JP2017208952A - Demand/supply operation support device and demand/supply operation support method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a demand/supply operation support device and demand/supply operation support method that are capable of predicting a power demand using a human flow even when a correlation coefficient between an air temperature and demand is small in a predetermined time zone and region, and are capable of using, as an explanatory variable, a human flow variation as a demand variation factor.SOLUTION: A demand/supply operation support device includes: a correlation analysis unit for analyzing correlation between a power demand prediction error, which is obtained by calculating a difference between a power demand prediction value and power demand measurement value, and a human flow amount prediction error, which is obtained by calculating a difference between a human flow amount prediction value and a human flow amount estimation value; a correlation determination unit for determining correlation; and a power demand prediction correction unit for correcting the power demand prediction value using an explanatory factor table based on the power demand error and human flow amount prediction error output as an analysis result of the correlation analysis unit. When it is determined that correlation exists between the power demand prediction error and the human flow amount prediction error, a power demand prediction correction value is calculated by inputting a variation in a human flow amount.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a supply and demand operation support apparatus and a supply and demand operation support method.

  In order to maintain a stable power supply, each transmission and distribution company needs to maintain a balance between power demand and power supply. In this case, since it is difficult to store power, the power demand is predicted, and the operation schedule of each power plant such as thermal power and hydropower is planned based on the prediction result. By operating the generator in accordance with the predicted power demand, it is possible to maintain a supply-demand balance and supply power stably.

  There exists patent document 1 as a related technique of this technique. In Patent Document 1, the temperature is first predicted using data correlated with the temperature, and then the power demand for 24 hours is predicted based on the correlation between the temperature and the demand using the predicted temperature value. Correlation data used here includes air consumption power consumption, soft drink purchase amount, water heater power demand and water supply amount, and other image data such as skin exposure, umbrella amount, It describes that similar cases are searched and predicted using outdoor brightness, sky color, amount of movement of trees and flags, and the like.

JP 2014-192910 A

  However, in Patent Document 1 described above, the correlation between the temperature and the data such as the power consumption of the air conditioning and the purchase amount of the soft drink is the season and time with respect to the predicted value of the temperature that is the basic data of the demand prediction. Since the correlation varies depending on conditions such as the band and region, there is a problem that depending on the conditions, the temperature prediction error is large and the demand prediction error is also large.

  Moreover, even if the temperature prediction error is large, it is difficult to explain the reason when the demand prediction error is small. In addition, for image data, there are many complex error factors in the temperature prediction itself, such as variations in skin exposure due to differences in sensory temperature and variations in the criteria for determining the amount of umbrellas. There is a problem that it is difficult to explain the reason for error generation and prediction results.

  From the above, in the present invention, it is possible to predict power demand by human flow even when the correlation coefficient between temperature and demand is small in a predetermined time zone and region, and explain human flow change as a factor of demand change. An object of the present invention is to provide a supply and demand operation support apparatus and a supply and demand operation support method that can be used as factors.

  In order to solve the above problems, for example, the configuration described in the claims is adopted. As an example, the disclosed supply-demand operation support device includes a power demand prediction error that calculates a difference between a power demand prediction value and a power demand measurement value, and a human flow rate that calculates a difference between a human flow prediction value and a human flow estimation value. Power using the correlation analysis unit for analyzing the correlation of the prediction error, the correlation determination unit for determining the correlation, the power demand error output as the analysis result of the correlation analysis unit, and the explanatory factor table by the human flow rate prediction error It has a power demand prediction correction unit that corrects the demand prediction value, and when the correlation determination unit determines that there is a correlation between the power demand prediction error and the human flow prediction error, it inputs the fluctuation amount of the human flow A power demand prediction correction value is calculated.

  In the disclosed supply and demand operation support method, as an example, a power demand prediction error that calculates a difference between a power demand prediction value and a power demand measurement value, and a person who calculates a difference between a human flow prediction value and a human flow estimation value Analyze the correlation of the flow prediction error, determine the correlation, correct the power demand prediction value using the explanatory table with the power demand error and human flow prediction error output as the correlation analysis result, and When it is determined that there is a correlation between the prediction error and the human flow rate prediction error, a fluctuation amount of the human flow rate is input and a power demand prediction correction value is calculated.

  According to the present invention, power demand can be predicted by human flow even when the correlation coefficient between temperature and demand is small in a predetermined time zone and region. In addition, changes in human flow can be used as an explanation factor as a factor of demand change.

The figure which shows an example of the supply-and-demand operation support apparatus which comprises a supply-and-demand operation system, and a peripheral device. The figure which shows an example of the electric power demand curve in each time of the day. The figure which shows an example which showed the electric power demand according to predetermined area in the list form. The figure which shows the model example of the mesh information which divided the map information and the map. The figure which shows the example of the human flow of each time in each mesh with a bar graph. The figure which shows an example of the electric power demand prediction value predicted by the electric power demand prediction part. The figure which shows an example of the correlation of the electric power demand calculated by the correlation analysis part or an electric power demand prediction error, and an explanatory variable (feature amount). The figure which shows the example which writes together and displays a map and calculation result information. The figure which shows the example which displays calculation result information in a table format. The figure which shows an example of the criterion in a correlation determination part. The figure which shows the example of a flowchart of a demand prediction part. The figure which shows the example of a flowchart of a human flow prediction part. The figure which shows the example of a flowchart of a demand prediction error calculating part. The figure which shows the example of a flowchart of a human flow prediction error calculating part. The figure which shows the example of a flowchart of a correlation analysis part. The figure which shows the example of a flowchart of an area setting part.

  Hereinafter, a supply and demand operation support apparatus and a supply and demand operation support method according to embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a diagram illustrating an example of a supply and demand operation support apparatus and peripheral devices that are elements constituting a supply and demand operation system in a power transmission and distribution company.

  The supply and demand operation support apparatus 10 shown in FIG. 1 includes a prediction unit 112 that includes a power demand prediction unit 101 and a human flow rate prediction unit 103, and an error that includes a power demand prediction error calculation unit 102 and a human flow rate prediction error calculation unit 104. A storage unit 114 including a calculation unit 113, a correlation analysis unit 105, a region setting unit 106, a system information storage unit 107, and a map information storage unit 108, a correlation determination unit 109, an explanatory factor storage table 11, and an explanatory factor storage unit 110 And a power demand prediction correction unit 111.

  Further, the supply and demand operation support apparatus 10 shown in FIG. 1 obtains power demand results from the power demand result storage device 20 as information from the outside, obtains human flow results from the human flow result storage device 21, and obtains weather information and calendar information. Yes.

  Of the prediction unit 112, the power demand prediction unit 101 reads weather forecast information, calendar information, and a past power demand value (power demand actual value) stored in the power demand record storage device 20, for example, a past maximum temperature. The demand is predicted from the temperature using a regression curve (demand prediction model) between the actual value of the minimum temperature and the actual power demand value, and the predicted power demand value is output. In addition to the temperature information such as the maximum temperature and the minimum temperature, weather information such as humidity, weather, amount of solar radiation, and sunshine hours and the actual power demand value may be used. The existing technology described in Patent Document 1 or the like can be applied to the processing here.

  Of the prediction unit 112, the human flow prediction unit 103 reads calendar information and past human flow values (human flow actual values) in a predetermined area stored in the human flow performance storage device 21, for example, for a specified predetermined area. Based on the calendar information, the person flow rate is read from the person flow record storage device 21 by reading the person flow rate at the same month / day of the week, day of the week, the same weekday / holiday classification, and the same time as the date / time to be predicted. Output the predicted value.

Of the information captured in the supply and demand operation support device 10, the weather forecast information is a forecast of the weather condition at a certain point in time, and the weather condition is the weather including sunny, cloudy, rain, snow, etc. Temperature (° C) and humidity (%) at each time, solar radiation (W / m ² ) and sunshine hours (seconds / minutes / hours), precipitation (mm), snowfall (cm) and snowfall (cm), It is preferable to include information on at least any three of wind speed (m / s), wind direction, and cloud cover.

  Of the information captured by the supply and demand operation support device 10, the calendar information should include date, day of the week (Sun / Mon / Tue / Wed / Thu / Fri / Sat), weekdays and holidays, and time information. .

  The area setting unit 106 reads the grid information stored in the grid information storage unit 108 and the grid information stored in the grid information storage unit 107 and predicts the power demand in the prediction unit 112 or the human flow rate. A predetermined area that is a target range when estimating the power, the entire power system corresponding to the predetermined area, or a substation of the same system is set. Note that the grid information may be matched with the map information, and the power grid may be matched on the map.

  The predetermined area is an area corresponding to one mesh when the map is divided into arbitrary rectangular meshes, or an area divided by administrative divisions such as prefectures, municipalities, and the like.

  In addition, the power system information stored in the system information storage unit 107 is, for example, a power system in which power transmission lines, distribution lines, substations, switches, control devices, power consumers, distributed power sources, and the like are connected. Configuration, impedance of transmission line and distribution line, reactance, admittance, tap position of transformer, on / off state of switch, control parameter of control device and output value Electricity load for power demand and distributed power generation output Information having a value or the like.

  The map information stored in the map information storage unit 108 includes, for example, mesh information obtained by dividing a map into squares having an arbitrary area, or mesh information divided by administrative divisions such as prefectures, municipalities, and the like. Any mesh can be set according to the user's specification.

  Of the error calculation unit 113, the power demand prediction error calculation unit 102 is a power demand prediction value calculated by the power demand prediction unit 101, and a power demand value measured by a measuring device installed in a power transmission / distribution system or a power consumer facility. Is calculated every hour, and a power demand prediction error that is a difference between the power demand prediction value and the power demand value is calculated and output.

  Here, the measuring device is an ammeter, a voltmeter, a wattmeter, a phaser measurement unit (PMU), a smart meter, or the like. The power demand value (active power) measured by the measuring device can be measured at arbitrary time intervals. For example, the power demand is measured at intervals of 10 minutes, 15 minutes, 30 minutes, and 60 minutes. Can be sent.

  The human flow prediction error calculation unit 104 of the error calculation unit 113 includes a human flow prediction value calculated by the human flow prediction unit 103, GPS information and population statistics of the mobile device, traffic volume / congestion rate / number of passengers at each station, Reads the estimated human flow rate in a predetermined area estimated using information such as images from an imaging device installed on the street every hour, and calculates the human flow rate prediction error, which is the difference between the predicted human flow rate and the estimated human flow rate. Output.

  The correlation analysis unit 105 reads the power demand prediction error calculated by the power demand prediction error calculation unit 102 and the human flow prediction error calculated by the human flow prediction error calculation unit 104, and calculates the power demand prediction error for each predetermined area and specified time. The regression equation or the multiple regression equation between the flow rate and the human flow rate prediction error is calculated, and information indicating the causal relationship between the power demand fluctuation amount and the human flow rate fluctuation amount is output as the explanatory factor table 11. In addition to the regression equation of power demand prediction error and human flow prediction error, for example, the regression formula and cluster are calculated by combining the power demand prediction value and human flow prediction value, or the combination of power demand actual value and human flow prediction value. May be. In addition to the human flow rate, numerical information such as demographic values, traffic volume / congestion rate, number of passengers at each station may be used.

The correlation determination unit 109 uses the power demand prediction error (power demand fluctuation amount) calculated by the correlation analysis unit 105 and the correlation coefficient R and the determination coefficient R ² of the human flow rate prediction error to formula (4) described later. Thus, the correlation between the two can be determined, and a regression equation (correction model) having a correlation can be selected from the explanatory factor table 11.

  In the explanatory factor table 11 created by the correlation analysis unit 105, the power demand prediction correction unit 111 uses, for example, only the regression equation determined to have a strong correlation in the correlation factor determination unit 109, and changes in human flow rate. From the amount (prediction error), a correction amount for the power demand prediction value at each time after 1 to 3 hours is calculated. By adding / subtracting the correction amount to the power demand prediction value for each time calculated by the power demand prediction unit 101, the error in the power demand prediction is reduced and the fluctuation factor is specified.

  The explanatory factor storage unit 110 stores the explanatory factor table 11 created by the correlation analysis unit 105. The explanatory factor table 11 is, for example, a regression equation between a fluctuation amount of power demand (prediction error) and a fluctuation amount of human flow rate (prediction error) at a certain time or in a predetermined area, a coefficient, a correlation coefficient, and a determination coefficient ( Contribution rate).

  Next, an example of the power demand record value stored in the power demand record storage device 20 and input to the power demand prediction unit 101 will be described. FIG. 2A is a diagram illustrating an example of a power demand curve (power load curve) 204 in which the vertical axis 201 is power demand (also referred to as power load) and the horizontal axis 202 is each time of day. The actual power demand value is measured as an hourly demand value 203. That is, the daily power demand curve 204 is composed of 24 demand values 203.

  The actual power demand value is active power measured by a measuring device installed at a predetermined location in the power system managed by the transmission / distribution company. For example, the total demand generated at the generator end of each power station connected to the power grid is the demand at the power generation end, and the total power demand measured at the connection point between the power station and the transmission system is the transmission end. Demand. Moreover, it is good also considering what totaled the power consumption measured with the smart meter installed in the consumer as an in-area demand.

  FIG. 2b is an example showing the power demand for each predetermined area in a list format. The list of actual power demand values shown in FIG. 2B includes grid information 221, map information 222 of a region where the grid is deployed, and information of time 223 when the demand is measured.

  For example, the grid information 221 includes a power system 224, a substation name 225 (A substation, B substation) or a substation management code 225 (310A, 311B) managed by a transmission / distribution company. In addition, information such as a power plant, a generator, a transmission line, and a distribution line may be used.

  The map information 222 includes mesh codes 226 (A0001, B0005) and subdivided mesh codes 227 (A0001-001, A0001-002, B0005-001 to B0005-004) of a predetermined area. About a mesh code, it shall have any one or more among the mesh divided arbitrarily, the mesh code added to the mesh, the administrative division of a prefecture municipality, and a national municipality code.

  For the predetermined area indicated by the mesh codes 226 and 227, the power demand actual value 228 at each time 223 is recorded. In this embodiment, the actual demand value is obtained every hour, but the time resolution is measured every 1 hour, 3 minutes, 5 minutes, 10 minutes, 15 minutes and 30 minutes.・ You may record. Further, the numerical value of the total daily demand may be calculated and recorded.

  Next, an example of a human flow rate (fluid population) actual value stored in the human flow history storage device 21 of FIG. 1 and input to the human flow rate prediction unit 104 will be described. FIG. 3A is a diagram illustrating a schematic example of the map information 301 and the mesh information 302 obtained by dividing the map.

  Of these, the map information 301 is stored in the map information storage unit 108. For example, the map information 301 can be obtained by purchasing in advance from the creator of the map information and updated every year. Further, map information published on the Internet may be downloaded and used.

  The mesh information 302 is information set to divide the map information 301 into predetermined areas. For example, the mesh information 302 may be set based on an administrative division such as a prefecture or a municipality, or may be arbitrarily set by a demand / supply operation system user. In the case of the mesh information 302 based on the administrative division, the size, shape, and resolution of the mesh are set based on the administrative division, and the information is obtained by adding a national local government code as a mesh code.

  On the other hand, when the user arbitrarily divides, the size of the mesh may be defined by meters (m), and the shape may be defined by an arbitrary polygon such as a square, a rectangle, a rhombus, or a parallelogram. The size of the mesh is related to the resolution of the division. For the size of the mesh, the user must set the minimum and maximum values and variable steps (intervals) in addition to entering numerical values uniquely. Thus, arbitrary mesh information 302 can be set.

  FIG. 3 b is an example in which a bar graph 313 indicates a human flow rate (fluid population) (unit: person) at each time of a predetermined day in each mesh information 302. The vertical axis is the human flow rate (person) 311 and the horizontal axis is the time 312. The time resolution is in units of 1 hour as in the demand record, but may be in units of 1 minute, 3 minutes, 5 minutes, 10 minutes, 15 minutes, and 30 minutes according to the demand measurement value.

  As described above, the human flow rate at each time in the predetermined area is stored in the human flow rate record device 21 for each arbitrary mesh information 302 on the map information 301.

  FIG. 4 is a diagram illustrating an example of a daily power demand prediction value predicted by the power demand prediction unit 101 using a demand prediction model generated by the correlation between the actual power demand value and the weather information. It is the graph which plotted the electric power demand predicted value 403 of each time of the day on the graph which made the vertical axis 401 the electric power demand predicted value [W] and the horizontal axis 402 made time.

  The predicted power demand value 403 is predicted based on a power demand prediction model. The predicted power demand value 403 is a demand value predicted every time, that is, every hour, and is a discrete numerical value as shown in FIG. The time resolution of the demand predicted value 403 can be changed according to the time resolution of the power demand record value stored in the power demand record storage device 20. For example, when the actual power demand value is measured at a time interval of 1 minute, 3 minutes, 5 minutes, 10 minutes, 15 minutes, or 30 minutes, the predicted power demand value is also predicted at the above time interval. It is possible.

  FIG. 5 is a diagram illustrating an example of the correlation between power demand or power demand prediction error calculated by the correlation analysis unit 104 and explanatory variables (features).

  In FIG. 5, when the vertical axis 501 is the power demand [MW] or the power demand prediction error [MW] and at least one or more feature values are set on the horizontal axis (502, 503, 504, 505), The graph which plotted quantity and the electric power demand or the electric power demand prediction error is shown.

  As the feature amount on the horizontal axis, for example, a human flow actual value [person] 502 stored in the human flow actual storage device 21, a human flow predicted value [person] 503 predicted by the human flow prediction unit 103, a human flow estimated value [ At least one feature quantity such as “person” is set. In addition to traffic flow, traffic flow information includes the congestion rate [%] and boarding rate [%] at each time, the number of passengers at each station at each time [people], or the total number of passengers at all stations in a given area. [People], vehicle temperature [° C], traffic volume of roads connecting at least two or more predetermined areas [units / hour], social event information such as changes in the number of visitors [People], hourly audience rating [%], Frequent occurrence of event-related keywords written on SNS and Web, and weather information, in addition to the conventional temperature [° C], at least two features such as hourly atmospheric pressure [hPa], cloud cover, etc. An amount may be set.

  In the example of the correlation shown in FIG. 5, the power demand actual value [MW] 501 with respect to the human flow rate [person] in a predetermined area set in advance on the horizontal axis 502 is plotted 506, and a regression curve 507 obtained from a plurality of plots 506 is obtained. A distribution range 508 is schematically shown. These models are similarly obtained for the other horizontal axis items.

Note that the regression curve 507 shown in FIG. 5 is a linear regression curve by the least square method, and can be calculated by the following general formula. Here, the expression (1) is an expression indicating a linear regression curve, the expression (2) is an expression indicating the slope α of the expression (1), and the expression (3) is an expression indicating the intercept β of the expression (1). It is. Here, the power demand on the vertical axis 501 is expressed as P _{L, k} , the human flow rate h _{m, k on} the horizontal axis 502, and the number of data is expressed as N. As shown by the horizontal axes 502, 503, 504, and 505, when a plurality of feature amounts are set, a linear regression curve can be calculated by multiple regression analysis. Also, with respect to the plot range such as the distribution 508, data (eg, 506) plotted using the hierarchical and non-hierarchical clustering methods may be classified into the number of clusters designated by the user.

  Next, a display example when outputting the calculation result of the supply and demand operation support apparatus 10 will be described.

  FIG. 6A is a diagram illustrating an example in which a map and calculation result information are displayed together. A display example when mesh information 302 is displayed on map information 301 and a mouse cursor 601 or the like is placed on a predetermined mesh 602 is shown. For example, when the cursor 601 is placed on a predetermined mesh 602 on the map 301, information on the calculation result associated with the mesh 602 is displayed in the window 603.

In the window 603, the mesh 602 code (A001-001), the current time (AM3: 00), the power demand in the area surrounded by the mesh 602 (665.00 [MW]), the human flow (221463 [person ]) Is displayed as basic information. Further, as a result of the correlation analysis between the power demand and the human flow rate, a correlation coefficient R (0.85) and a determination coefficient (contribution rate: 0.72) R ² indicating the validity of the correlation are displayed.

Note that the correlation determination unit 109 shown in FIG. 1 determines the correlation between the power demand and the human flow rate in the area of the mesh 602 based on the correlation coefficient R and the determination coefficient R ² , and displays the presence or absence of the correlation. Good. In this case, the user can arbitrarily specify a threshold value as a criterion for determining the correlation, but it is preferable to adopt the general threshold value and criterion shown in FIG. 7 as initial values. For example, in the table of FIG. 7, (1) is “no correlation”, (2) is “substantially no correlation”, (3) is “weak correlation”, (4) is “ In the case of “with correlation” and (5), “strong correlation” is displayed. Note that it may be used as a determination criterion for determining whether or not to use a regression curve as a prediction model. For example, it can be used as a demand prediction model only when there is a strong correlation.

  FIG. 6B is a diagram illustrating an example of displaying the calculation result information in a table format. For the mesh information 302 arranged on the map information 301, the correlation between the power demand and the human flow rate in the area designated by each mesh is displayed.

  The list table displays, for example, mesh code 611, time 612, power demand 613, human flow 614, correlation coefficient 615, and determination coefficient 616 as a causal relationship table. In addition to the power demand 613 and the human flow rate 614, the power demand prediction error and the human flow rate prediction error, the power demand and the traffic flow information, social event information, and weather information described above may be used.

  FIG. 8 is an example of a flowchart showing power demand prediction processing by the power demand prediction unit 101.

  In the first processing step S701 of the power demand prediction unit 101, the region information set by the region setting unit 106 is read, and the process proceeds to processing step S702. The area information is, for example, a mesh code added to the mesh information 302 shown in FIGS. 3a and 6a. In addition to the mesh code, when it is classified by administrative division, it may be a national local government code, a prefecture city name, a substation or power plant name in the area, a management code, and the like.

  In processing step S702, the power demand prediction unit 101 reads calendar information from the external device, and proceeds to processing step S703. The calendar information may include, for example, date, day of the week (Sun / Mon / Tue / Wed / Thu / Fri / Sat), weekdays and holidays, and time information.

  In processing step S703, the power demand prediction unit 101 reads the power demand prediction model from the power demand record storage device 20, and proceeds to processing step S704. The power demand prediction model is composed of a regression curve as described with reference to FIG. 5, and a power demand prediction model created in advance with the above-described calendar information and actual power demand values corresponding to the regional information and weather information is constructed. deep.

  In processing step S704, the hourly temperature forecast value provided from the outside in the weather forecast is read, and the process proceeds to processing step S705.

  In step 705, the predicted temperature value read in processing step S704 is substituted into the regression curve of the actual power demand and the actual temperature as expressed by equation (1), and the predicted power demand at each time (the predicted power demand). Value).

  In processing step S706, the above-described power demand prediction value is output to the power demand prediction error calculation unit 102 or the correlation analysis unit 105, and the power demand prediction is terminated.

  In addition, you may perform the process which reads the information regarding the prediction conditions of process step S701 and S702 as a parallel process.

  According to the above power demand prediction process, the power demand prediction unit 101 can predict the power demand for 24 hours a day on the day of the predetermined area.

  FIG. 9 is a flowchart showing the human flow prediction process in the human flow prediction unit 103.

  In the first processing step S801 in the human flow prediction unit 103, the region information set by the region setting unit 106 is read, and the process proceeds to processing step S802. The area information is, for example, a mesh code added to the mesh information 302 shown in FIGS. 3a and 6a. In addition to the mesh code, when it is classified by administrative division, it may be a national local government code, a prefecture city name, a substation or power station name in the area, and a management code.

  In process step S802, the human flow prediction unit 103 reads calendar information from the external device, and proceeds to process step S803. The calendar information includes, for example, date, day of the week (Sun / Mon / Tue / Wed / Thu / Fri / Sat), weekdays and holidays, and time information.

  In processing step S803, the human flow prediction unit 103 reads the human flow prediction model from the human flow performance storage device 21, and proceeds to processing step S804. The human flow rate prediction model is a regression curve type prediction model represented by a regression curve 507 as shown in FIG. 5 or a cluster in which weather information such as calendar information and mesh code or weather is tagged as tag information (for example, FIG. 5). 5 having a cluster type prediction model 508). The human flow rate prediction model is constructed in advance.

  In processing step S804, a cluster having a tag that matches the acquired calendar information and area information is selected, and a past human flow actual value at the same time is calculated as a predicted human flow.

For example, equation (4) is executed as a method for calculating the predicted human flow rate. In equation (4), h _p (T) is the estimated human flow rate at a certain time T, and h _{p is the} actual human flow rate at the same time, the same day, the same area, and the same weather as the target date and time for predicting the human flow. (T, D, A, W), and the elongation as a correction coefficient is ω.

  In equation (4), the growth rate ω is the rate of population growth or the year-on-year increase / decrease ratio, for example, the ratio obtained by dividing the average flow rate of people up to 30 days before the target date by the average flow rate of previous year May be used. In addition to calendar information and regional information, weather information such as weather (sunny, cloudy, rain, snow, etc.) may be read and used for cluster selection.

  In process step S805, the above-described human flow rate prediction value is output to the human flow rate prediction error calculation unit 104 or the correlation analysis unit 105, and the human flow rate prediction is terminated.

  In addition, you may perform the process which reads the information regarding the prediction conditions of process step S801 and S802 as a parallel process.

  According to the power demand prediction procedure described above, the human flow rate predicting unit 103 can predict the human flow rate for 24 hours per day on the day in a predetermined area.

  FIG. 10 is a flowchart showing power demand prediction error calculation processing by the power demand prediction error calculation unit 102.

  In the first processing step S901 of the power demand prediction error calculation unit 102, the power demand prediction error calculation unit 102 reads calendar information from the external device, and proceeds to processing step S902. The calendar information includes, for example, date, day of the week (Sun / Mon / Tue / Wed / Thu / Fri / Sat), weekdays and holidays, and time information.

  In the repetitive process of process step S902, the power demand prediction error calculation unit 102 repeatedly executes the processes of process steps S903 to S909 using the power demand measurement value measured by the external measurement device. Here, the measurement device that provides the power demand measurement value read by the power demand prediction error calculation unit 102 is a telemeter (TeleMeter) installed at the power station and transmission line connection point or at the generator end, or by the power consumer. It is a smart meter installed in equipment.

  In processing step S903, the power demand prediction error calculation unit 102 is time-synchronized, sets time T = 0 at time 0:00 when the prediction target day changes from the previous day, and proceeds to processing step 904.

  In processing step S904, the demand predicted value at time T predicted by the power demand prediction unit 101 is read, and the process proceeds to processing step S905.

  In process step S905, the electric power demand measured value measured with the measuring device mentioned above is read, and the process proceeds to process step S906.

In the processing step S906, the power demand predicted value at the time T read in the processing step S904 and the power demand measurement value at the same time T read in the processing step S905 are used to calculate the power based on the equations (5) and (6). Calculate the demand forecast error. Here, it is assumed that the power demand prediction error is ε _P (T), the power demand prediction value is PL _{, F} (T), and the power demand measurement value is PL _{, M} (T). Also, the increase / decrease direction of the prediction error is ρ (T), 1 for the increasing direction and −1 for the decreasing direction.

In process step S907, the power demand prediction error ε _P (T) at time T calculated in process step S906 is output to the correlation analysis unit 105, and the process proceeds to process step S908.

  In processing step S908, it is determined whether or not the time T has reached 23:00 on the prediction target date for the time T using the determination formula (7).

  In addition to the determination at time T, if the time resolution is hourly, since one day is 24 sections, a counter may be provided to determine whether the counter reaches 24 times.

  If the determination is true (YES) in process step S908, the power demand prediction error calculation unit 102 ends the repetitive process of process step S902. On the other hand, when the determination is not satisfied (NO), the power demand prediction error calculation unit 102 proceeds to processing step S909, advances time T by 1 hour in processing step S909 (T = T + 1), and then proceeds to processing step S904 again. Then, the same process is executed.

  Thus, the prediction error calculation process at each time for the power demand prediction value in the power demand prediction error calculation unit 102 is completed.

  FIG. 11 is a flowchart showing human flow prediction error calculation processing by the human flow prediction error calculation unit 104.

  In the first processing step S1001 of the human flow rate prediction error calculation unit 104, the human flow rate prediction error calculation unit 104 reads calendar information from the external device, and proceeds to processing step S902. The calendar information includes, for example, date, day of the week (Sun / Mon / Tue / Wed / Thu / Fri / Sat), weekdays and holidays, and time information.

  In the iterative process of process step S1002, the human flow rate prediction error calculation unit 104 repeatedly executes the processes of process steps S1003 to S1009 using the estimated human flow rate estimated by the external device.

  In process step S1003, the human flow rate prediction error calculation unit 104 is time-synchronized, sets time T = 0 at time 0:00 when the prediction target day changes from the previous day, and proceeds to process step 1004.

  In processing step S1004, the predicted human flow rate at time T predicted by the human flow prediction unit 103 is read, and the process proceeds to processing step S1005.

  In process step S1005, the estimated human flow rate estimated by the external device is read, and the process proceeds to process step S1006.

In the processing step S1006, the human flow prediction error at the time T read in the processing step S1004 and the human flow estimation value at the same time T read in the processing step S1005 are used to calculate the human flow prediction error based on the formulas 9 and 10. Calculate. Here, it is assumed that the human flow rate prediction error is ε _H (T), the human flow rate prediction value is h _P (T), and the human flow rate estimation value is h _M (T). Also, the increase / decrease direction of the prediction error is ρ (T), 1 for the increasing direction and −1 for the decreasing direction.

In process step S1007, the human flow rate prediction error ε _H (T) at time T calculated in process step S1006 is output to correlation analysis section 105, and the process proceeds to process step S1008.

  In processing step S1008, it is determined whether or not the time T has reached 23:00 on the prediction target date with the determination formula of Formula 8 for the time T.

  In addition to the determination at time T, if the time resolution is hourly, since one day is 24 sections, a counter may be provided to determine whether the counter reaches 24 times.

  If the determination is true (YES) in process step S1008, the human flow rate prediction error calculation unit 104 ends the repeat process of process step S1002. On the other hand, if the determination is not satisfied (NO), the human flow rate prediction error calculation unit 104 proceeds to processing step S1009, advances time T by 1 hour in processing step S1009 (T = T + 1), and then proceeds to processing step S1004 again. Then, the same process is executed.

  Thus, the prediction error calculation process at each time for the power demand prediction value in the human flow prediction error calculation unit 104 is completed.

  FIG. 12 is a flowchart illustrating a correlation analysis process between the power demand and the human flow rate by the correlation analysis unit 105.

  In the first processing step S1101 of the correlation analysis unit 105, the area information set by the area setting unit 106 is read, and the process proceeds to processing step S1102. The area information is, for example, a mesh code added to the mesh information 302 shown in FIGS. 3a and 6a. In addition to the mesh code, when it is classified by administrative division, it may be a national local government code, a prefecture city name, a substation or power station name in the area, and a management code.

  In processing step S1102, the power demand prediction unit 101 reads calendar information from the external device, and proceeds to processing step S1103. The calendar information includes, for example, date, day of the week (Sun / Mon / Tue / Wed / Thu / Fri / Sat), weekdays and holidays, and time information.

  In processing step S1103, the correlation analysis unit 105 reads at least one of the power demand prediction error calculated by the power demand prediction error calculation unit 102 or the actual power demand value, and proceeds to processing step S1104.

  In processing step S1104, the correlation analysis unit 105 reads at least one of the human flow prediction error calculated by the human flow prediction error calculation unit 104 or the human flow estimation value, and proceeds to processing step S1105.

  In processing step S1105, a regression curve expressed in the form of Formula 1 is calculated for any combination of power demand prediction error and human flow rate prediction error, or power demand actual value and human flow rate estimation value, and processing step The process moves to S1106.

In the processing step S1106, with respect to the regression curve calculated in the processing step S1105, a determination coefficient (contribution which means the validity of the correlation between the power demand prediction error and the human flow rate prediction error, or the correlation between the actual power demand value and the human flow rate estimation value) Rate) R ² is calculated.

  Next, in the processing step S1107, with respect to the regression curve calculated in step S1105, the correlation coefficient R means the correlation between the power demand prediction error and the human flow rate prediction error or the power demand actual value and the human flow rate estimation value. Calculate Note that the processing steps S1101 and S1102 may be executed as parallel processing.

  FIG. 13 is a flowchart showing the region setting process by the region setting unit 106.

  In the first processing step S1201 of the region setting unit 106, the region information set by the region setting unit 106 is read, and the process proceeds to processing step S1202. The area information is, for example, a mesh code added to the mesh information 302 shown in FIGS. 3a and 6a. In addition to the mesh code, when it is classified by administrative division, it may be a national local government code, a prefecture city name, a substation or power station name in the area, and a management code.

  In process step S1202, the area setting unit 106 reads the system information stored in the system information storage unit 107, and proceeds to process step S1203. System information refers to the location information of power equipment including transmission lines, distribution lines, generators, substations, transformers, switches, and measuring devices in the transmission and distribution systems, connection information between each equipment, and information on each equipment and measurement point. It is preferable to include at least one of measured values of voltage, current, power, and frequency.

  In processing step S1203, the maximum area (boundary line on the map) to be mesh-divided is set according to the laying range of the transmission / distribution lines and the interconnection information of the power consumers included in the system information.

  In step 1204, a map information dividing method by mesh is set, and the process proceeds to processing step S1205.

  In processing step S1205, it is determined whether or not to divide by an arbitrary mesh with respect to the division method by mesh. If a division method for dividing by an arbitrary mesh is set (YES), the process proceeds to processing step S1206. In step 1206, the minimum distance [km] and the maximum distance [km] in the latitude direction and the minimum distance [km] and the maximum distance [km] in the longitude direction on the map as a mesh unit, and the scale for scaling the mesh ( Change amount, interval) [km] is set. Thereby, the minimum and maximum areas of the mesh and the size of each mesh at the time of enlargement / reduction are set. On the other hand, when it is not divided by an arbitrary mesh in the determination in step 1205 (NO), the map information is divided using the administrative division stored in advance in the map information storage unit 108 as a mesh. The administrative division is a division of a prefecture city, and a predetermined mesh code (national local government code) is added.

  In step 1207, the map information is divided by the set mesh unit, and the processing of the region setting unit 106 is terminated.

10: Supply and demand operation support device 11: explanatory factor table 20: power demand result storage device 21: human flow result storage device 31: causal table storage device 41: power demand prediction correction amount calculation device 101: power demand prediction unit 102: power demand prediction Error calculation unit 103: human flow rate prediction unit 104: human flow rate prediction error calculation unit 105: correlation analysis unit 106: region setting unit 107: system information storage unit 108: map information storage unit 109: correlation determination unit 110: explanatory factor storage Unit 111: Power demand prediction correction unit 611: Correlation table between power demand and human flow rate

Claims (14)

A correlation analysis unit that analyzes a correlation between a power demand prediction error that calculates a difference between a power demand prediction value and a power demand measurement value, and a human flow prediction error that calculates a difference between a human flow prediction value and a human flow estimation value; A power demand prediction correction that corrects a power demand prediction value using an explanatory factor table based on a correlation determination unit that determines the correlation, a power demand error output as an analysis result of the correlation analysis unit, and a human flow rate prediction error Part
In the correlation determination unit, when it is determined that the power demand prediction error and the human flow rate prediction error are correlated, a fluctuation amount of the human flow rate is input and a power demand prediction correction value is calculated. Supply / demand operation support device. The supply and demand operation support apparatus according to claim 1,
The correlation analysis unit includes, as a correlation between the power demand prediction error and the human flow rate prediction error, a regression equation and a coefficient of the power demand prediction error and the human flow rate prediction error with respect to a predetermined time zone in a predetermined region specified in advance. A supply and demand operation support apparatus characterized by calculating at least one of a relation number and a determination coefficient. The supply and demand operation support apparatus according to claim 1 or 2,
Calendar information that has at least one of date, day of the week, and time, and any of weather, temperature, humidity, sunshine duration, solar radiation, wind direction, wind speed, atmospheric pressure, precipitation, snowfall, and snowfall Power demand forecasting unit for calculating a power demand forecast value at each time based on weather information including one or more of the above, weather information and past power demand actual value, and power demand to be measured hourly with the power demand forecast value A power demand prediction error calculation unit that calculates the power demand prediction error as a difference from the measured value,
The power demand prediction error calculation unit obtains a power demand prediction value at each time predicted by the power demand prediction unit and a power demand measurement value that is actually measured at the same time, and then calculates the power demand prediction value. The supply and demand operation support apparatus, wherein the power demand prediction error, which is a difference from the power demand measurement value, is calculated. The supply and demand operation support apparatus according to any one of claims 1 to 3,
Using a calendar information and weather information, a human flow prediction unit that calculates a human flow prediction value that predicts a human flow at each time in a predetermined area, and a human flow prediction error calculation unit,
The human flow rate prediction error calculation unit obtains a human flow rate prediction value at each time predicted by the human flow rate prediction unit for a predetermined area and a human flow rate estimation value estimated and calculated at the same time, and then the human flow rate prediction value A supply and demand operation support apparatus that calculates a human flow rate prediction error that is a difference between a value and the human flow rate estimated value. The supply and demand operation support apparatus according to claim 4,
The human flow prediction unit obtains the calendar information and the weather information for a past human flow estimated value stored in a human flow record storage device, and has the same date, day of the week, day of week, A supply and demand operation support apparatus for predicting a human flow rate by searching a past human flow rate actual value corresponding to the same time or weather information as a prediction target. The supply and demand operation support apparatus according to claim 4 or 5,
The estimated human flow rate is obtained by using at least one information of position information measured by GPS (Global Positioning System), human flow image information captured by an imaging device, and demographic values. A supply and demand operation support apparatus characterized by being estimated. The supply and demand operation support apparatus according to any one of claims 1 to 6,
The predicted power demand value is calculated using a power demand record value stored in a power demand record storage device, and the power demand record value is a numerical value obtained by measuring the amount of power consumed in a predetermined region or power system. A supply / demand operation support apparatus characterized by being stored in a power demand record storage device as a power demand record value at each time in combination with at least one of calendar information and map information / system information. The supply and demand operation support apparatus according to any one of claims 1 to 7,
The power demand prediction value or the human flow rate prediction value is a prediction value within a predetermined range indicated by map information, and the map information is an administrative division of a prefecture, a municipality, or a map of an arbitrary size It is stored as a collection of areas divided into shapes, and is stored in the map information storage unit with at least one of address, postal code, and power plant and substation location information. Supply and demand operation support device. The supply and demand operation support apparatus according to any one of claims 1 to 8,
The predicted power demand value is calculated using system information of the transmission / distribution system, and the system information is the transmission line / distribution line / generator / substation / transformer / switch / measurement device of the transmission / distribution system. It has at least one or more of position information of power equipment including any one or more of them, connection information between each equipment, and measured values of voltage, current, power, and frequency at each equipment and measurement point. Supply and demand operation support device. The supply and demand operation support apparatus according to any one of claims 1 to 9,
The correlation analysis unit includes a regression equation and a coefficient when the correlation between the power demand prediction error and the human flow prediction error is analyzed, a correlation coefficient, and a determination coefficient included in a predetermined area or system information included in map information. A supply and demand operation support apparatus that is classified and held by a combination of one or more information of a power system corresponding to a substation, a date, a day of the week, a time, and a time zone. The supply and demand operation support apparatus according to any one of claims 1 to 10,
In addition to the human flow rate information, the correlation analysis unit includes railway schedules, boarding rates, number of passengers at each station, traffic volume / flow rate, traffic rate / congestion rate, arrival time, travel trajectory / location information by GPS. Using the traffic flow information that has at least one of them, and the event time held along the railway and road, social event information that has the expected customer acquisition and past customer acquisition results, A supply and demand operation support apparatus characterized by estimating a human flow rate and performing a correlation analysis with the power demand prediction error. The supply and demand operation support apparatus according to any one of claims 1 to 11,
The power demand prediction value or the human flow rate prediction value is for a region on a map given by a region setting unit, and in the region setting unit, a mesh division code or an administrative division, an address, Supply and demand operation support characterized by setting a region for predicting the predicted power demand value or the predicted human flow rate by designating at least one of a postal code, a power plant name, and a substation name apparatus. The supply and demand operation support apparatus according to claim 12,
The region setting unit uses the map information and system information to collate power plant and substation name and location information, and to match the area divided by the map information with the power system. Supply / demand operation support device.   Analyzing the correlation between the power demand prediction error that calculates the difference between the power demand prediction value and the power demand measurement value, and the human flow prediction error that calculates the difference between the human flow prediction value and the human flow estimation value, the correlation The power demand prediction value is corrected using an explanatory factor table based on the power demand error and the human flow rate prediction error output as the correlation analysis result, and is correlated with the power demand prediction error and the human flow rate prediction error. A supply and demand operation support method, comprising: calculating a power demand prediction correction value by inputting a fluctuation amount of a human flow rate when it is determined that there is a possibility.

Images