Apparatus operation planning device, apparatus operation planning method and program

Abstract

To provide an apparatus operation planning device, an apparatus operation planning method and a program that appropriately evaluate a possible prediction error and then realize apparatus operation based on rational replanning even at re-prediction parallel usage time referring to a last actual value.SOLUTION: An apparatus operation planning device comprises: a demand prediction unit; a prediction demand scenario setting unit; and a plan calculation unit. The demand prediction unit predicts prediction demand for a future fixed period. The prediction demand scenario setting unit sets a plurality of prediction demand scenarios using a past prediction error in a similar day, in which arbitrary time past result demand is extracted in a similar order, and a latest prediction demand predicted by the demand prediction unit. The plan calculation unit calculates an operation plan of an apparatus, in which a future desired evaluation function value is minimized or maximized, by adding a prediction error and its occurrence probability using the plurality of prediction demand scenarios set by the prediction demand scenario setting unit.SELECTED DRAWING: Figure 4

Description

[Technical field]

[0001] An embodiment of the present invention relates to an equipment operation planning device, an equipment operation planning method, and a program. [Background technology]

[0002] Conventionally, in facilities such as factories and buildings, operation plans for equipment are calculated to optimize desired evaluation index values ​​over a certain future period (e.g., to minimize operating costs) while observing certain operational constraints (e.g., energy supply and demand balance). Equipment for which operation plans are calculated in this way includes equipment that requires time and cost to start and stop, equipment that is required to continue operating or stop for a certain period of time (e.g., absorption chillers, CGS (Co-Generation Systems), etc.), and equipment that stores energy to accommodate energy over time (e.g., heat storage tanks, storage batteries, etc.).

[0003] To operate such equipment ideally, it is necessary to operate the equipment based on an operation plan that takes into account a certain future period (for example, the next day, etc.). To achieve this, it is essential to predict energy demand over a certain future period as a given condition of the plan, and various methods have been developed to improve the accuracy of this prediction. For example, Non-Patent Document 1 discloses an example of air conditioning heat load prediction that employs the ARIMA (Auto-Regressive Integrated Moving Average) model, which is a representative time series model, and discloses a technology that enables the revision of predicted values ​​by referring to the most recent heat load actual value (hereinafter referred to as the immediately preceding actual value) by introducing the time series model.

[0004] On the other hand, it is inevitable that predictions of energy demand over a certain period of time in the future will contain a certain amount of error. For this reason, methods have been developed that aim to realize rational equipment operation while taking into consideration the occurrence of such prediction errors in advance. For example, Patent Document 1 discloses a technology that realizes rational equipment operation even if prediction errors occur by calculating an equipment operation plan using multiple scenarios created from prediction errors of power demand, etc. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Patent No. 6109631 [Non-patent literature]

[0006] [Non-Patent Document 1] Akiomi Kanehara, Shigeru Kurosu, Fusachika Miyasaka, and Kazuyuki Kamimura, "Air-conditioning load prediction using ARIMA model," Transactions of the Society of Instrument and Control Engineers, Vol. 26, No. 6, pp. 721-728 (1990) Summary of the Invention [Problem to be solved by the invention]

[0007] However, when taking into account the review of forecast values ​​by referring to the most recent actual values ​​as shown in Non-Patent Document 1 (hereinafter, this will be referred to as "reforecasting"; furthermore, the review of a plan based on reforecast values ​​will be referred to as "re-planning"), reforecasting acts to eliminate the most recent prediction error, so creating a scenario using only the forecast error as in Patent Document 1 may be inappropriate.

[0008] The present invention has been made in consideration of the above, and provides an equipment operation planning device, an equipment operation planning method, and a program that realize equipment operation through rational re-planning after appropriately evaluating possible prediction errors, even when re-prediction referring to the most recent actual values ​​is also used. [Means for solving the problem]

[0009] An equipment operation planning device according to an embodiment includes a demand forecasting unit, a forecast demand scenario setting unit, and a plan calculation unit. The demand forecasting unit forecasts forecast demand over a certain period of time in the future. The forecast demand scenario setting unit sets multiple forecast demand scenarios using past forecast errors on similar dates extracted in order of similarity of past actual demand at any time and the latest forecast demand predicted by the demand forecasting unit. The plan calculation unit uses the multiple forecast demand scenarios set by the forecast demand scenario setting unit to calculate an equipment operation plan that minimizes or maximizes a desired future evaluation function value while taking into account forecast errors and their occurrence probability. [Brief description of the drawings]

[0010] [Figure 1] FIG. 1 is a diagram showing an overall configuration of a monitoring and control system including an equipment operation planning device according to a first embodiment. [Diagram 2] FIG. 2 is a diagram illustrating an example of a target device. [Diagram 3] FIG. 3 is a block diagram illustrating a hardware configuration of the equipment operation planning device. [Figure 4] FIG. 4 is a functional block diagram showing a functional configuration of the equipment operation planning device. [Diagram 5] FIG. 5 is a flowchart showing the flow of the operation planning process of the equipment operation planning device. [Figure 6] FIG. 6 is a flowchart showing the flow of the forecast demand scenario setting process performed by the forecast demand scenario setting unit. [Figure 7] FIG. 7 is a diagram for explaining why the prediction error is inappropriate. [Figure 8] FIG. 8 is a diagram showing an example of extraction of similar dates. [Figure 9] FIG. 9 is a diagram illustrating an example of sampling times of the past actual demand difference. [Figure 10] FIG. 10 is a diagram showing an example of creating a forecast demand scenario. [Figure 11]FIG. 11 is a flowchart showing the flow of the forecast demand scenario setting process of the forecast demand scenario setting unit 21 according to the second embodiment. [Figure 12] FIG. 12 is a diagram showing an example of creating a prediction error distribution. [Figure 13] FIG. 13 is a diagram showing an example of creating a forecast demand scenario. [Figure 14] FIG. 14 is a flowchart showing the flow of the forecast demand scenario setting process of the forecast demand scenario setting unit 21 according to the third embodiment. [Figure 15] FIG. 15 is a diagram showing an example of calculation of past prediction errors. [Figure 16] FIG. 16 is a diagram showing an example of similar date extraction. [Figure 17] FIG. 17 is a diagram showing a method for normalizing a prediction error. [Figure 18] FIG. 18 is a diagram illustrating an example of a target device according to the fourth embodiment. [Figure 19] FIG. 19 is a flowchart showing the flow of the forecast demand scenario setting process performed by the forecast demand scenario setting unit. [Figure 20] FIG. 20 is a diagram showing an example of selection of a priority target. [Figure 21] FIG. 21 is a block diagram illustrating a functional configuration of an equipment operation planning device according to the fifth embodiment. As illustrated in FIG. [Figure 22] FIG. 22 is a flowchart showing the flow of the forecast demand scenario setting process of the forecast demand scenario setting unit. [Diagram 23] FIG. 23 is a diagram showing an example of extraction of similar dates. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] (First embodiment) Fig. 1 is a diagram showing the overall configuration of a monitoring and control system 100 including an equipment operation planning device 1 according to a first embodiment. As shown in Fig. 1, the equipment operation planning device 1 of this embodiment included in the monitoring and control system 100 is installed in, for example, a target facility 2 together with a monitoring and control terminal 3, a target device 4, and a sensor 5.

[0012] 1 is merely an example, and does not limit the present embodiment. For example, the monitoring and control system 100 may be configured such that either or both of the equipment operation planning device 1 and the monitoring and control terminal 3 are cloud servers installed outside the target facility 2.

[0013] First, the target device 4 that is the target of the device operation plan by the device operation planning device 1 will be described.

[0014] Fig. 2 is a diagram showing an example of target equipment 4. The target equipment 4 in the example shown in Fig. 2 is, for example, water-cooled chillers 11 and 12 and an air-cooled HP (Heat Pump) chiller 13 that operate with power received by power receiving facility 10, and an absorption chiller 14 that is driven by city gas. The water-cooled chillers 11 and 12, the air-cooled HP chiller 13, and the absorption chiller 14 are devices that supply the produced chilled water to chilled water demand facility 16 via a chilled water header 15, which is a tank that stores chilled water.

[0015] In order to operate such target equipment 4 efficiently, the chilled water demand of the chilled water demand facility 16 over a certain future period (this period differs depending on the target equipment 4) is predicted, and then a desired evaluation index (for example, one-day equipment operating cost, CO 2 An optimal operation plan for the target device 4 is created in advance so as to minimize or maximize the environmental impacts (e.g., emissions, primary energy consumption, system efficiency, etc.), and the target device 4 is operated based on the operation plan.

[0016] Therefore, if the chilled water demand of the chilled water demand facility 16 were to completely match the predicted demand, the ideal evaluation index value assumed in the prior operation plan would be realized. However, in reality, the predicted demand never completely matches its true value, so the evaluation index value will be worse than the ideal evaluation index value. To minimize this deterioration, re-planning methods have been proposed, as described in the prior art, in which the operation plan is revised based on re-forecasting with reference to the most recent actual demand value, or operation planning methods that take into account the effects of forecast errors in advance.

[0017] In the following description of this embodiment, the target facility 2 described here is taken as an example, but this embodiment is merely an example and does not limit the prediction target or equipment configuration of the present invention. For example, the contents of this embodiment are applicable even when the prediction target is power demand or power generation from renewable energy sources such as PV (photovoltaic) and wind power, and power supply and demand are planned.

[0018] Next, the equipment operation planning device 1 will be described.

[0019] Fig. 3 is a block diagram showing a hardware configuration of the equipment operation planning device 1. As shown in Fig. 3, the equipment operation planning device 1 includes a control device such as a CPU (Central Processing Unit) 86, a storage device such as a ROM (Read Only Memory) 88, a RAM (Random Access Memory) 90, and a HDD (Hard Disk Drive) 92, an I / F unit 82 which is an interface with various devices, an output unit 80 which is a display unit or the like which outputs various information such as output information, an input unit 94 which accepts operations by a user, and a bus 96 which connects each unit, and has a hardware configuration using a normal computer.

[0020] In the equipment operation planning device 1, the CPU 86 reads out a program from the ROM 88 onto the RAM 90 and executes it, thereby realizing various functions, which will be described later, on the computer.

[0021] The programs for executing the various functions executed by the equipment operation planning device 1 may be stored in the HDD 92. Also, the programs for executing the various functions executed by the equipment operation planning device 1 may be provided by being pre-installed in the ROM 88.

[0022] Furthermore, the programs for executing the various functions executed by the equipment operation planning device 1 may be stored in a computer-readable storage medium such as a CD-ROM, CD-R, memory card, DVD (Digital Versatile Disc), or flexible disk (FD) in an installable or executable format file and provided as a computer program product. Furthermore, the programs for executing the various functions executed by the equipment operation planning device 1 may be stored on a computer connected to a network such as the Internet and provided by downloading the programs via the network. Furthermore, the programs for executing the various functions executed by the equipment operation planning device 1 may be provided or distributed via a network such as the Internet.

[0023] Next, various functions that are realized by the CPU 86 of the equipment operation planning device 1 reading a program from the ROM 88 onto the RAM 90 and executing it will be described.

[0024] Fig. 4 is a functional block diagram showing a functional configuration of the equipment operation planning device 1. As shown in Fig. 4, the equipment operation planning device 1 realizes a demand forecasting unit 20, a forecast demand scenario setting unit 21, a plan calculation unit 22, and a display control unit 23 by the CPU 86 executing a predetermined program. The equipment operation planning device 1 may realize a part or all of these functions by one or a plurality of processing circuits.

[0025] The demand forecasting unit 20 forecasts the future demand for a certain period of time based on the past actual demand up to the input prediction execution time. In addition to the past actual demand, weather forecast values ​​for the prediction target time may be input.

[0026] The forecast demand scenario setting unit 21 sets multiple forecast demand scenarios using past forecast errors on similar days extracted in order of similarity of past actual demand at any time, and the latest forecast demand predicted by the demand forecasting unit 20.

[0027] The plan calculation unit 22 uses the equipment conditions and operating conditions, which are parameters at the time of plan creation, and multiple forecast demand scenarios set by the forecast demand scenario setting unit 21, to calculate an equipment operation plan that minimizes or maximizes the desired future evaluation function value, taking into account forecast errors and their occurrence probability.

[0028] The display control unit 23 displays the operation plan calculated by the plan calculation unit 22 on the output unit 80.

[0029] Next, the flow of the operation planning process of the equipment operation planning device 1 will be described.

[0030] Fig. 5 is a flowchart showing the flow of the operation planning process of the equipment operation planning device 1. A series of sequences shown in the flow of Fig. 5 is executed at any time set in advance.

[0031] For example, in general, the equipment operation planning device 1 first sets 22:00 the night before the transition to the nighttime power unit price as a starting time, and immediately before that, performs demand forecasting and planning up to 22:00 the following day. The equipment operation planning device 1 executes the sequence shown in FIG. 5 at every predefined arbitrary timing on the following day, and performs re-forecasting and re-planning up to 22:00 on that day. In general, the equipment operation planning device 1 starts re-forecasting and re-planning a certain time after the demand rises on that day (for example, if the demand rises from 7:00 a.m., it starts from 9:00 a.m.). Thereafter, the equipment operation planning device 1 executes re-forecasting and re-planning every 30 minutes or every hour.

[0032] 5, first, the demand forecasting unit 20 of the equipment operation planning device 1 performs demand forecasting based on past actual demands using a known method such as that shown in Non-Patent Document 1 (step S1). Note that the demand forecasting unit 20 may take into account weather forecast data for the forecast target time, depending on the demand forecasting method.

[0033] Next, the forecast demand scenario setting unit 21 of the equipment operation planning device 1 sets a plurality of forecast demand scenarios using the past actual demand, the past forecast error, and the forecast demand output by the demand forecasting unit 20 in step S1 (step S2). Details of the processing operation in the forecast demand scenario setting unit 21 will be described later.

[0034] Next, the plan calculation unit 22 of the equipment operation planning device 1 calculates an optimal operation plan for the target equipment 4 using the equipment conditions and operation conditions, which are parameters at the time of plan creation, and multiple forecast demand scenarios set by the forecast demand scenario setting unit 21 (step S3).

[0035] Here, the mathematical problem and solution method for searching for a decision variable (here, the operating state of equipment) that minimizes or maximizes a desired evaluation function, such as an operational plan, is called mathematical programming. In particular, when the objective function or the coefficients of the constraint conditions include elements that vary probabilistically, it is called stochastic programming. In stochastic programming, it is common to convert the problem into a problem that does not include random variables (equivalent deterministic problem) and solve it, and a typical example is the redemption claim problem (also called a two-stage stochastic programming problem). A stochastic programming problem in which the coefficients of the constraint conditions are random variables for the probabilistic event w is shown in equations (1-1) to (1-3).

[0036]

number

[0037] The decision variable x in equation (1-1) is a discrete variable (0: stopped, 1: running) that represents the operating / stopped state of each device at each time, and a continuous variable that represents the device output. If the objective is to minimize the device operating costs, the coefficient vector c of the objective function is the unit price of the input energy required for the operation of the target device 4 by time period, and is the unit price of purchased electricity for electrically powered target device 4, and is the unit price of purchased gas for gas powered target device 4.

[0038] Equation (1-2) is an equality constraint including a probability event w, and is a constraint on the supply and demand balance of chilled water at each time. h(w) corresponds to a plurality of forecast demand scenarios output from the forecast demand scenario setting unit 21.

[0039] Equation (1-3) is an equality constraint and an inequality constraint that do not include the probability event w. For example, the former is a conversion equation for the input and output energy of the target device 4, and the latter is an upper and lower limit equation for the output of the target device 4.

[0040] Here, the constraint equation containing a random variable is not necessarily satisfied for all realized values. Therefore, in the constraint equation (1-2) containing this random variable, a recourse is introduced to correct the deviation from the difference between the two sides, and the equation is transformed into the following equation (1-4).

[0041]

number

[0042] From the above, the redemption claim problem is formulated as the following equations (1-5) to (1-7). The objective function (1-5) is the sum of the expected value of the redemption claim for the newly introduced probability event w (the second-stage optimization problem) in addition to the original equation (1-1) (the first-stage optimization problem).

[0043]

number

[0044] The newly introduced recourse variable y(w) is an element that is adjusted to satisfy the constraint condition equation (1-6) including the random variable, and is the output correction value of each target device 4 that is adjusted to maintain the chilled water supply and demand balance when an individual forecast demand scenario is realized.

[0045] q is the unit cost required for this output correction, and similarly to equation (1-1), it is the unit cost of purchased electricity for electrically driven target equipment 4, and the unit cost of purchased gas for gas driven target equipment 4. Therefore, the second term in equation (1-5) is the sum of the expected values ​​of the output correction costs of target equipment 4 corresponding to multiple forecast demand scenarios, and this can be calculated as the sum of the products of the output correction costs for each forecast demand scenario and the probability of that scenario occurring, added together for all scenarios.

[0046] The operation planning problem formulated above can be calculated using known methods such as a general-purpose mathematical optimization solver or the L-type method, which is a type of decomposition method. For details, please refer to Non-Patent Document 2 (Makoto Tanaka, Ryuta Takashima, Shigeki Shimaumi: "Mathematical Models of Energy Risk Management" (Asakura Publishing), (2018)).

[0047] The display control unit 23 of the equipment operation planning device 1 displays the operation plan calculated as described above on the output unit 80 (step S4).

[0048] In addition, re-planning can also be calculated in the same manner as above, with only the target time length of the plan being different. A distinctive element of this embodiment is the forecast demand scenario setting (step S2) in the forecast demand scenario setting unit 21, which is pre-processing for the operation plan calculation (step S3) during re-forecasting and re-planning. The operation of this function will be described in detail below.

[0049] Next, a method for setting a forecast demand scenario in the forecast demand scenario setting unit 21 in step S2 will be described in detail.

[0050] Fig. 6 is a flowchart showing the flow of the process of setting a forecast demand scenario by the forecast demand scenario setting unit 21. As shown in Fig. 6, the forecast demand scenario setting unit 21 first calculates the similarity of the past actual demand at the same time on a past day to the past actual demand on the current day (step S21), and extracts similar days in the past (step S22).

[0051] Here, the reason why the past actual demand is used as an index of similarity in this embodiment will be explained with reference to FIG.

[0052] FIG. 7 is a diagram for explaining the reason why the forecast error is inappropriate. FIG. 7 shows a past actual demand 30 for one day on a certain past day, a reforecast result 31 at 10:00 on the same day (assuming that the reforecast updates only the forecast demand for the next two hours), and its forecast error 32, and a reforecast result 33 at 12:00 on the same day, and its forecast error 34. If the current time is 12:00, the forecast error 32 is used to extract similar days and is a forecast error before the current time on the past day. The forecast error 34 is used to create a forecast demand scenario and is a forecast error after the current time on the past day. It is assumed that the past actual demand 30 is a typical actual demand pattern for past days over a certain period (taking into account seasonal fluctuations, data used for forecasting is usually a period of at most several months).

[0053] The actual demand 35 is the actual demand for the current day when explaining the operation of reforecasting and replanning. The actual demand 35 is smaller in the morning than the typical past actual demand 30, and tends to increase to almost the same level as the typical past actual demand 30 in the afternoon.

[0054] In this case, since demand forecasting generally tends to predict demand close to typical past actual demand 30, it can be assumed that a large prediction error occurred at 8:00, although this is not shown in Fig. 7. However, for this prediction error, as shown in Patent Document 1, by considering multiple predicted demand scenarios from past dates in which similar prediction errors occurred, it is possible to create an operation plan in advance that minimizes the impact of this prediction error.

[0055] Here, assume that predicted demand 36 has been obtained by reforecasting at 10:00. In reforecasting, the predicted value is revised by also referring to the actual demand immediately prior to that day (8:00 to 10:00 in this example), so the obtained predicted demand 36 is considered to be smaller than the prediction error at 8:00, and is a predicted demand that fits to a certain extent to the actual demand on that day. At 12:00, the prediction error 37 in the above-mentioned reforecast at 10:00 can be calculated (because the actual demand 35 after 10:00 is unknown at 10:00). The prediction error 37 is used as the reference value for extracting similar days, and is the prediction error before the current time on that day.

[0056] Here, assuming that the forecast error 37 observed immediately before and the forecast error 32 at the same time on the past day are almost equal, when creating a forecast demand scenario from a past date with a similar forecast error as in the conventional method, the forecast error 34 of the day on which a typical past actual demand 30 occurred is used as the standard to set the forecast demand scenario for the following time. Specifically, there is a risk that a forecast demand scenario 39 that takes into account the forecast error 34 of the past day into account for the latest reforecast demand 38 at 12 o'clock may be adopted. However, as described above, in reality, the demand on the same day increases from the afternoon to almost the same as the typical past actual demand 30, so the forecast demand scenario 39 becomes a scenario that is significantly different from the actual situation.

[0057] As described above, when reforecasting, the general tendency is to revise forecast values ​​so as to eliminate forecast errors immediately prior to the day. Therefore, when reforecasting is used in conjunction with the conventional method of creating a forecast demand scenario from past dates with similar forecast errors, a scenario that deviates significantly from reality may be set, which may ultimately cause a deterioration in the operational performance of the equipment.

[0058] For the reasons described above, the equipment operation planning device 1 in this embodiment proposes a method of creating a scenario of forecast demand from a past date in which actual demand is similar.

[0059] First, extraction of similar dates will be described.

[0060] FIG. 8 is a diagram showing an example of similar day extraction. FIG. 8 shows an example of similar day extraction when the current time is October 1st, 12:00 and the population of past data is the past six days. Here, the actual demand difference is used as an index of similarity. The actual demand difference shown in FIG. 8 is a value calculated, for example, by equation (1-8). The forecast error shown in FIG. 8 is a value calculated, for example, based on equation (1-9). The forecast error shown in FIG. 8 is used to create a forecast demand scenario, and is a forecast error from the current time onwards on a past day.

[0061]

number

[0062] In addition, in formula (1-8), the actual demand difference is calculated based on the cumulative demand at the evaluation target time, but this method is not limited to this. For example, if the similarity is calculated based on the past actual demand value, such as using the demand average value, this method is not limited to this.

[0063] 8 shows an example in which the actual demand differences for the past week are sorted in ascending order. The forecast demand scenario setting unit 21 of this embodiment extracts an arbitrary number of top past days as similar days.

[0064] The top three days extracted in the example of Figure 8 are days where the difference in actual demand from actual demand 35 in Figure 7 is small, in other words, days where actual demand in the morning is small compared to the typical demand 30. On such days, a tendency for large positive forecast errors (actual demand values ​​are larger than forecast values ​​when compared with formula (1-9)) to occur in the following hours can be extracted as similar days.

[0065] In addition, in formula (1-8), similar days are extracted by evaluating the actual demand difference for the previous two hours, but this target time length (any time to be evaluated) may be changed according to the degree of demand change at the operation execution timing. In other words, when calculating the similarity of past actual demand for any time, the forecast demand scenario setting unit 21 may be able to adjust the any time to be evaluated based on the degree of demand change immediately before.

[0066] Here, Fig. 9 is a diagram showing an example of sampling time of the past actual demand difference. The horizontal axis of Fig. 9 represents the demand change in the last 30 minutes of the day, and is calculated based on, for example, formula (1-10).

[0067]

number

[0068] In Figure 9, the demand change ΔD for the last 30 minutes of the day A exceeds a certain threshold value Δa or Δb, the sampling time for the past actual demand difference is shortened from 2.0 (h) to 1.5 (h) and 1.0 (h) based on a predefined set time 40. In this way, for example, in a time period after 12:00 in Fig. 7 where there is a large change in demand, it becomes possible to extract past days whose recent trends show a similar trend as similar days in response to this change.

[0069] Returning to FIG. 6, after the above processes, the forecast demand scenario setting unit 21 creates a forecast demand scenario based on the extracted similar dates (step S23).

[0070] Here, Fig. 10 is a diagram showing an example of creating a forecast demand scenario. The forecast errors occurring on the top three similar days shown in Fig. 8 are used to create a forecast demand scenario, and are forecast errors from the current time on a past day onward. For forecast demand scenarios 41 to 43, forecast demand scenarios 41 (September 28th), forecast demand scenario 42 (September 29th), and forecast demand scenario 43 (September 27th) are created based on the forecast errors occurring on the top three similar days shown in Fig. 8, for example, by adding the re-forecast errors for each time on past similar days to the re-forecast demand 38 for the current day, using equation (1-11).

[0071]

number

[0072] In the case of this embodiment, it is sufficient to formulate the expected value, which is the second term of the objective function, equation (1-5), on the assumption that each forecast demand scenario occurs with a probability of 1 / 3 in the operation plan.

[0073] In this way, according to the equipment operation planning device 1 of the first embodiment, by creating a forecast demand scenario from past dates with similar actual demand, even when re-forecasting, which acts to eliminate last-minute forecast errors, is used in combination, possible forecast errors can be appropriately evaluated and equipment operation can be realized through rational re-planning.

[0074] Second embodiment Next, a second embodiment will be described.

[0075] The second embodiment differs from the first embodiment in the flow of the process of setting a forecast demand scenario by the forecast demand scenario setting unit 21. In the following explanation of the second embodiment, the explanation of the same parts as in the first embodiment will be omitted, and only the parts that are different from the first embodiment will be explained.

[0076] FIG. 11 is a flowchart showing the flow of the forecast demand scenario setting process of the forecast demand scenario setting unit 21 according to the second embodiment.

[0077] In the first embodiment, the forecast demand scenario setting unit 21 extracts past days with similarity for an arbitrary number of days, and creates a forecast demand scenario by directly using the forecast errors for those past days. In other words, there is a one-to-one relationship between the extracted similar days and the created forecast demand scenario.

[0078] On the other hand, as shown in Fig. 11, the forecast demand scenario setting unit 21 of this embodiment, like the first embodiment, first calculates the similarity of the past actual demand at the same time on a past day to the past actual demand of the current day (step S31), then extracts multiple similar past days (step S32) and creates a forecast error distribution for the following time (step S33). Through the above processes, the forecast demand scenario setting unit 21 creates a forecast demand scenario based on the extracted similar days (step S34).

[0079] Fig. 12 is a diagram showing an example of creating a prediction error distribution. The example of creating a prediction error distribution in step S33 shown in Fig. 12 is an example in which the current time is set to 12 o'clock on a certain day, 46 similar days are extracted in the same manner as in the first embodiment, and a prediction error distribution of those similar days at subsequent times is created. Note that the prediction error here is a value calculated, for example, by the above-mentioned formula (1-9).

[0080] In Fig. 12, the prediction error class of 150 to 250 kWh is extracted for four days (reference number 50). An example of the prediction error by time for every 30-minute planning unit for the four days included in these specific distribution classes is shown in Fig. 13.

[0081] FIG. 13 is a diagram showing an example of creating a forecast demand scenario. As shown in the rightmost column of FIG. 13, the cumulative forecast error from 12:00 to 14:00 for four days included in these specific distribution classes is in the range of 150 to 250 kWh. In this embodiment, for example, the forecast errors for four days included in these specific distribution classes are averaged by time to obtain the hourly past forecast error 51 in this forecast error class of 150 to 250 kWh. A forecast demand scenario corresponding to this class is created by adding the hourly past forecast error 51 to the latest re-forecast result executed at 12:00 on the day, as in the formula (1-11) of the first embodiment. In the operation plan, it is sufficient to formulate the expected value assuming that this forecast demand scenario occurs with a probability of 4 / 46.

[0082] In this way, according to the equipment operation planning device 1 of the second embodiment, it is possible to create a wide range of forecast demand scenarios from past dates with similar actual demands. Furthermore, by setting a forecast demand scenario for each class of the forecast error distribution, it is possible to reduce the problem scale of the operation plan compared to the first embodiment in which a forecast demand scenario is set for each similar date to be extracted. Therefore, according to the equipment operation planning device 1 of the second embodiment, even when re-forecasting that acts to eliminate the most recent forecast error is used in combination, it is possible to calculate an operation plan in a shorter time after appropriately and widely evaluating possible forecast errors.

[0083] (Third embodiment) Next, a third embodiment will be described.

[0084] The third embodiment differs from the first and second embodiments in the flow of the forecast demand scenario setting process of the forecast demand scenario setting unit 21. In the following explanation of the third embodiment, explanations of the same parts as in the first and second embodiments will be omitted, and only the parts that differ from the first and second embodiments will be explained.

[0085] FIG. 14 is a flowchart showing the flow of the forecast demand scenario setting process of the forecast demand scenario setting unit 21 according to the third embodiment.

[0086] In the first and second embodiments, a forecast demand scenario was created from a past date with similar immediately preceding actual demand, but in the third embodiment, in addition to this, similar dates are extracted by taking into account the immediately preceding past forecast error, and a forecast demand scenario is created.

[0087] As shown in Figure 14, the forecast demand scenario setting unit 21 of this embodiment, like the first embodiment, first calculates the similarity between the past actual demand at the same time on a past day and the current day's past actual demand (step S41), and then, in addition, calculates the similarity to the past forecast error (step S42). This is a difference.

[0088] Fig. 15 is a diagram showing an example of calculation of past prediction errors. The calculation example of past prediction errors in step S42 shown in Fig. 15 is a compilation of past reprediction errors in which reprediction is performed at 30-minute intervals from 10:00, assuming that the present time is 12:00 on a certain day. The column direction shown in Fig. 15 indicates the prediction target time of the reprediction performed at each time.

[0089] For example, in the case of 10:00 re-prediction, there are 4 time steps of re-prediction errors in 30-minute increments from 10:00 to 12:00. On the other hand, in the case of 10:30 re-prediction, since 10:00 to 10:30 has already passed, there are 3 time steps of re-prediction errors in the remaining 30-minute increments from 10:30 to 12:00. In this way, when calculating past re-prediction errors, even for the same prediction target time, there are multiple prediction errors depending on the execution time of re-prediction. Here, to calculate the re-prediction error from 10:00 to 12:00, the average value and integrated value calculated using all the numerical values ​​shown in FIG. 15 are used, or the average or integrated value calculated using only the latest prediction error 65 at the past time indicated by a circle in FIG. 15 may be evaluated as the prediction error.

[0090] In other words, the forecast demand scenario setting unit 21 uses the average error, mean absolute error, root mean square error, and accumulated error calculated from all forecast errors of multiple past re-forecasts, or the average error, mean absolute error, root mean square error, and accumulated error calculated using only the latest forecast value at each past point in time, as similarity indexes of past forecast errors when extracting similar days.

[0091] After the above processes, the forecast demand scenario setting unit 21 extracts similar dates in the past using the calculated similarities between the past actual demands and the past forecast errors (step S43).

[0092] Fig. 16 is a diagram showing an example of similar day extraction. In the example of similar day extraction in step S43 shown in Fig. 16, the vertical axis is the demand difference ratio A, which is a similarity index calculated from the above-mentioned past actual demand, and the horizontal axis is the forecast error ratio P, which is a similarity index calculated from the past forecast error, and the data of the past days to be evaluated are plotted at the corresponding positions with circles. Here, considering the case where the similarity calculation methods for the demand difference and the forecast error are different, it is better to normalize both indicators according to the maximum value 61 / minimum value 60 of the forecast error of the data population. More specifically, the demand difference is normalized by the maximum value 61, and the forecast error is normalized by the difference between the maximum value 61 and the minimum value 60.

[0093] Fig. 17 is a diagram showing a method for normalizing a forecast error. If the minimum value 60 of the forecast error p of the data population is -200 kWh and the maximum value 61 is 600 kWh, the normalized forecast error ratio can be normalized using the method shown in Fig. 17. On the plane of Fig. 16 obtained in this way, similar days can be extracted in order of closest straight-line distance from the current day data 62. This straight-line distance can be calculated using formula (1-12).

[0094]

number

[0095] Here, by introducing sensitivity adjustment coefficients a and b into formula (1-12), it is possible to adjust whether to place more weight on past actual demand or past forecast error as the similar days to be extracted. For example, if it is the morning before the start of reforecasting on the day, by placing more weight on past forecast error, a forecast demand scenario can be created from a past date with similar past forecast error, which is the conventional technology, and if it is after the start of reforecasting on the day, by placing more weight on past actual demand, a forecast demand scenario can be created from a past date with similar past actual demand as described in the first and second embodiments.

[0096] By setting the above adjustment coefficients to different values ​​for each time of day in this way, it is possible to use the conventional method and the proposed method in a unified manner without the need to switch between them.

[0097] Note that the forecast demand scenario setting unit 21 creates a forecast demand scenario based on the extracted similar dates (step S44), similar to the first and second embodiments.

[0098] In this way, according to the third embodiment of the equipment operation planning device 1, it is possible to uniformly use the conventional method of creating a forecast demand scenario from past dates with similar past forecast errors, and the proposed method of creating a forecast demand scenario from past dates with similar past actual demand, and it is possible to provide an equipment operation planning device 1 that can be applied throughout the entire day, regardless of before or after the start of re-prediction.

[0099] (Fourth embodiment) Next, a fourth embodiment will be described.

[0100] In the fourth embodiment, there are multiple forecast targets, and the flow of the forecast demand scenario setting process of the forecast demand scenario setting unit 21 differs from the first to third embodiments. In the following explanation of the fourth embodiment, explanations of the same parts as in the first to third embodiments will be omitted, and only the differences from the first to third embodiments will be explained.

[0101] Fig. 18 is a diagram showing an example of a target device 4 according to the fourth embodiment. As shown in Fig. 18, in this embodiment, in addition to the cold water demand facility 16 described in the first to third embodiments, a hot water demand facility 71 is included as a prediction target.

[0102] 18 includes, for example, water-cooled chillers 11 and 12 and an air-cooled HP chiller 13 that operate on power received by a power receiving facility 10, as well as an absorption type chiller / heater 72 that runs on city gas. The water-cooled chillers 11 and 12 are devices that supply produced chilled water to chilled water demand facility 16 via a chilled water header 15. The air-cooled HP chiller 13 and the absorption type chiller / heater 72 are devices that supply produced hot water to hot water demand facility 71 via a hot water header 70, which is a tank that stores hot water. That is, this embodiment shows a case where there are multiple targets (chilled water / hot water demand in this case) for which fluctuation probabilities are considered, and these are non-independent events.

[0103] 18, when there are multiple forecast targets, if the two are independent (not correlated with each other) events, similar days can be extracted individually and a forecast demand scenario can be created in the same way as in the first embodiment. However, in general, there are many cases where demands within the same facility are correlated with each other, and these are non-independent events (for example, a correlation may arise between the two demands due to a common cause such as the outside temperature or the number of people in the facility).

[0104] In this embodiment, a case will be described in which the random variables taking into account the prediction errors are multiple and non-independent.

[0105] Fig. 19 is a flowchart showing the flow of the process of setting a forecast demand scenario in the forecast demand scenario setting unit 21. As shown in Fig. 14, unlike the first embodiment, the forecast demand scenario setting unit 21 of this embodiment adds a process (step S51) of selecting a priority target for extracting similar dates prior to calculating the similarity of past demand results (step S52).

[0106] Note that, while FIG. 19 adds a process (step S51) for selecting a priority target for extracting similar days to the base of the first embodiment, the equipment operation planning device 1 of this embodiment is not limited to this operating procedure, and it is also possible to add a process (step S51) for selecting a priority target for extracting similar days to the second and third embodiments.

[0107] Fig. 20 is a diagram showing an example of a priority target selection. Fig. 20 shows a daily cold water demand 73 and hot water demand 74, and the total amounts of these are SC and SH. One example of a priority target selection by the forecast demand scenario setting unit 21 in step S51 is to prioritize the one with the larger total amount depending on the magnitude of SC and SH. Other examples of selection by the forecast demand scenario setting unit 21 in step S51 include a case where the judgment is made based on the magnitude of the supply cost of each of SC and SH, a case where the judgment is made based on the magnitude of the marginal cost obtained by dividing the supply cost by the total demand amount, or a case where the user arbitrarily selects one.

[0108] In other words, the forecast demand scenario setting unit 21 allows the user to select the priority target used to extract similar dates from the size of the target demand amount, the cost required to supply the target demand or the marginal cost, or any other value selected by the user.

[0109] After selecting the priority targets by the above method, the forecast demand scenario setting unit 21 creates forecast demand scenarios in the same manner as in the first embodiment (steps S52 to S54). Therefore, forecast demand scenarios for non-priority targets are determined subordinately from the similar dates extracted for the priority targets.

[0110] In this way, according to the equipment operation planning device 1 of the fourth embodiment, even when there are multiple non-independent prediction targets, by creating a forecast demand scenario from a past date with similar prioritized actual demand, it is possible to appropriately evaluate possible prediction errors and realize rational re-planning of equipment operation, even when re-prediction that acts to eliminate immediate prediction errors is used in combination.

[0111] Fifth embodiment Next, a fifth embodiment will be described.

[0112] The fifth embodiment differs from the first to third embodiments in that a forecast demand scenario is created based on past integrated index values ​​in the forecast demand scenario setting unit 21. In the following explanation of the fifth embodiment, explanations of the same parts as the first to third embodiments will be omitted, and only differences from the first to third embodiments will be explained.

[0113] Fig. 21 is a block diagram showing a functional configuration of an equipment operation planning device 1 according to the fifth embodiment. As shown in Fig. 21, the fifth embodiment is different from the first to third embodiments in that a forecast demand scenario setting unit 21 creates a forecast demand scenario based on a past integrated index value.

[0114] The forecast demand scenario setting unit 21 of this embodiment creates a forecast demand scenario based on a past integrated index value, instead of selecting a priority target as described in the third embodiment. This past integrated index value is a converted index value on the upper side of the target system, and is, for example, the total equipment operating cost, primary energy consumption, and CO 2 It may also be an environmental indicator value such as an emission amount.

[0115] Fig. 22 is a flowchart showing the flow of the process of setting a forecast demand scenario by the forecast demand scenario setting unit 21. As shown in Fig. 22, unlike the first embodiment, the forecast demand scenario setting unit 21 of this embodiment first calculates the similarity of past integrated index values ​​(step S61). Next, the forecast demand scenario setting unit 21 extracts similar days based on the calculated similarity of the past integrated index values ​​(step S62).

[0116] Fig. 23 is a diagram showing an example of similar day extraction. The similar day extraction example in step S62 shown in Fig. 23 shows an example in which the past integrated index value is broken down by energy type, such as electricity bill and gas bill. Of course, the combined operating cost of these electricity bills and gas bills may be used as the past integrated index value. The vertical axis of Fig. 23 is electricity bill, and the horizontal axis is gas bill. If the present time is 12 o'clock on a certain day, the accumulated electricity and gas costs from 10 to 12 o'clock on the current day and on past days are plotted in Fig. 23.

[0117] The forecast demand scenario setting unit 21 extracts, as similar days, days that are closest in straight-line distance to the day's estimated cost 85 in the same manner as described in the third embodiment.

[0118] After extracting the similar dates, the forecast demand scenario setting unit 21 creates a forecast demand scenario in the same manner as in the first to third embodiments (step S63).

[0119] In this way, according to the equipment operation planning device 1 of the fifth embodiment, even when there are multiple non-independent prediction targets, a scenario of predicted demand is created from past dates for which the integrated index value incorporating the trends of both is similar, without selecting a prioritized prediction target, so that even when re-prediction that acts to eliminate immediately preceding prediction errors is used in combination, possible prediction errors can be appropriately evaluated and equipment operation can be realized through rational re-planning.

[0120] Although some embodiments of the present invention have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their modifications are included in the scope and spirit of the invention, and are included in the scope of the invention and its equivalents described in the claims. [Explanation of symbols]

[0121] 1. Equipment operation planning device 20 Demand Forecasting Department 21 Demand forecast scenario setting section 22 Planning Calculation Department

Claims

1. a demand forecasting unit for forecasting a forecast demand for a certain period of time in the future; a forecast demand scenario setting unit that sets a plurality of forecast demand scenarios using past forecast errors on similar dates extracted in order of similarity of past actual demands at any given time and the latest forecast demand predicted by the demand forecasting unit; a plan calculation unit that calculates an operation plan for an apparatus that minimizes or maximizes a desired evaluation function value in the future, taking into account a forecast error and its occurrence probability, by using the plurality of forecast demand scenarios set by the forecast demand scenario setting unit; An equipment operation planning device comprising:

2. a demand forecasting unit for forecasting a forecast demand for a certain period of time in the future; a forecast demand scenario setting unit that sets a plurality of forecast demand scenarios using past forecast errors by time for each distribution class obtained by averaging forecast errors for a plurality of similar days included in a specific distribution class in a forecast error distribution for a plurality of similar days extracted in order of similarity of past actual demand at an arbitrary time, and the latest forecast demand predicted by the demand forecasting unit; a plan calculation unit that calculates an operation plan for an apparatus that minimizes or maximizes a desired evaluation function value in the future, taking into account a forecast error and its occurrence probability, by using the plurality of forecast demand scenarios set by the forecast demand scenario setting unit; An equipment operation planning device comprising:

3. the forecast demand scenario setting unit is capable of adjusting the arbitrary time to be evaluated based on a degree of change in demand immediately before, when calculating a similarity of past actual demand for an arbitrary time; 3. The equipment operation planning apparatus according to claim 1 or 2.

4. a demand forecasting unit for forecasting a forecast demand for a certain period of time in the future; a forecast demand scenario setting unit that sets a plurality of forecast demand scenarios using past actual demands at any time and past forecast errors on similar dates extracted in order of similarity of past forecast errors, and the latest forecast demand predicted by the demand forecasting unit; a plan calculation unit that calculates an operation plan for an apparatus that minimizes or maximizes a desired evaluation function value in the future, taking into account a forecast error and its occurrence probability, by using the plurality of forecast demand scenarios set by the forecast demand scenario setting unit; An equipment operation planning device comprising:

5. a demand forecasting unit for forecasting a forecast demand for a certain period of time in the future; a forecast demand scenario setting unit that sets a plurality of forecast demand scenarios using past forecast errors by time for each distribution class obtained by averaging forecast errors for a plurality of similar days included in a specific distribution class in a forecast error distribution for a plurality of similar days extracted in order of similarity of past actual demand and past forecast errors at an arbitrary time, and the latest forecast demand predicted by the demand forecasting unit; a plan calculation unit that calculates an operation plan for an apparatus that minimizes or maximizes a desired evaluation function value in the future, taking into account a forecast error and its occurrence probability, by using the plurality of forecast demand scenarios set by the forecast demand scenario setting unit; An equipment operation planning device comprising:

6. the forecast demand scenario setting unit has an adjustment coefficient capable of changing the weighting of the past actual demand and the past forecast error when extracting similar dates; 6. An equipment operation planning apparatus according to claim 4 or 5.

7. The average error, mean absolute error, root mean square error, and cumulative error calculated from all forecast errors of multiple past re-forecasts, or the average error, mean absolute error, root mean square error, and cumulative error calculated using only the latest forecast value at each past point in time are used as a similarity index of past forecast errors when extracting similar days.

6. An equipment operation planning apparatus according to claim 4 or 5.

8. a demand forecasting unit for forecasting a forecast demand for a certain period of time in the future; a forecast demand scenario setting unit that selects from a plurality of priority targets to be used for extracting similar days, and sets a plurality of forecast demand scenarios for priority targets and non-priority targets using past forecast errors for the similar days extracted in order of similarity of past actual demands at any time selected by priority and the latest forecast demand predicted by the demand forecasting unit; a plan calculation unit that calculates an operation plan for an apparatus that minimizes or maximizes a desired evaluation function value in the future, taking into account a forecast error and its occurrence probability, by using the plurality of forecast demand scenarios set by the forecast demand scenario setting unit; An equipment operation planning device comprising:

9. a demand forecasting unit for forecasting a forecast demand for a certain period of time in the future; a forecast demand scenario setting unit which sets a plurality of forecast demand scenarios for the priority and non-priority targets using past forecast errors by time for each distribution class obtained by selecting from a plurality of priority targets to be used for extracting similar days, and averaging the forecast errors of the plurality of similar days included in a specific distribution class in the forecast error distribution of the priority and non-priority targets for the plurality of similar days extracted in order of similarity of past actual demand for an arbitrary time selected by priority, and the latest forecast demand predicted by the demand forecasting unit; a plan calculation unit that calculates an operation plan for an apparatus that minimizes or maximizes a desired evaluation function value in the future, taking into account a forecast error and its occurrence probability, by using the plurality of forecast demand scenarios set by the forecast demand scenario setting unit; An equipment operation planning device comprising:

10. The forecast demand scenario setting unit can select a priority target used for extracting similar dates from the size of the target demand amount, the cost required to supply the target demand, or the marginal cost, or can select it arbitrarily by the user.

10. The equipment operation planning apparatus according to claim 8 or 9.

11. a demand forecasting unit for forecasting a forecast demand for a certain period of time in the future; a forecast demand scenario setting unit that sets a plurality of forecast demand scenarios using past forecast errors on similar dates extracted in order of similarity of past actual demands at any time including the influence of a plurality of targets and the latest forecast demand predicted by the demand forecasting unit; a plan calculation unit that calculates an operation plan for an apparatus that minimizes or maximizes a desired evaluation function value in the future, taking into account a forecast error and its occurrence probability, by using the plurality of forecast demand scenarios set by the forecast demand scenario setting unit; An equipment operation planning device comprising:

12. a demand forecasting unit for forecasting a forecast demand for a certain period of time in the future; a forecast demand scenario setting unit that sets a plurality of forecast demand scenarios using past forecast errors by time for each distribution class obtained by averaging forecast errors for a plurality of similar days included in a specific distribution class in a forecast error distribution for a plurality of similar days extracted in order of similarity of past integrated index values ​​at any time including the influence of a plurality of targets, and the latest forecast demand predicted by the demand forecasting unit; a plan calculation unit that calculates an operation plan for an apparatus that minimizes or maximizes a desired evaluation function value in the future, taking into account a forecast error and its occurrence probability, by using the plurality of forecast demand scenarios set by the forecast demand scenario setting unit; An equipment operation planning device comprising:

13. The forecast demand scenario setting unit uses the past integrated index value to calculate the total equipment operating costs, primary energy consumption, and CO ２ Use the amount of emissions as an environmental indicator value.

13. The equipment operation planning device according to claim 12.

14. The forecast demand scenario setting unit breaks down the past integrated index value by energy type and extracts similar days.

14. The equipment operation planning device according to claim 13.

15. A demand forecasting step in which a demand forecasting unit forecasts a forecast demand for a certain period of time in the future; a forecast demand scenario setting step in which a forecast demand scenario setting unit sets a plurality of forecast demand scenarios using past forecast errors on similar dates extracted in order of similarity of past actual demands at any given time and the latest forecast demand predicted by the demand forecasting unit; a plan calculation step in which a plan calculation unit calculates an operation plan of an equipment that minimizes or maximizes a desired evaluation function value in the future by using the plurality of forecast demand scenarios set by the forecast demand scenario setting unit and taking into account a forecast error and its occurrence probability; An equipment operation planning method comprising:

16. Computer, a demand forecasting unit for forecasting a forecast demand for a certain period of time in the future; a forecast demand scenario setting unit that sets a plurality of forecast demand scenarios using past forecast errors on similar dates extracted in order of similarity of past actual demands at any given time and the latest forecast demand predicted by the demand forecasting unit; a plan calculation unit that calculates an operation plan for an apparatus that minimizes or maximizes a desired evaluation function value in the future, taking into account a forecast error and its occurrence probability, by using the plurality of forecast demand scenarios set by the forecast demand scenario setting unit; A program to function as a

Images