | surface which shows the example of load performance data in the 5th Embodiment, 6th Embodiment, etc. in the case of performing the time type shift and determining the day type.

  Hereinafter, a divergence determination device, a day type determination device, a prediction device, a program, and a divergence determination method according to embodiments will be described with reference to the drawings.

(First embodiment)
First, the first embodiment will be described.
FIG. 1 is a block diagram illustrating a schematic functional configuration of a deviation determination apparatus according to the first embodiment. As shown in the figure, the divergence determination device 2000 determines whether or not the series data is deviated from a predetermined reference. As shown in the figure, the divergence determination device 2000 includes a divergence determination unit 2001, a determination target data storage unit 2010, a reference data storage unit 2011, an allowable data storage unit 2012, and a determination result data storage unit 2021. Composed.

The divergence determination unit 2001 determines whether the determination target data stored in the determination target data storage unit 2010 is deviated from a reference determined by data stored in the reference data storage unit 2011. Note that the divergence determination unit 2001 determines whether or not the time series of the load values represented by the determination target data deviates from the reference represented by the reference data based on whether or not the allowable range represented by the allowable data is exceeded.
The determination target data storage unit 2010 stores data that is a target of determination processing by the deviation determination unit 2001. The determination target data storage unit 2010 stores determination target data that is time-series data of the load values to be determined.
The reference data storage unit 2011 stores information indicating a reference range as data regarding the determination target data stored in the determination target data storage unit 2010. The reference data storage unit 2011 stores reference data representing a reference related to a load variation pattern.
The allowable data storage unit 2012 stores, as data, information indicating the allowable level in the determination by the divergence determination unit 2001. The permissible data storage unit 2012 stores permissible data indicating the degree of tolerance when the load variation pattern deviates from the standard determined by the standard data.
The determination result data storage unit 2021 stores data of a determination result by the divergence determination unit 2001. The determination result data stored in the determination result data storage unit 2021 is information indicating whether the determination target data is deviated from the reference.

  FIG. 2 is a schematic diagram illustrating an example of a configuration of determination target data stored in the determination target data storage unit 2010. As shown in the figure, the determination target data is data including a set of values associated with the index t. Here, the index t is a value for distinguishing a plurality of time intervals (for example, time intervals of 1 hour) within a predetermined period (for example, 1 day). In the illustrated example, t is an integer and 0 ≦ t ≦ 23. In the illustrated data, t = 0 corresponds to the time interval “from 0:00 to 1:00”. T = 1 corresponds to the time interval “from 1:00 to 2:00”. Thereafter, the same applies, and t = 23 corresponds to the time interval “from 23:00 to 24:00”. In the example shown in the figure, the time period is divided into 24 equal parts, but the time period may be divided into 48 equal parts, or a time period of another length may be used.

  The value W (t) is a load actual value in the time interval indicated by t. As an example, the value W (t) is a load actual value (unit: kilowatt hour) of the electric energy in a predetermined building in the time interval. In addition, although the case where W (t) is the numerical value of electric energy was illustrated here, it may be a load of heat amount, for example, and may be a load of water usage.

FIG. 3 is a schematic diagram illustrating an example (first aspect) of the configuration of the reference data stored in the reference data storage unit 2011. As illustrated, the reference data of this aspect includes an upper limit value U (t) and a lower limit value L (t) in association with the index t. The index t in the reference data in this figure is the same as the index t in the determination target data already described, and they correspond to each other. The reference data of this aspect represents a predetermined range by the upper limit value U (t) and the lower limit value L (t) in the time zone indicated by the index t.
The actual reference data (first aspect) is generated based on, for example, the average value and standard deviation of each index t based on the series data of a plurality of samples. For example, in each index t, a value such as (average value−2.0 × standard deviation) or (average value−3.0 × standard deviation) is set as the lower limit L (t). Further, values such as (average value + 2.0 × standard deviation) and (average value + 3.0 × standard deviation) are set as the upper limit value U (t).

FIG. 4 is a schematic diagram illustrating an example (second mode) of the configuration of the reference data stored in the reference data storage unit 2011. Instead of the first mode, the reference data of the second mode may be used. As illustrated, the reference data of this aspect includes an average value Av (t) and a range width Am (t) in association with the index t. The reference data of this aspect represents a predetermined range by the average value Av (t) and the range width Am (t) in the time zone indicated by the index t. That is, the reference data of this aspect represents a range from the value (Av (t) −Am (t)) to the value (Av (t) + Am (t)).
The actual reference data (second mode) is generated based on, for example, an average value and a standard deviation for each index t based on series data of a plurality of samples. For example, in each index t, a value such as (2.0 × standard deviation) or (3.0 × standard deviation) is set as the range width Am (t).

  It should be noted that the first mode (FIG. 3) and the second mode (FIG. 4) of the reference data differ only in how they are expressed, and the information shown is equivalent.

FIG. 5 is a graph showing an example of a range serving as a reference for the actual load value and an example of a time transition of the actual load value. In the graph of the figure, the horizontal axis is time, and the vertical axis is the actual load value. This graph, as the time indicator _{includes t 1, t 2, t 3} , t 4. The indices t ₁ , t ₂ , t ₃ and t ₄ are not necessarily adjacent indices. In this graph, the upper and lower limits of the reference range are indicated by broken lines. Further, this graph shows curves A, B, and C by solid lines. Curves A, B, and C show the transition of the actual load value, respectively.

In the graph shown in FIG. 5, the load actual value of the curve A is always within the range between the upper limit and the lower limit.
Moreover, the load actual value of the curve B is in the range between the upper limit and the lower limit for most of the time. And the load actual value of the curve B has a part which has exceeded the upper limit line on the way. Specifically, the curve B is greater than the upper limit of the line between the t ₂ and t _3, which exceeds the value of the upper limit of t _3, and again lower than the upper limit of the line between the t ₃ and t _4, It is back within the standard range.
Further, the actual load value of the curve C exceeds the upper limit line in the middle, and thereafter gradually deviates greatly from the reference range while exceeding the upper limit line.

The divergence determining unit 2001 also refers to the allowable data storage unit 2012 to determine whether or not the time series of actual load values is deviated from the reference.
For example, the curve A is located between the lower limit and the upper limit in all the intervals t shown in the figure. Therefore, the deviation determination unit 2001 determines that the data of the series represented by the curve A does not deviate from the reference. Further, the curve B falls between the lower limit and the upper limit in almost all of the section t shown in the figure. However, the value indicated by the curve B exceeds the upper limit in the vicinity of t = t ₃ . In this case, the divergence determination unit 2001 determines whether the degree to which the value of the series represented by the curve B exceeds the upper limit is allowable by referring to the allowable data. In this example, the divergence determining unit 2001 determines that the series represented by the curve B is within an allowable range. In this case, the divergence determining unit 2001 determines that the data of the series represented by the curve B is not deviated from the reference. The curve C exceeds the upper limit line between t = t ₂ and t = t _3, and the value increases as it is. In this case, the divergence determination unit 2001 determines whether or not the degree to which the value of the series represented by the curve C exceeds the upper limit is allowable by referring to the allowable data. In this example, the divergence determining unit 2001 determines that the series represented by the curve C is in an unacceptable range. In this case, the divergence determination unit 2001 determines that the data of the series represented by the curve C is deviated from the reference.

  The permissible data represents, for example, a numerical value indicating a range in which data that exceeds the reference is permissible. The allowable data defines, for example, the range in which the data that exceeds the reference is allowed by information such as the absolute value of the value, the ratio of the value, and the integrated value of the value (by integration in the time direction).

(Second Embodiment)
Next, a second embodiment will be described. However, descriptions of matters common to the already described embodiments may be omitted. Here, the description will focus on matters specific to the present embodiment.

FIG. 6 is a block diagram showing a schematic functional configuration of the deviation determination apparatus according to the second embodiment. As illustrated, the divergence determination device 2100 includes a divergence determination unit 2101, a determination target data storage unit 2010, a reference data storage unit 2011, a weight data storage unit 2112, and a determination result data storage unit 2021. Composed.
The divergence determination unit 2101 refers to the weight data storage unit 2112, and the determination target data stored in the determination target data storage unit 2010 differs from the reference represented by the reference data stored in the reference data storage unit 2011. Judge whether or not.
The weight data storage unit 2112 stores weight data, which is a specific example of “allowable data” described in the first embodiment. The weight data storage unit 2112 stores weight information that is associated with the time index and represents the degree of tolerance. Note that the weight data storage unit 2112 stores weight data representing weight values determined corresponding to the time or the time zone. An allowable range for determining whether the determination target data is deviated from the reference is determined by the load value range represented by the reference data and the weight value for each time or time zone. For example, the weight value is a value indicating how much the upper limit value and the lower limit value determined for each time or time zone are allowed to be widened (or narrowed). Details of the weight data will be described later.

  Note that each of the determination target data storage unit 2010, the reference data storage unit 2011, and the determination result data storage unit 2021 is the same as that of the previous embodiment, and thus description thereof is omitted here.

  FIG. 7 is a schematic diagram illustrating an example of a configuration of weight data stored in the weight data storage unit 2112. As illustrated, the weight data is data of a weight value x (t) associated with the index t. The weight value x (t) is a positive number. As an example, the weight value is defined as in the following expression (1) and stored in the weight data storage unit 2112 in advance.

  The divergence determination unit 2101 according to the present embodiment determines whether or not the determination target data is deviated from the reference as follows. That is, when the determination target data W (t) at the index t satisfies the condition of the following expression (2), the deviation determination unit 2101 determines that the determination target data is deviated from the reference at the index t.

  In Expression (2), Av (t) is an average value in the reference data of the second aspect. Am (t) is a range width in the reference data of the second mode. X (t) is weight data.

  The index t corresponds to the time or the time zone. By giving the weight data as a function of the index t as described above, the allowable level can be changed depending on the time or the like. For example, in a time zone where the actual value is small (a time zone relatively close to 0), the weight data x (t) is increased (that is, the allowable level is reduced), and a time zone where the actual value is large (relatively 0). The weight data x (t) can be reduced (that is, the allowable level is increased) in the time zone far from the time zone. As a result, it is possible to allow variations in the value fluctuation pattern in a time zone where the actual value is large.

(Third embodiment)
Next, a third embodiment will be described. However, descriptions of matters common to the already described embodiments may be omitted. Here, the description will focus on matters specific to the present embodiment.

FIG. 8 is a block diagram showing a schematic functional configuration of the deviation determination apparatus according to the third embodiment. As shown in the figure, the divergence determination device 2200 includes a divergence determination unit 2201, a determination target data storage unit 2010, a reference data storage unit 2011, a time shift data storage unit 2212, and a determination result data storage unit 2021. Consists of.
The divergence determination unit 2201 refers to the time shift data storage unit 2212 and the determination target data stored in the determination target data storage unit 2010 is based on the reference represented by the reference data stored in the reference data storage unit 2011. Determine whether there is a divergence.
The time shift data storage unit 2212 stores time shift data, which is a specific example of “allowable data” described in the first embodiment. The time shift data storage unit 2212 stores information (time shift data) that is associated with an index of time and that represents an allowable amount of time shift. The time shift data represents a shift amount in the time direction that is allowed with respect to time or a time zone. The allowable range for determining whether the determination target data deviates from the reference is determined by shifting the range of the load value represented by the reference data in the time direction according to the shift amount represented by the time shift data. Details of the time shift data will be described later.

  Note that each of the determination target data storage unit 2010, the reference data storage unit 2011, and the determination result data storage unit 2021 is the same as that of the above-described embodiment, and thus description thereof is omitted here.

FIG. 9 is a schematic diagram illustrating an example of a configuration of time shift data stored in the time shift data storage unit 2212. As shown in the figure, the time shift data is data of the value of the time shift allowable amount in both positive and negative directions. In the figure, the negative direction of the time shift data is TS _N, the positive direction of the time shift data is TS _P. Incidentally, both the TS _N and TS _P is non-negative. Also, it may be values and is always equal to the value of TS _N and TS _P, may store only one value either time shift data storage unit 2212 in that case. Further, the time shift data may be stored as a part of a program code for realizing the function of the deviation determination unit 2201, for example. In this case, the area including the time shift data in the memory for storing the program code is the time shift data storage unit 2212.
As an example, TS _N = TS _P = 1 may be set.

The divergence determination unit 2201 performs divergence determination including the time zone shifted forward and backward based on the above time shift data. Thereby, the divergence determination unit 2201 allows a time shift with respect to time-series determination target data. A specific determination method by the divergence determination unit 2201 is as follows. That is, for all integers i satisfying the condition (−TS _N ≦ i ≦ TS _P ), when the condition of the following expression (3) is true, if the determination target data is deviated from the reference at the index t, The divergence determination unit 2201 determines. In other words, in any one or more of the integers i satisfying (−TS _N ≦ i ≦ TS _P ), when the condition of the expression (3) is false, the determination target data is not deviated from the reference in the index t. The divergence determining unit 2201 determines.

  By performing the above determination method, the divergence determination unit 2201 can allow a time-direction shift of time-series determination target data. That is, the deviation judgment by the deviation judgment unit 2201 can be prevented from being too strict.

(Fourth embodiment)
Next, a fourth embodiment will be described. However, descriptions of matters common to the already described embodiments may be omitted. Here, the description will focus on matters specific to the present embodiment.

FIG. 10 is a block diagram illustrating a schematic functional configuration of the deviation determination apparatus according to the fourth embodiment. As shown in the figure, the deviation determination device 2400 includes a deviation determination unit 2401, a determination target data storage unit 2010, a reference data storage unit 2011, a weight data storage unit 2112, a time shift data storage unit 2212, and determination result data. And a storage unit 2021.
The divergence determination unit 2401 stores the determination target data stored in the determination target data storage unit 2010 in the reference data storage unit 2011 while also referring to the weight data storage unit 2112 and the time shift data storage unit 2212. It is determined whether or not a deviation from the standard represented by the standard data. That is, the divergence determining unit 2401 determines whether or not the determination target data is deviated from the reference based on the weight data associated with the index t, similarly to the divergence determining unit 2101 in the second embodiment. At the same time, the divergence determination unit 2401 considers the reference data of the time zones before and after the time zone indicated by the index t, based on the time shift data, similarly to the divergence determination unit 2201 in the third embodiment. Then, it is determined whether the determination target data is deviated from the reference.

  Note that each of the determination target data storage unit 2010, the reference data storage unit 2011, the weight data storage unit 2112, the time shift data storage unit 2212, and the determination result data storage unit 2021 is the same as those in the already described embodiment. Therefore, explanation here is omitted.

Details of the divergence determination process by the divergence determination unit 2401 of the present embodiment are as follows. That is, the divergence determination unit 2401 uses the weight data corresponding to the index t to determine whether or not the determination target data is deviated from the reference while performing allowance based on the weight value corresponding to the time zone. . At that time, the divergence determination unit 2401 performs divergence determination including the time zone shifted forward and backward based on the time shift data. Thereby, the divergence determination unit 2201 allows a time shift while allowing a divergence within a predetermined range of the value itself with respect to time-series determination target data. A specific determination method by the divergence determination unit 2401 is as follows. That is, for all integers i satisfying the condition (−TS _N ≦ i ≦ TS _P ), when the condition of the following expression (4) is true, if the determination target data is deviated from the reference at the index t, The divergence determination unit 2401 determines. In other words, when any one of the integers i satisfying (−TS _N ≦ i ≦ TS _P ) and the condition of the expression (4) is false, the determination target data is not deviated from the reference at the index t. The divergence determining unit 2401 determines.

  By performing the above determination method, the divergence determination unit 2401 can determine whether or not the determination target data is divergence by shifting the time back and forth. For example, when the actual value rises or falls earlier (or later) than the standard value fluctuation pattern, it is possible to make a deviation judgment by allowing such a time shift. It becomes possible.

(Fifth embodiment)
Next, a fifth embodiment will be described. However, descriptions of matters common to the already described embodiments may be omitted. Here, the description will focus on matters specific to the present embodiment.

  FIG. 11 is a block diagram illustrating a schematic functional configuration of the day type determination device according to the fifth embodiment. As shown in the figure, the day type discriminating apparatus 3000 includes a day type discriminating unit 1101, a load result data storage unit 1106, a weight data storage unit 1018, a time shift data storage unit 2212, a day type data storage unit 1107, It is comprised including. The load record data storage unit 1106 includes a determination target data storage unit 2010 and a reference data storage unit 2011.

  The day type determination unit 1101 determines the day type of the determination target day based on the load result data stored in the load result data storage unit 1106. Note that the load result data storage unit 1106 includes a determination target data storage unit 2010 that holds the load result data of the determination target day, and a reference data storage unit 2011 that holds the load result data of the day whose day type is known. Have. That is, the reference data storage unit 2011 stores reference data representing a reference for each day type related to the load fluctuation pattern. Note that the judgment target data storage unit 2010 can only store the minutes up to the time zone in the middle of the judgment target date (for example, the time zone from 0:00 am to 1:00 am to the time zone from 11:00 am to noon). Data may not be retained. Then, the day type discriminating unit 1101 determines the similarity between the date of the determination target and the day of the known day type based on the load actual data of the determination target day and the day actual load data of the known day type. Calculate The day type discriminating unit 1101 estimates the day type of the day to be determined based on the calculated similarity.

  That is, the day type discriminating unit 1101 determines whether or not the time series of the load value represented by the determination target data deviates from the reference represented by the reference data based on whether or not the allowable range represented by the allowable data is exceeded. Then, the day type determination unit 1101 determines the day type of the determination target day.

  At this time, the day type discriminating unit 1101 refers to the weight data storage unit 1018 and allows variation in load results by assigning weights for each time zone. The day type discriminating unit 1101 refers to the time shift data storage unit 2212 and allows a time zone shift within a predetermined range. That is, the day type determination unit 1101 has the function of the deviation determination unit described in any of the first to fourth embodiments as part of the process for determining the day type. In other words, the day type discriminating unit 1101 determines to which date type the time-series load record data stored in the judgment target data storage unit 2010 fits while performing the allowance described here. To do.

As described above, the load record data storage unit 1106 stores load record value data for each day and for each time period. Of the load record data storage unit 1106, the determination target data storage unit 2010 stores data of load record values of dates for which the day type determination unit 1101 is a target for determining the day type.
The day type data storage unit 1107 stores the day type for each day. The day type data storage unit 1107 can store a day type set in advance by a person. Further, the day type data storage unit 1107 may be updated by the day type determination unit 1101. That is, for a specific day, when the day type determined by the day type determination unit 1101 does not match the preset day type, the day type determination unit 1101 stores the day stored in the day type data storage unit 1107. Update the day type data.

  The weight data storage unit 1018 is the same as the weight data storage unit 2112 described with reference to FIG. The time shift data storage unit 2212 is the same as the time shift data storage unit 2212 described in FIG. The weight data storage unit 1018 and the time shift data storage unit 2212 are both allowable data storage units that store allowable data indicating the degree to which the load variation pattern is allowed when it deviates from the standard.

Specific processing contents by the day type discrimination unit 1101 are as follows.
The load performance in the time index (time zone) t of the determination target day is defined as W (t). The data of W (t) can be read from the determination target data storage unit 2010. The day type is I. I is one of the elements of the set such as {Saturday (but non-holiday), Sunday or holiday, weekday after holiday, premium Friday, other weekdays}, and the like. Let W _I (t) 実 績 be the average value of the day load performance index t belonging to day type I (the actual ￣ is located above “W _I (t)” as shown in equation (5)) .) In addition, the standard deviation in the index t of the load performance of the day belonging to the day type I is defined as SD _I (t).

The day type discriminating unit 1101 divides the judgment target data from the reference at the index t when the condition of the following expression (5) is true for all integers i satisfying the condition (−TS _N ≦ i ≦ TS _P ). Judge that you are doing. In other words, in any one or more of the integers i satisfying (−TS _N ≦ i ≦ TS _P ), when the condition of the expression (5) is false, the determination target data is not deviated from the reference in the index t. Then, the day type determination unit 1101 determines. Here, the reference is determined by statistics represented by the actual load data of the day type I stored in the reference data storage unit 2011.

In Expression (5), x (t + i) is weight data for the index (t + i).
As a modification of the present embodiment, the weight value stored in the weight data storage unit 1018 may be different for each day type. In this case, the day type discriminating unit 1101 uses the formula (6) instead of the formula (5) to determine whether the determination target data is deviated from the standard of the day type I.

In Expression (6), x _I (t + i) is a weight value in the index (t + i) for the day type I.
When judging the presence or absence of deviation using Formula (6), a different weight value can be used for each day type I.

FIG. 12 is a flowchart showing a processing procedure for determining the day type. Hereinafter, it demonstrates along this flowchart.
The day type information and the weight information are input and set in advance by the human system. This day type information is day type information set in correspondence with the calendar. Further, the weight information is information on the value of the weight for each time period stored in the weight data storage unit 1018 as described above.

  In step S <b> 1101, the day type determination unit 1101 refers to the determination target data storage unit 2010 and determines whether or not the determination target date matches the set day type. If the day to be judged does not match the set day type, the same judgment is made for other day types. Thereby, the day type discriminating unit 1101 determines the day type of the day to be determined.

  If the inside of the process of step S1101 is looked at in detail, the loop of judgment of a day type will be started in step S1120. This loop is repeated for each day type. In step S1121, the day type discriminating unit 1101 determines whether or not the load record value deviates from the reference for each index t of the load record data of the target day. In step S1122, the day type discriminating unit 1101 regards each index t of the load performance data of the target day, the time zone represented by the preceding and following time zones (that is, the index (t + i), where i ≠ 0 .) To be compared, it is determined whether or not the actual load value deviates from the reference. The day type discriminating unit 1101 performs the determinations in steps S1121 and S1122 according to the formula (5) or the formula (6) already described.

If there is no corresponding day type even after performing the processing in step S1101, the day type determination unit 1101 determines that the target day is a peculiar day.
After the process of step S1101, the process proceeds to step S1102, and the day type discriminating unit 1101 performs a process of detecting an abnormal value of the load pattern.

  In the present embodiment, the day type determination unit 1101 determines the day type using the deviation determination described in the fourth embodiment. That is, the day type discriminating unit 1101 determines the day type of the actual load data while performing both allowance by weight for each time zone and allowance by time shift. However, as a modification, only one of tolerance by weight for each time zone (data stored in the weight data storage unit 1018) and tolerance by time shift (data stored in the time shift data storage unit 2212) is used. The determination may be made. In other words, the day type determination unit 1101 may determine the day type using the deviation determination described in either the second embodiment or the third embodiment.

  According to the present embodiment, when comparing the load result of the determination target day with the statistical standard for a specific day type, allowance by weight data and allowance by time shift are taken into consideration. By such a deviation determination, it is possible to avoid an increase in the number of day-type classifications more than necessary.

(Sixth embodiment)
Next, a sixth embodiment will be described. However, descriptions of matters common to the already described embodiments may be omitted. Here, the description will focus on matters specific to the present embodiment.

FIG. 13 is a block diagram illustrating a schematic functional configuration of a prediction device according to the present embodiment and an energy crisis optimum operation system configured using the prediction device.
As shown in the figure, the prediction device 1100 includes a day type determination unit 1101, a load pattern abnormal value detection unit 1102, a prediction load record pattern extraction unit 1103, a day type change unit 1104, a load pattern prediction unit 1105, Load result data storage unit 1106, day type data storage unit 1107, pattern determination date type storage unit 1108, prediction load result pattern storage unit 1109, weather information (forecast) storage unit 1110, and load prediction pattern storage unit 1111, a weight data storage unit 1018, and a time shift data storage unit 2212.

  The energy device optimum operation system 1001 includes a prediction device 1100, an input / output unit 1012, a load record collection unit 1013, a weather information collection unit 1014, an energy device planning unit 1015, an energy device control unit 1016, and a calendar. A set date type storage unit 1017 and an energy device operation pattern storage unit 1019 are included.

  Further, the energy equipment optimum operation system 1001 acquires load record data from the external load facility 1002 and stores the load record data in the load record data storage unit 1106. Further, the energy equipment optimal operation system 1001 acquires weather information (forecast) from the external weather information providing facility 1003 and stores it in the weather information (forecast) storage unit 1110 at least temporarily. In addition, the energy device control unit 1016 in the energy device optimum operation system 1001 controls the external energy facility 1004.

  The load record data storage unit 1106 receives and stores load record data for each time period from the load record collection unit 1013.

  The day type discriminating unit 1101 refers to the load result data of the target day stored in the load result data storage unit 1106, and the day type of the load pattern of the target day (for example, {holiday, weekday, unique day}, etc. Type).

  The method of discrimination by the day type discriminating unit 1101 is as already described in the fifth embodiment. In other words, the day type determination unit 1101 refers to the load result data of the target day stored in the load result data storage unit 1106 and the load result data of the day type other than the target day. Further, the day type determination unit 1101 refers to the weight data stored in the weight data storage unit 1018 and the time shift data stored in the time shift data storage unit 2212 as necessary. Each of the weight data storage unit 1018 and the time shift data storage unit 2212 is as described above. Further, the day type discriminating unit 1101 performs the functions from the first embodiment to the fourth embodiment as a function for determining whether or not the load result data of the target day is deviated from the standard of each day type. It has the function of the described divergence determination unit.

  The day type discriminating unit 1101 refers to the day type of the target date stored in the calendar setting day type storage unit 1017, and if it is the same as the day type of its own discrimination result, the day type data storage unit It is stored in 1107. Also, the day type discriminating unit 1101 refers to the day type of the target date stored in the calendar setting day type storage unit 1017. If the day type is different from the day type of its own discrimination result, the day type discriminating result is discriminated. The day type is corrected and stored in the day type data storage unit 1107. The day type data storage unit 1107 records the day type of the load pattern of the target day output by the day type discrimination unit 1101 in association with the date of the target day.

  The pattern discrimination date type storage unit 1108 stores a load pattern statistical value for each day type. The statistical value of the load pattern for each day type includes, for example, an average load pattern that is an average value of the load pattern for each day type, the maximum value and the minimum value of the load pattern for each day type, and the like. Note that the pattern discrimination date type storage unit 1108 is updated as appropriate based on the actual load data of the day for which the day type is known.

  The prediction load record pattern extraction unit 1103 refers to the day type information stored in the day type data storage unit 1107, refers to the load record data stored in the load record data storage unit 1106, and performs the target period. The load pattern from which the day type is extracted is extracted as a prediction load pattern. The prediction load record pattern extraction unit 1103 stores the extracted prediction load pattern data in the prediction load record pattern storage unit 1109.

  The load pattern abnormal value detection unit 1102 refers to the day type statistical value stored in the pattern determination day type storage unit 1108 and refers to the day type information stored in the day type data storage unit 1107. Further, referring to the load actual data of the target day stored in the load actual data storage unit 1106, based on the statistical value corresponding to the day type of the target day, for each time zone in the load pattern of the target day Detects the presence or absence of abnormal values. Further, when detecting the abnormal value, the load pattern abnormal value detection unit 1102 corrects the abnormal value of the load pattern and causes the predicted load result pattern storage unit 1109 to store the corrected load pattern.

  The day type change unit 1104 compares the day type of the target date recorded in the day type data storage unit 1107 with the day type of the target day set in the calendar setting day type storage unit 1017, and determines both day types. Are different, the input / output unit 1012 is notified that the day type has been changed. The input / output unit 1012 outputs information indicating that the day type has been changed in a form presented to the user. Thereby, the user of the energy equipment optimal operation system 1001 can know that there is an error in the setting of the day type of the calendar information.

  The prediction load record pattern storage unit 1109 stores the prediction load pattern extracted by the prediction load record pattern extraction unit 1103. Further, when receiving the corrected load pattern from the load pattern abnormal value detection unit 1102, the prediction load result pattern storage unit 1109 updates and stores the load pattern at that date and time as the corrected load pattern.

  The weather information (forecast) storage unit 1110 records the weather information collected by the weather information collection unit 1014 in association with the date. The load pattern prediction unit 1105 refers to prediction load pattern data stored in the prediction load record pattern storage unit 1109, and refers to weather information stored in the weather information (forecast) storage unit 1110. The load pattern is predicted and the load prediction pattern is output. Note that the load pattern prediction unit 1105 predicts the load pattern of the target day based on the day type and weather information (forecast) using, for example, a neural network method or a regression model. The load prediction pattern storage unit 1111 stores the load prediction pattern output from the load pattern prediction unit 1105.

  By the operation described above, the prediction device 1100 determines the day type of the target day. As a result, even if the date type stored in the calendar setting date type storage unit 1017 is incorrect, the corrected correct day type can be recorded in the day type data storage unit 1107.

  Moreover, using the load pattern predicted by the prediction device 1100 by the operation described above, the energy equipment optimal operation system 1001 controls the energy equipment as follows. That is, the energy equipment planning unit 1015 refers to the load prediction pattern storage unit 1111 and acquires the predicted load pattern. Then, the energy device planning unit 1015 plans an operation pattern of the energy device according to the acquired load pattern, and writes the operation pattern in the energy device operation pattern storage unit 1019. Then, the energy device control unit 1016 refers to the energy device operation pattern storage unit 1019 and controls the energy equipment 1004 according to the read operation pattern.

  In the present embodiment, the number of classifications by day type can be reduced by allowing variation in load results in a time zone with a heavy load (for example, from 8:00 to 17:00) and time zone shift. If the number of classifications by day type can be reduced, the data for each day type will be accumulated sooner, and this function can be used sooner after the system is installed. In addition, it is possible to use the actual load data closer to the prediction target date for the prediction, and the prediction accuracy can be improved. Since it is possible to reduce the classification of the day type as “others”, it is possible to use the load actual data on the day closer to the prediction target date for the prediction, and the prediction accuracy can be improved.

  Next, based on actual data examples, the effect of “allowance in judgment of deviation” described in the above embodiment will be described.

First, an example of effects related to application of weight data will be described.
FIG. 14 is a graph illustrating an example of load performance data related to day type determination in a case where tolerance by weight data is not applied. In the figure, the horizontal axis is time, and the vertical axis is the numerical value of the load performance data. The time includes a range from 0:00 to around 23:00. A slightly thick solid line in the graph shown in FIG. In the figure, six other lines are drawn, which correspond to the reference data of the day type “weekday” and the day type “holiday”. Specifically, these six lines are an average value, a lower limit value, and an upper limit value for the day type “weekday”, and an average value, a lower limit value, and an upper limit value for the day type “holiday”. In this example, the time transition (solid line) of the load performance of the target day is approximately between the lower limit value and the upper limit value of the day type “weekday”. However, for example, in the area indicated by the arrow near 14:00, the load performance value of the target day is below the lower limit value of the day type “weekday”. That is, it is determined that a deviation has occurred with respect to the standard of the day type “weekday”. That is, when the day type of the target day is determined according to the criteria shown in this graph, the day type of the target day is not classified as “weekday” or “holiday”. In other words, the day type of the target date is treated as a classification such as “others” or a special day.
On the other hand, in the case of the following diagram, it is possible to solve the inconvenience that the target day is not classified into the day type “weekday” or the day type “holiday”.

FIG. 15 shows an example of load performance data related to day type determination when applying allowance by weight data (described in the second embodiment, the fourth embodiment, the fifth embodiment, and the sixth embodiment). It is a graph which shows. Also in the figure, the horizontal axis is time, and the vertical axis is the numerical value of the actual load data. And the transition of the load actual value shown with the graph of a continuous line is the same as that of the graph in FIG. However, in the example shown in FIG. 15, the actual value of the target day is the day type “Weekday” even in the time zone (the area indicated by the arrow) where the actual value of the target day was below the lower limit of the day type “Weekday” in FIG. Is not below the lower limit. This is because the weight value is set and used.
That is, in the example of FIG. 15, the load result data of the target day is classified into the day type “weekday”. That is, as compared with the case of FIG. 14, in the case of FIG. 15, it is possible to reduce the degree that the number of classifications increases unnecessarily or a certain day is classified as a peculiar day.

Next, an example of an effect related to application of time shift will be described.
FIG. 16 is a graph illustrating an example of load performance data related to day type determination. In the figure, the horizontal axis is time, and the vertical axis is the numerical value of the load performance data. The time includes a range from 0:00 to around 23:00. A slightly thick solid line in the graph shown in FIG. Here, the actual values from 0 o'clock to 7 o'clock are plotted as the actual load on the target day. In the figure, three other lines are drawn, which correspond to the reference data of the day type “weekday”. Specifically, these three lines indicate an average value, a lower limit value, and an upper limit value for the day type “weekday”, respectively. In addition, in the same figure, the reference ranges at 6 o'clock, 7 o'clock and 8 o'clock are indicated by the vertical line of the month. The reference range at each time is a range from the value of “weekday lower limit” to the value of “weekday upper limit”.

The load actual value at each time point from 0 o'clock to 6 o'clock is within the range from the lower limit to the upper limit at each time point. Therefore, during the period from 0 o'clock to 6 o'clock, the load actual value of the target day does not deviate from the reference.
The load actual value at the time of 7 o'clock shows a value slightly exceeding 4000 (indicated by a white diamond in the graph). This value is a value exceeding the reference range at 7 o'clock. Therefore, if the allowance by the time shift is not performed, the load result value of the target day is deviated from the reference at 7 o'clock.
However, the actual load value at the time of 7 o'clock is within the standard range at 8 o'clock (the range from the weekday lower limit to the weekday upper limit). In other words, when a shift of one hour or more in the positive direction is allowed, it can be said that the load result data on the target day shown in the figure has not deviated from the standard even at the time of 7 o'clock. .
As described above, by allowing the time shift, the load record data may not be regarded as deviating from the standard pattern reference. In other words, it is possible to prevent the number of day types from increasing darkly.

  FIG. 17 is a graph including a portion after 8 o'clock of the load record data on the target day shown in FIG. As shown in the figure, the load result data of the target day exceeds the upper limit at each of the time points of 8:00, 14:00, and 23:00. Further, the actual load data on the target day is below the lower limit at 16:00 and 17:00. However, the actual load values at 8 o'clock, 14 o'clock, 16 o'clock, 17 o'clock, and 23 o'clock listed here do not deviate from the reference when the allowance due to time shift is considered. In the present example, the time shift allowable range is 1 hour in both positive and negative directions.

  FIG. 18 is a schematic diagram showing numerical values at each time point (time zone) corresponding to the graph of FIG. As shown in the figure, the measured values (load actual values) at each time point are recorded in increments of one hour from 0:00 to 23:00. Of these time points, 0 o'clock, 1 o'clock, 2 o'clock, 3 o'clock, 4 o'clock, 5 o'clock, 6 o'clock, 9 o'clock, 10 o'clock, 11 o'clock, 12 o'clock, 13 o'clock, 15 o'clock, 18 o'clock, 19 o'clock , 20:00, 21:00, and 22:00 actually measured values are within the range from “weekday lower limit” to “weekday upper limit” at the time. That is, at each of these times, it is determined that the actual measurement value does not deviate from the reference. For each of these times, a “◯” mark is written in the “comparison in the same time zone” field. On the other hand, the measured values at other times are not within the range from “weekday lower limit” to “weekday upper limit” at the time. For each of these times, an “x” mark is written in the “comparison in the same time zone” field.

Each time at which the “x” mark is written in the “comparison in the same time zone” field will be further described. In the present example, the time shift allowable range is 1 hour in both positive and negative directions. That is, TS _N = TS _P = 1. The actually measured value at 7 o'clock is within the range from “weekday lower limit” to “weekday upper limit” at 8 o'clock. The actually measured value at 8 o'clock falls within the range from “weekday lower limit” to “weekday upper limit” at 9 o'clock. The actually measured value at 14:00 falls within the range from “weekday lower limit” to “weekday upper limit” at 13:00 and 15:00. The actually measured value at 16:00 falls within the range from “weekday lower limit” to “weekday upper limit” at 17:00. The actually measured value at 17:00 falls within the range from “weekday lower limit” to “weekday upper limit” at 18:00. The actual measurement value at 23:00 falls within the range from “weekday lower limit” to “weekday upper limit” at 22:00.

  By allowing the time shift in this manner, the load actual data may not be regarded as deviating from the standard pattern standard. In other words, it is possible to prevent the number of day types from increasing darkly.

  According to at least one embodiment described above, the allowable data storage unit that stores the allowable data (for example, weight data, time shift data, etc.) indicating the allowable level when the load fluctuation pattern deviates from the standard. Have. Thus, the deviation determination unit determines whether or not the time series of the load values represented by the determination target data deviates from the reference represented by the reference data, based on whether or not the allowable range represented by the allowable data is exceeded. As a result, the deviation can be determined while allowing a predetermined allowable range.

  In addition, you may make it implement | achieve at least one part of functions of apparatuses or systems, such as a deviation determination apparatus in the embodiment mentioned above, a day type determination apparatus, a prediction apparatus, and an energy equipment optimal operation system, with a computer. In that case, a program for realizing this function may be recorded on a computer-readable recording medium, and the program recorded on this recording medium may be read into a computer system and executed. Here, the “computer system” includes an OS and hardware such as peripheral devices. The “computer-readable recording medium” is a portable medium such as a flexible disk, a magneto-optical disk, a ROM, a CD-ROM, a DVD-ROM, a USB memory, or a storage device such as a hard disk built in a computer system. That means. Furthermore, a “computer-readable recording medium” dynamically holds a program for a short time, like a communication line when transmitting a program via a network such as the Internet or a communication line such as a telephone line. In this case, a volatile memory inside a computer system serving as a server or a client in that case may be included and a program held for a certain period of time. The program may be a program for realizing a part of the above-described functions, or may be a program that can realize the above-described functions in combination with a program already recorded in a computer system.

  Although several embodiments of the present invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the spirit of the invention. These embodiments and their modifications are included in the scope and gist of the invention, and are also included in the invention described in the claims and the equivalents thereof.

DESCRIPTION OF SYMBOLS 1001 ... Energy equipment optimal operation system, 1002 ... Load equipment, 1003 ... Meteorological information provision equipment, 1004 ... Energy equipment, 1012 ... Input / output part, 1013 ... Load results collection part, 1014 ... Weather information collection part, 1015 ... Energy equipment plan , 1016 ... energy equipment control part, 1017 ... calendar setting date type storage part, 1018 ... weight data storage part, 1019 ... energy equipment operation pattern storage part, 1100 ... prediction device, 1101 ... day type discrimination part, 1102 ... load pattern Abnormal value detection unit, 1103 ... Prediction load result pattern extraction unit, 1104 ... Day type change unit, 1105 ... Load pattern prediction unit, 1106 ... Load result data storage unit, 1107 ... Day type data storage unit, 1108 ... Pattern discrimination date Type storage unit, 1109 ... Prediction load record pattern , 1110 ... Weather information (forecast) storage unit, 1111 ... Load prediction pattern storage unit, 2000 ... Deviation determination device, 2001 ... Deviation determination unit, 2010 ... Determination target data storage unit, 2011 ... Reference data storage unit, 2012 ... Permitted Data storage unit 2021... Judgment result data storage unit 2100. 2400 ... Deviation judgment device 2401 ... Deviation judgment unit 3000 ... Day type discrimination device

Claims (7)

A determination target data storage unit that stores determination target data that is time-series data of load values of the determination target;
A reference data storage unit for storing reference data representing a reference for a load variation pattern;
An allowable data storage unit for storing allowable data representing a degree allowed when a load fluctuation pattern is out of the reference;
A divergence determination unit that determines whether or not a time series of load values represented by the determination target data deviates from a reference represented by the reference data, based on whether or not the allowable range represented by the allowable data is exceeded;
A divergence determination device comprising: The allowance data includes weight data representing a weight value determined corresponding to a time or a time zone,
The allowable range is determined for each time or time zone by a range of load values represented by the reference data and the weight value.
The deviation determination apparatus according to claim 1. The tolerance data includes time shift data representing a shift amount allowed for a time or a time zone,
The allowable range is determined by shifting the range of the load value represented by the reference data in the time direction according to the shift amount represented by the time shift data.
The deviation determination apparatus according to claim 1 or 2. A determination target data storage unit that stores determination target data that is time-series data of load values of the determination target day;
A reference data storage unit that stores reference data representing a reference for each day type related to a load fluctuation pattern;
An allowable data storage unit for storing allowable data representing a degree allowed when a load fluctuation pattern is out of the reference;
By determining whether or not the time series of load values represented by the determination target data deviates from the reference represented by the reference data, whether the time series of the determination target data exceeds the allowable range represented by the allowable data. A day type determination unit for determining the day type of the day;
A day type discrimination device comprising: A day type discrimination device according to claim 4,
Predictive load actual pattern extraction for extracting a predictive load pattern for the day type of the determination target day specified by the day type discriminating device from the load actual data stored in association with the day type and the load actual pattern And
A load pattern prediction unit that predicts a load pattern based on the prediction load pattern extracted by the prediction load result pattern extraction unit;
A prediction device comprising: Computer
The deviation determination device according to any one of claims 1 to 3,
Program to function as. Store the determination target data, which is time-series data of the load value of the determination target, in the determination target data storage unit,
Reference data representing a reference related to the load variation pattern is stored in the reference data storage unit,
Storing allowable data representing a degree allowed when a load fluctuation pattern is out of the reference in the allowable data storage unit;
The divergence determination unit determines whether or not the time series of the load value represented by the determination target data deviates from the reference represented by the reference data based on whether or not the allowable range represented by the allowable data is exceeded.
Deviation judgment method.

Images