JP7184716B2 - Distributed resource adjustability management device and adjustability management method - Google Patents

Description

TECHNICAL FIELD The present invention relates to a distributed resource control system and control control method, and supplies control power to a system by means of energy resources such as a group of power storage equipment, a group of electric vehicles, and a group of loads that are distributed. The present invention relates to a coordination power management device and a coordination power management method for distributed resources.

There is a technique disclosed, for example, in US Pat. Patent Literature 1 states, "Even if the power consumption of a home appliance to be controlled in this system information monitoring system is changed instantaneously, if it consumes the energy required for each specified period, the intended use of the home appliance is not achieved." achievable controllable load. In such a controllable load, the power consumption rate γ calculated from the power _consumption P, the future power consumption average value P _fut , the maximum power consumption P _max and the minimum power consumption P min = (P _fut - P _min )/(P _max - P _min ). To reduce power consumption, select the controllable loads with the lowest power consumption rate γ, and to increase the power consumption, select the controllable loads with the highest power consumption rate γ. maintain. By always maintaining γ between 0 and 1, it is possible to consume the energy required for each prescribed period, and the convenience of home appliance users is maintained.” See paragraph 0005). In addition, in Patent Document 1, "Basically, a plurality of controllable loads calculates the power consumption rate γ = (P _fut - P _min ) / (P _max - P _min ) of the load, and the system operation device based _{on the} power consumption rate γ received from each controllable load. Based on the histogram, the operation device calculates a threshold value of γ from the power consumption adjustment amount ΔP necessary for system operation, and the controllable load performs power consumption control based on this threshold value. ] is disclosed (see paragraph 0010 of Patent Document 1).

JP 2010-068704 A

However, in the above-mentioned Patent Document 1, it is shown that when power consumption is to be reduced, loads should be selected in order from the lowest power consumption rate. It is necessary to build a model for each home appliance that can monitor power consumption and calculate future average power consumption. In the case of home appliances such as single-use water heaters, it is possible to build a relatively accurate model by learning based on the actual usage data of individual consumers. It cannot be applied when it is difficult to obtain detailed information on operation from the viewpoint of

The purpose of the present invention is to consider the above-mentioned points, and it is difficult to construct a detailed operation model for each device, and it is desirable to construct an operation model due to operational constraints such as personal information protection. It is an object of the present invention to provide a load balance management device and load balance management method for distributed resources that enable prediction and control of load balance as a group of loads even when they are not mixed.

In order to solve such a problem, in the present invention, as a means for solving the object, the adjustability management device for managing the adjustability of the distributed resources includes a behavior simulation system that comprehensively simulates the time distribution of the operation state of the distributed resources. and a controllability trial calculation unit that estimates the controllability that can be supplied by the distributed resource based on the time distribution of the operating state of the distributed resource that is simulated by the behavior simulating unit.

According to the present invention, for example, even if it is difficult to build a detailed operation model for each device, or if construction of an operation model is not desired due to operational constraints such as personal information protection, load groups can be allows prediction and control of the adjustability of

1 is a diagram showing an example of a configuration of a control power management device according to Embodiment 1; FIG. FIG. 4 is a diagram showing an example of a configuration method of a state transition probability model when an EV is used as a control power supply resource in the behavior simulating unit of the first embodiment; 4 is a flowchart showing an example of control power trial calculation processing according to the first embodiment; FIG. 4 is an explanatory diagram of an example of adjusting force calculation (first step) of the maximum supplyable amount of the first embodiment; FIG. 5 is a diagram showing an example of a result of adjustment force calculation (first step) of the maximum suppliable amount according to the first embodiment; FIG. 5 is a diagram showing an example of a result of adjustment force calculation (second step) according to the first embodiment; The figure which shows an example of a mutual transition. The figure which shows an example of the adjustment calculation method of mutual transition. FIG. 8 is a diagram showing a modification of a model configuration method when EVs are used as control power supply resources in the behavior simulating unit of the first embodiment; The figure which shows an example of a structure of the control power management apparatus of Embodiment 2. FIG. The figure which shows an example of the hardware constitutions of a control power management apparatus.

Embodiments of the present invention will be described below with reference to the drawings. In the following description, the same or similar elements and processes are in principle given the same reference numerals, and duplicate descriptions are omitted. Also, the configuration and processing described below are merely examples, and are not intended to limit the embodiments according to the present invention to the following embodiments. Moreover, each embodiment and modifications can be combined in whole or in part within the scope of the technical idea of the present invention and within the scope of matching.

Each drawing referred to in the following description outlines the embodiments to the extent necessary for understanding and implementing the present invention, and the scope of the present invention is not limited to the configuration shown in each drawing.

[Embodiment 1]
Hereinafter, prediction and management of controllability when using an EV (Electric Vehicle) as a target distributed resource will be described as a first embodiment.

FIG. 1 is a diagram showing an example of the configuration of a control power management device 100 according to the first embodiment. As shown in FIG. 1, the control power management device 100 includes a behavior simulation unit 110, a parameter storage unit 120, a control power trial calculation unit 130, a control power control command unit 140, a result collection unit 150, and a model correction unit 160. be.

In FIG. 1, EVs 171, 172, 173, and 174 shown as examples of distributed resources are connected to the controllability management device 100 by a communication means (not shown), and the states before and during controllability activation are displayed as resources. control force management device 100 from the control power management device 100, and control force control commands are transmitted from the control force management device 100 to EVs 171-174. Also, although not shown, the EVs 171, 172, 173, and 174 are connected to the power system when connected to the charging facility. Although four EVs EV171 to EV174 are shown in FIG. 1, the number is not limited to four.

The behavior simulating unit 110 consists of one or more virtual avatar groups 111 , 112 , 113 . In FIG. 1, there are three virtual avatar groups, but the number is not limited to this.

In the behavior simulating unit 110 according to the present embodiment using an EV as a resource, the SoC (State of Charge) of the EV and operational states such as charging, plugging in (non-charging), stopping (plugging out), and running. , a state transition probability model or the like as shown in FIG. FIG. 2 is a diagram showing an example of a configuration method of a state transition probability model when EVs 171 to 174 are used as control power supply resources in the behavior simulating unit 110 of the first embodiment.

FIG. 2 shows a state 311 in which the SoC is i at time t. ], R[t, i] during running, and assumes that only one state transition occurs in one unit time step. The 〇 (circle) symbol represents the state, and the arrow represents the transition.

State 311 at time t is not only transition from state 312 where SoC is i at time t−1, but also state 322 where SoC is i−1 at time t−1 and state 322 where SoC is i−1 at time t−1. It indicates that there is a transition out of state 332 at i+1. Conversely, the transition from state 311 includes not only a transition to state 313 with SoC i at time t+1, but also a transition to state 333 with SoC i+1 and state 323 with SoC i−1 at time t+1. It is shown that.

In the case of the state transition probability model, some discretization is required for SoC and operational states, but in the case of SoC, it is possible to use battery segments or even or non-uniformly with respect to the percentage of the amount of charge to 100% capacity. You may express so that it may be divided into several steps.

The operating state may also be divided into stopping and charging at the home location and stopping and charging at other locations, or charging at a high-speed charging facility and charging at a normal speed charging facility.

Time is also discretized and modeled. It is assumed that the state transition occurs at the moment when the time changes without considering the state change between the times t−1, t, and t+1. It is assumed that the time difference between time t−1 and time t and the time difference between time t and time t+1 are equal to Δt.

Returning to FIG. The behavior simulating unit 110 calculates the time distribution of discrete operational states of the plurality of resources (EV171-EV174) using a state transition probability model. For the calculation, each EV is used for typical operational use cases, such as commercial vehicles such as taxis, vehicles for the transport of cargo and mail, private cars, and vehicles that operate regularly, such as buses. It may be modeled by dividing it into multiple groups according to.

The behavior simulating unit 110 includes three types of virtual avatar groups 111, 112, and 113 corresponding to three types of groups assuming such differences in operation. The virtual avatar groups 111, 112, and 113 are groups of individual distributed resources that are grouped according to use cases and combinations, and are resources that are collectively targeted for predicting and commanding coordinating power.

Avatar group parameters 122, 123, and 124 corresponding to the virtual avatar groups 111, 112, and 113 are stored in the parameter storage unit 120 in order to simulate different behaviors for each group. In addition, the parameter storage unit 120 stores avatar group meta-parameters 125 in a form in which parameters common to the avatar group parameters 122, 123, and 124 are factored out separately.

In addition, the parameter storage unit 120 stores a forecast period 211, which will be described later, for confirming the suppliable amount of control power, and a reserve rate that determines how much margin is taken into consideration for the control power request amount when issuing commands. Avatar group real resource association parameters 121 such as .

When simulating behavior with a state transition probability model, a simulation using pseudo-random numbers is used. For this reason, for each of the virtual avatar groups 111, 112, and 113, not only are a plurality of resources assumed, but simulations under the same conditions are performed a plurality of times, and an average expected adjustability and a predetermined probability or more are obtained. The adjustability and the like that can be implemented in , are calculated by the adjustability trial calculation unit 130, which will be described later.

The control power trial calculation unit 130 calculates power consumption by charging as control power at each time based on the discretized states of the plurality of resources belonging to the virtual avatar groups 111, 112, and 113 calculated by the behavior simulating unit 110 at each time. A trial calculation is made of the controllability that can be supplied to the controllability market in the case of negawatt operation for power reduction and power generation, and posiwatt operation for charging and increasing power consumption.

Next, the adjustment force trial calculation method of the first embodiment will be described.

The behavior simulator 110 performs prediction calculations of future operational states and SoC states under current avatar group parameters 122 , 123 , 124 and avatar group meta-parameters 125 . Behavior simulating unit 110 calculates the state distribution of virtual avatar groups 111, 112, and 113 at the time of collection of results so as to match or approximate the states of resources (EVs 171 to 174) collected by result collection unit 150. , the subsequent operational status and SoC status prediction calculations may be performed.

Next, the processing of the control power trial calculation unit 130 will be described with reference to the flowchart of FIG. 3 showing an example of the control power trial calculation process of the first embodiment.

The controllability trial calculation unit 130 calculates the possible amount of controllability supply by the virtual avatar groups 111, 112, and 113, starting from the time at which the controllability transaction is assumed.

In the present embodiment, first, in step S11, the control force trial calculation unit 130 calculates the maximum amount of control force that can be supplied for each discretized supply start time (first step). Next, in step S12, the controllability trial calculation unit 130 calculates the percentage of the maximum possible amount at the time at which controllable power supply is assumed to start, and how long it continues after the controllable power supply start time. Evaluate availability (second step).

The maximum amount of controllable power that can be supplied calculated in the first step is the maximum possible amount of controllable power supply under forecasted operating conditions in which controllability is not supplied. can provide less adjustability over time. Therefore, in a second step, the effect of a decrease in available controllability over time is evaluated when controllability is actually supplied.

In this embodiment, since a probabilistic model such as a state transition probability model is used, both in the first step (calculation of the maximum possible amount) and in the second step (calculation of possible amount prediction assuming actual control) , multiple simulation evaluations are performed to create data for multiple cases.

Then, in step S13, the controllability trial calculation unit 130 calculates the maximum possible amount for each supply start time and a constant controllability after each supply start time based on the data of the plurality of cases obtained as described above. Statistically calculate a combination of supplyable period and controllability with a certain probability or higher.

Next, with reference to FIG. 4, a method of predicting the maximum possible supply of controllable power under operation prediction without supply of controllable power (first step described above) will be described. FIG. 4 is an explanatory diagram of an example of adjustment force calculation (first step) of the maximum supplyable amount according to the first embodiment.

FIG. 4 shows an image of the SoC state 210 and the operational state 220 of each virtual resource obtained by the state transition probability model and simulation. A time T(Ir) 221 at which the operational state 220 becomes R for the first time in a predetermined outlook period 211 is detected. Then, before the time T (Ir) 221, the operating state 220 finally becomes Po during stop (plug-out) at the time T (Ipo) 222, and after the time T (Ir) 221, the operating state 220 changes to Time T(Ic) 223 at which charging C is reached for the first time is detected.

Using these time T (Ir) 221, time T (Ipo) 222, and time T (Ic) 223, the raising margin 214 is calculated as in the following formula (1), and as in the following formula (2) Calculate the raising adjustment power [kW]. Here, “P+” in the following formula (2) is the charge power that makes the increase margin zero in the period from the supply start time T (Io) to the time T (Ipo) when the SoC reaches the peak of the upper limit. It is calculated by the following formula (3).

Also, the lowering margin 215 is calculated as in the following formula (4), and the lowering adjusting power [kW] is calculated as in the following formula (5). Here, "P-" in the following formula (5) is such that the lowering margin is zero in the period from the supply start time T (Io) to the time T (Ic-1) at which the SoC reaches the lower limit peak. It represents the discharge power and is calculated by the following formula (3).

In addition, in the supply of control power, as shown in the above formula (2) and the above formula (5), in addition to the control power [kW] from the viewpoint of operation, connection is made at the control power supply start time T (Io) The installed capacity of the installed charging facility and the battery performance of the vehicle are also related.

In the above formulas (2) and (5), the vehicle C value and the battery capacity are the upper limits in terms of vehicle battery performance. may be used as the charge/discharge facility capacity and the operation limit of the vehicle battery.

In any case, the possible amount is estimated based on charging/discharging with the lowest power in terms of the upper operating limit 212 and the above-mentioned charge/discharge facility capacity and vehicle battery operation limit. In addition to the elements represented by the above formulas (2) and (5), a scenario may be adopted in which the minimum value is evaluated including charging and discharging at a predetermined upper limit power value.

In addition, there is also a means of stopping charging with regard to the supply of the downward adjustment force. Regarding this, only those in which the state at the control power supply start time T (Io) is in the charging state are targeted, but the reduction margin is the lowest SoC among the states immediately before the start of charging within the current and forecast period The difference in the SoC operating lower limit 213 can be calculated as in the following formula (7). It is possible to know when to stop charging by calculating where the current charge should be stopped so that the lowering margin calculated by the following formula (7) becomes 0.

The above formula (7) is the difference between the SoC and the lower operating limit when the SoC is the minimum before the last charging time T (Ic) during the forecast period from the current time Io. indicates that it is required as

Also, the index k of the time to stop charging can be calculated as in the following equation (8).

ΣΔSoC(t(i)) in the above equation (8) is the total amount of change in SoC from the current time when ΔSoC(t(i)) is integrated from the time index Io (current). Further, argmin(k)ΣΔSoC(t(i)) in the above formula (8) is in a range where the total amount of change ΣΔSoC(t(i)) of this SoC is smaller than the lowering margin 215 shown in FIG. Denote to find the maximum time index k. Also, ∧∀i, State(i)=(C or Pi) in the above equation (8) indicates that the state during the integration period of ΔSoC(t(i)) is C (charging) or Pi (plug-in (non-charging) ) inside). The above formula (8) is for calculating by when the simulated charging should be stopped without any problem if the control power is not provided.

Using the time index k obtained in this way, the possible reduction amount [kWh] (charging stop) is obtained as in the following formula (9), and the reduction control power [kW] is obtained as in the following formula (10). (charging stop) can be calculated.

Equation (9) above indicates the amount of power that should have been charged during the period when charging was stopped. Further, the above formula (10) is a value obtained by dividing the amount of electric power ΔSoC (Io) charged from the current time (target time, time index Io) to the next time by the period ΔT, that is, the value at the target time It calculates the charging power (W). The maximum value of controllability that can be provided by stopping charging during the period from time T(Io) to T(Io+1) is the value obtained by dividing the value of formula (9) above and the value of formula (10) above by ΔT. be the smaller of

FIG. 5 shows an example of the result of obtaining the adjustment force by the above calculation. FIG. 5 is a diagram showing an example of the result of the adjustment force calculation (first step) of the maximum suppliable amount according to the first embodiment. Fig. 5 shows the maximum deliverable amount of megawatt controllability at each time. As an example, the average of the results of 20 controllability evaluations for a group of virtual avatars on the assumption that there are 500 EVs with a capacity of 27 kWh. value. In FIG. 5, the state transition is modeled with ΔT=5 minutes.

In the calculation of the supplyable amount of controllability by the state transition simulation of the virtual resource shown in FIG. .

Next, the case where the virtual resource provides the actual resilience is simulated, and the resilience that can be supplied is evaluated (the second step described above).

In the present embodiment, based on the possible amount obtained by the calculation in which the control power is not supplied as shown in FIG. The time is provisionally set, and evaluation is made on the case where the adjustment power supply that satisfies these conditions is performed.

FIG. 6 is a diagram showing an example of the result of adjustment force calculation (second step) according to the first embodiment. FIG. 6 shows the results of simulating the state of control power supply on the premise that control power of 100 kW is supplied to the virtual avatar group used for the calculation of FIG. 6 from 16:00 to 19:00. . Targeting the resources being charged at each time, calculate the amount of negawatts supplied due to suspension of charging, and calculate the shortage from the resources being charged and plugged in (not charging), the performance of the vehicle, and the charging to which the vehicle is connected. A case was evaluated assuming control to discharge at the maximum power based on the installed capacity.

In addition, in the calculation of the maximum control power (first step) described using the above formulas (1) to (10), the operation Trial calculations are made within a range that does not deviate from the upper limit 212 and the operating lower limit 213 . However, in actual control, future operations are not always predictable. For this reason, in the control reserve prediction calculation (second step) assuming control reserve supply similar to the actual one, the past history and current state of each virtual resource are used, but the information about the future is not used.

In the present embodiment, among the resources that can supply power to the grid at each time, priority is given to stopping the charging of the resources that are being charged, and if the power that can be delivered is insufficient, the resources that have the greater controllability are sequentially selected. Evaluation is performed assuming control that discharges at maximum power.

In consideration of delays in control commands, etc., the virtual avatar group may be divided into a plurality of resource groups, and commands may be given alternately to these resource groups. For example, a resource group with high coordinating power and a resource group with low coordinating power may be alternately selected. Alternatively, the resources may be divided into several groups in advance, and control may be performed such that the groups are sequentially instructed. Also, based on the operation status of each virtual resource, if the SoC at the start of charging is set to 1 during charging, 1. A method of commanding within a range of x (x is, for example, 2, etc.), or a method of commanding discharge in descending order of SoC at the start of the time period. .

Instead of the simple state transition probability model in which the state of the next time is determined only by the state of the previous time shown in Fig. 2, the state of the next time determines the transition probability to the state of the next time. In the case of such a model, based on the series of past states of the virtual resources, it is also possible to select resources from those with a low possibility that the supply of controllability will affect the future.

In the example of FIG. 6, the number of resources used increases with the passage of time, and finally the divergence widens in the direction of less than the requested amount (approximately between 18:00 and 19:00). FIG. 6 shows that the target value of the control power delivery is 100 kW x 3 hours, whereas the virtual avatar group in this evaluation case does not have enough control power to deliver the target value. It represents that.

In such a case, the target value is reduced, and for a predetermined period (in this case, between 16:00 and 19:00), the requested amount and the requested period that can continue to supply the target adjustment force within a predetermined error range , to find a set of request start times.

Next, similar evaluations are performed multiple times for the set of requested amount, requested period, and requested start time. Then, from a statistical analysis of those results, a set of the requested amount, the requested period, and the requested start time that can realize supply within a predetermined error range with a predetermined probability is derived.

In addition, assuming a longer control power supply period in advance, multiple patterns are set for the target value of control power supply and evaluated, and the time until the control power supply deviates from the target value is determined. A method of setting the continuation possible time may be used.

In the evaluation assuming actual controllability supply as shown in FIG. 6, although not shown, there are virtual resources in which the SoC falls below the lower limit of operation or exceeds the upper limit of operation at a time after the start of supply of controllability. The probability of occurrence may also be evaluated.

Such an evaluation is repeated multiple times for each set of provisionally set control capacity request magnitude and period, and from the obtained results, the supply amount and its continuation that can realize the provisionally set request amount and period with a predetermined probability or more. Based on time, trading of adjustment power is performed.

The start time of the provisionally set control force request is determined based on the record of the control force request being invoked, and the magnitude of the control force is determined based on the above-mentioned maximum control force supply possible amount. %, 20%, 30%, and so on.

Next, the adjustment force control command section 140 that issues commands to real resources will be described.

The control power control command unit 140 predicts control power for the resources that can contribute to the control power supply among the information such as the operational status described above and the actual resources (EV171 to EV174) at the time of the control power request. In order to satisfy the supply amount assumed at the stage, multiple real resources are selected, the control power request is distributed, and the period corresponding to one time step of the prediction calculation using the virtual avatar group is stopped, discharged, or Command charging. Regarding discharge, commands are issued in accordance with the control method assumed at the time of predictive calculation of control power using the virtual avatar group.

Such directives are updated at each time step of the prediction calculation. In the case of a model that predicts the state of the next time slice from the state of one time slice as shown in Figure 2, the virtual resource simulates the consistent behavior of individual EVs. Therefore, there is no choice but to determine only the command value in the relevant time section based only on the state in the relevant time section.

In such a case, a method of changing the allocation amount of commands according to the SoC level may be adopted. That is, the maximum controllable power supply capacity calculated in the first step and the trial calculation of the controllable power calculated in the second step are evaluated for each SoC level of the virtual resource when the command is issued. By doing so, the SoC level of the virtual resource can be reduced in the form of reducing the command to the virtual resource of the SoC level where the reserve supply capacity of the control power runs out early and the SoC level where the fluctuation of the control power supply is large. It is also possible to take a method of giving weight to each adjustment force distribution.

On the premise of such control, when activating the controllability obtained by performing the evaluation multiple times as shown in FIG. When performing the step), it may be performed in the same way.

Considering the difference in behavior between the behavior simulating unit 110 using the virtual avatar groups 111, 112, and 113 and the behavior of the real resources (EVs 171 to 174), a reserve is made to issue a larger command than the required amount of adjustment power. It is also possible to set a rate and allocate more by that amount.

Next, the model modification unit 160 that modifies the model will be described.

The performance collection unit 150 collects the resource status after the control power is activated and the resource status for which the control power command has not been issued. The model correction unit 160 corrects the avatar group parameters 122, 123, 124, divides the virtual avatar groups 111, 112, 113, divides the virtual avatar groups 111, 112, 113 and the actual resources (EV171-174) are corrected, such as the avatar group meta-parameters 125.

Based on the difference between the operation state of the real resource group after the control power is activated and the operation state of the virtual avatar group after the control power is activated, the operation state of the virtual avatar group after the control power is activated is the real resource after the control power is activated. The avatar group parameters 122, 123, 124 are adjusted so as to approximate the operational state of the group.

In addition, when adjusting the avatar group parameters 122, 123, and 124 based on the difference between the operational state of the real resource group during the control power non-activation period and the operational state of the virtual avatar group during the control power non-activation period, a plurality of Compare the operating status results of the real resource group and the distribution of the operating status of the corresponding virtual avatar group obtained over a period of 10 days, and try to make the operating status distribution of the virtual avatar group closer to the operating status of the real resource group. , the state transition probability of each time slice is corrected.

Ratios (Pc(s, t), Pr(s, t), Ps(s, t), Pp( s, t)) and the proportion of the corresponding operational state of the virtual avatar group for each SoC state, and ΔOP (s, t) = (ΔPc (s, t), ΔPr (s, t), ΔPs ( s, t) and ΔPp(t)) are calculated for a predetermined period, and the time change rate of the error is calculated based on the following equation (11).

In the above equation (11), ΔPc(s, t) represents the difference between the charging ratio in the virtual avatar group and the charging ratio in the real resource group when the SoC is at level s at time t. ΔPr(s, t) represents the difference between the running ratio in the virtual avatar group and the running ratio in the real resource group when the SoC is at level s at time t. Also, ΔPs(s, t) represents the difference between the ratio of stopped virtual avatars and the ratio of stopped in the real resource group at time t and when the SoC is at level s. ΔPp(s, t) is the difference between the plug-in (non-charging) ratio in the virtual avatar group and the plug-in (non-charging) ratio in the real resource group when the SoC is at level s at time t. represents

The transition probability PT(s, t), which is one of the avatar group parameters of the corresponding virtual avatar group, is corrected so that the error in the denominator on the left side of Equation (11) changes little over time. Here, PT (s, t) is the transition probability PTc (s, t) in the charging state (see formula (12) below), the transition probability PTr (s, t) in the running state (see formula (13) below) ), the transition probability PTs (s, t) in the stopped state (see formula (14) below), and the transition probability PTp (s, t) in the plug-in (non-charging) state (see formula (15) below). Become.

The second subscript p on the right side of each of the above equations (12) to (13) represents a transition in the direction in which the SoC increases, the subscript h represents the status quo of the SoC, and the subscript m represents the SoC. represents a transition in the decreasing direction.

For example, on the right side of the above equation (12), PTcp(s, t) is the SoC at time t at the s level and the operating state is charging (the first subscript c), and at time t+1, the operating The state represents the transition probability that the charging continues and the SoC becomes one level higher.

PTch(s, t) is the probability that when the SoC at time t is at level s and the operating state is charging, the operating state continues to be charging at time t+1 and the SoC remains at level s. show.

PTcm(s, t) is the probability that the operating state transitions from charging to plug-in (not charging) at time t+1 when the SoC is at level s at time t and the operating state is charging. show.

Similarly, on the right side of the above equation (13), PTrp(s, t) is at time t+1 when the SoC is at s level at time t and the operational state is running (first subscript r), Represents the probability that the operating state will transition from running to stopped.

PTrh(s, t) is the probability that when the SoC at time t is at level s and the operating state is running, the operating state continues to be running at time t+1 and the SoC level remains at level s. represents

PTrm(s, t) is the transition probability that, when the SoC at time t is at level s and the operational state is running, the operational state continues to be running at time t+1 and the SoC goes to a level one step lower. represents

Similarly, on the right side of the above equation (14), PTsp(s, t) is, at time t, when the SoC is at s level and the operation state is stopped (first subscript s), at time t+1, Represents the probability that the operating state will transition from being stopped to being plugged in (not charging).

PTsh(s, t) represents the probability that, when the SoC at time t is at level s and the operating state is stopped, the operating state continues to be stopped at time t+1.

PTsm(s, t) represents the probability that the operating state transitions from stopped to running at time t+1 when the SoC is level s at time t and the operating state is stopped.

Similarly, on the right side of the above equation (15), PTpp(s, t) is when the SoC at time t is at s level and the operating state is plug-in (non-charging) (first subscript p). , represents the probability that the operating state transitions from plug-in (non-charging) to charging at time t+1.

PTph(s, t) is the probability that the operating state continues to be plugged in (non-charging) at time t+1 when the SoC is at level s at time t and the operating state is plugging in (non-charging). represents

PTpm(s, t) indicates that when the SoC at time t is at level s and the operating state is plugged in (not charging), the operating state changes from plugging in (not charging) to stopping at time t+1. Represents the probability of transition.

The time change of the error shown on the right side of the above formula (11) and the transition probability can be expressed using the symbols on the right side of the above formula (12), and the following formulas (16) to (19) There is a relationship like

The above equation (16) expresses the difference between the real resource and the virtual avatar in the ratio of charging operation, in this embodiment, the ratio of the charging resource in the virtual avatar Pc(s−1, t), Pc(s, t) ), and the ratio Pp(s, t) of the plug-in state (not charging), and the sum of products of transition probabilities from those states.

Since the error between the real resource and the virtual avatar is considered to be caused by the error in this transition probability, for example, in the case of the above equation (16), the error on the left side is Pc (s−1, t), Pc PTcp(s-1,t), PTch(s,t) up to that time by multiplying the values prorated by the ratio of (s,t) and Pp(s,t) by a factor sufficiently smaller than 1 , PTpp(s, t), the transition probabilities can be corrected.

Note that errors are also accumulated in Pc(s, t), Pr(s, t), Ps(s, t), and Pp(s, t) as time t elapses, so PTc(s, t ), PTr(s, t), PTs(s, t), and PTp(s, t), after recalculating the state transitions, ΔOP(s, t+1)-ΔOP(s , t) may be re-evaluated.

Alternatively, when the transition probability is corrected for the time t when ΔOP(s, t+1)−ΔOP(s, t) exceeds a predetermined value, the correction at the time after the correction is made so that the correction amount becomes zero for a certain period of time. A value multiplied by a coefficient that attenuates in the time direction may be used.

The above is a method that assumes a case where the operation history of individual resources cannot be obtained from a contractual point of view such as protection of personal information.

When considering building or updating a model of transition probability while only the SoC state and operation state in each time section are obtained, unknown parameters such as transition probability to be estimated rather than obtained information becomes more frequent. In such a case, a transition probability model is constructed from detailed operational history obtained based on an exploratory method or some special contract.

When the search method is used, the transition probabilities PTxx [s, t] at each time t and each SoC state s constituting the left sides of the following equations (21) to (24) (where xx=rr-, rr, rs, ss, sp, sr, pp, pc, ps, cc, cc+, cp) are obtained by solving the minimization problem of the following equation (31) under constraint conditions.

Regarding the subscripts, the subscripts used for parameter correction are changed using the above equations (11) to (19). In the above formulas (11) to (19), the subscripts x and y are defined as x = [c (charging), r (running), s (stopping), p (plugging in (not charging))]. , y = [p (transition in the direction of increasing SoC), h (transition in which SoC is maintained), m (transition in the direction of decreasing SoC)]. x).

On the other hand, in the above equations (20) to (31), the second subscript is also the same as the first subscript, "+" for a transition in which the SoC segment increases, and "-" for a transition in which the SoC segment decreases. is added.

Constraints for solving the minimization problem of formula (31) below are based on definitions of transition probabilities such as formulas (21) to (24) below. Alternatively, it may be a constraint condition based on the definition of the state probability as shown in Equation (20) below. Alternatively, it may be a constraint condition based on the transition model (see FIG. 2) represented by the following formulas (25) to (28). Alternatively, it may be a constraint condition based on the definition of the number of divisions of the SoC state as shown in Equation (29) below. Alternatively, it may be a constraint condition representing the finiteness of resources such as storage batteries as shown in the following equation (30).

Note that Po[t], Pi[t], Pc[t], and Pr[t] in the above equation (31) represent the operating state composition ratio at each time t, and P[t, i] at time t It represents the composition ratio of the SoC state s. In addition to the transition probability PTxx[s, t], the variables to be obtained include the values of each SoC state s and each operation state (c, p, s, r) at time 0 Py[s, 0] (where y=c, p, s, r).

Exploratory methods such as genetic algorithms and PSO (Particle Swarm Optimization) can be used. Aiming is not realistic. Therefore, it is preferable to use a method of searching until the error falls below a preset threshold. Note that the coefficients a _s , a _p , a _c , b, etc. in the above formula (31) are parameters that determine the balance in which to evaluate each error element, and are appropriately calculated based on the magnitude of the observation results. should be set.

For example, in the transition model of FIG. 7 showing an example of mutual transition, the probabilities of observing S(t, 1) and S(t, 2) are respectively P(t, 1) and P(t, 2). and At this time, consider transition probabilities q11, q12, q21, q22 where the probabilities of observing S(t+1, 1), S(t+1, 2) are P(t+1, 1), P(t+1, 2). There is a balanced relationship between the transition from S(t,1) to S(t+1,2) and the transition from S(t,2) to S(t+1,1), and it is not uniquely determined. .

Such equilibrium transitions can be obtained exploratoryly, or if there are resources for which detailed operation history information can be obtained based on specific contracts, etc., the transition probability estimated based on the operation history of those resources. Based on the (possible solutions), the equilibrium transition amount mmt represented by the following equation (32) and the net transition amount Δpq represented by the following equation (33) are obtained. Then, assuming the ratio a of the equilibrium transition described above, q11 is replaced with q11′, q12 is replaced with q12′, q21 is replaced with q21′, and q22 to q22'.

When the ratio a of balanced transitions is large, it indicates that the probability of state switching between resources or within the resource itself is high. For this reason, if a method is adopted in which resources that can be used in each time interval are used for control, the resources as a whole are expected to reach a steady state (highly uniform state) quickly. Therefore, when the control power supply is performed, the variance of the control power supply amount predicted by the virtual avatar group is compared with the variance of the control power supply amount in the actual results. The rate of balanced transition is increased according to the amount proportional to the difference in variance or the ratio of variances, and when the variance of performance is large, the rate of balanced transition of the virtual avatar group can be decreased.

If it is possible to collect information from which the operation history of each resource can be obtained for the operation of each actual resource, the transition probability may be obtained directly from these histories. At time t, the SoC level is s level, collect the data how the running resource became at time t+1, calculate the ratio of SoC state, operational state at t+1, and statistics of such values It can be regarded as a transition probability if a typical average value is obtained.

The above Embodiment 1 is not the behavior of individual specific distributed resources, but the energy generated by a large number of resources contained in a virtual distributed resource group that uses a specific application or a specific physical device for negawatts or posiwatts. A group of virtual avatars simulating the consumption situation, a coordination capacity trial calculation unit that estimates the coordination capacity that the group of virtual avatars can provide, assuming coordination capacity requirements for the group of virtual avatars, and similarity between the calculation results and the state of individual resources. For actual distributed resources determined to be similar to a group of virtual avatars or a part of them based on gender, commands to the group of virtual avatars, or commands are modified based on the difference between the state of the group of virtual avatars and the actual resources. A control force control command unit that issues a command, a control force result collection unit that collects results when the control force is activated, and a model of the virtual avatar group or similarity between the virtual avatar group and the actual resource based on the collected results. and a model modification unit that modifies parameters for determining

Therefore, according to the first embodiment, the coordination capacity prediction of the distributed resource group is improved according to the actual situation without information such as the history of the state from the past to the present of each distributed resource and the operation intention for several hours ahead. become able to do.

[Modification of Embodiment 1]
In FIG. 2, one discretized state is represented by a circle symbol, but this is not limiting, and a transition model relating to a state combination combining a plurality of operation states at consecutive times may be constructed. An example of the state transition model in this case is shown in FIG. FIG. 9 is a diagram showing a modification of a model configuration method when EVs are used as control power supply resources in the behavior simulating unit of the first embodiment.

In FIG. 9 , ○ symbols representing states are CC+ (the transition from charging to charging increases SoC), CC (there is no SoC increase at transition from charging to charging), and PiC (from plug-in to charging), PoPi (transition from stopped (plugged out) to plugged in), PiPi (transition from plugged in to plugged in), CPi (transition from charging to plugged in) , RPo (transition from running to stopped), PoPo (transition from stopped to stopped), PiPo (transition from being plugged in to stopped), PoR (transition from being plugged out to running), There are 12 types of RR (no SoC change at transition from running to running) and RR- (SoC decrease at transition from running to running). State transitions between times (t−1, t, t+1) are indicated by arrows. These transition probabilities may be estimated by the exploratory method described above, and corrected based on the results when the control power is exercised and when the control power is not implemented. In this case, the result collection unit 150 collects, as results, a set of the discretized operation state and SoC state one hour ago and the operation state and SoC state at the current time.

In addition, a plurality of discretized time intervals Δt in modeling the state transition are defined, and each time interval is Δt1, Δt2, Δt3 (n Δt1 = Δt2, m Δt1 = Δt3, n, m∈N, 1<n , m, n<m), multiple transition models for the main operational state and the SoC state may be constructed simultaneously.

By using a transition model related to the state transition sequence of continuous time and a transition model with different multi-time resolutions, it is possible to improve the accuracy of the correspondence between the actual resources and the virtual resources that make up the virtual avatar group. , can improve the efficiency of correction.

In order to select the state of the transition model related to the state transition series of consecutive times and the state of the real resource corresponding to the combination of the states of the transition model with different multi-time resolutions, clustering or other methods are used to select the state of the real resource. and the state transition sequence of the virtual avatar group.

On the other hand, when using a transition model related to a state transition series over a plurality of continuous time sections in the past shown in FIG. Similar to the weighting of commands according to SoC level above, for states across a range of states, we evaluate the duration and stability of possible quantities, and for successive past states and combinations of states at multi-time resolution , weighting may be applied to the adjustment force distribution.

In such a case, the behavior simulation unit 110 using the corresponding virtual avatar groups 111, 112, and 113 calculates the From among the results obtained, the selection of the use of real resources, i.e., the allocation, is determined in proportion to the amount of selection of virtual resources in the same state sequence as the real resources or combination of states at multiple time resolutions. good.

In addition, when there are multiple types of state transition probability models that simulate the behavior of a resource group for one resource group, the behavior of the resource group in each state transition probability model is presented, and the real resource selected according to the presentation is presented. It is also possible to simulate the time distribution of the operating state of the distributed resources based on a state transition probability model that exhibits behavior close to , calculate the adjustability, and issue commands to the real resources.

[Embodiment 2]
Embodiment 2 of the present invention will be described with reference to FIG. FIG. 10 is a diagram showing an example of the configuration of a control power management device 800 according to the second embodiment. The coordination power management device 800 uses factories, offices, commercial facilities, etc. having facilities such as storage batteries and private power generators as target distributed resources. Hereinafter, distributed resources in this embodiment are referred to as C&I (Commercial And Industry) resources.

As shown in FIG. 10, the control power management device 800 includes a behavior simulating unit 810, a parameter storage unit 820, a control power trial calculation unit 830, a control power control command unit 840, a result collection unit 850, and a model correction unit 860. be.

In FIG. 10, C&I resources 871, 872, 873, and 874 shown as examples of distributed resources are connected to the control power management device 800 by a communication means (not shown), and are in a state before and during control power activation. is transmitted from the resource side to the coordinating power management device 800, and commands for coordinating power control are transmitted from the coordinating power management device 800 to the C&I resources 871-874. Although FIG. 10 shows four C&I resources 871 to 874, the number is not limited to four.

The behavior simulating unit 810 consists of one or more virtual avatar groups 811 , 812 , 813 . In FIG. 10, there are three virtual avatar groups, but the number is not limited to this.

In the behavior simulation unit 810 in this embodiment using C&I resources, a standard operation pattern model for the operation state of C&I resources is applied to the business type of the C&I resource owner and the operation method of the generator and storage battery used for supply of adjustment power. A plurality of virtual avatar groups are prepared according to the requirements.

In many cases, the specifications of generators and storage batteries used for load adjustment are known, but there are differences in usage such as for BCP and peak cut, and other loads that affect the baseline, solar power generation equipment, etc. There is also the influence of the variable power supply, etc., and the prediction of individual adjustability has the same difficulty as the case of EV described in the first embodiment.

In addition, in the case of a configuration that implements aggregation hierarchically, it is not a lower-level aggregator that analyzes the performance of storage batteries and generators and issues control commands, but rather bundles them and communicates with users of adjustment power. There is a high-level aggregator (aggregation coordinator) who plans, controls, and manages the whole under the contract. In many cases, the upper aggregator does not know the load other than the resource, its operation status, the configuration of the renewable energy power generation facility, etc., and predicts the behavior of the consumer side from the point of view of the power receiving point of the C&I consumer who supplies the adjustment power. becomes more difficult, especially for high-level aggregators.

Based on such characteristics of C&I resources or hierarchical characteristics of aggregation, in the present embodiment, virtual avatar groups 811, 812, and 813 are not only generators and storage batteries, but also facilities operated in combination with them. Model your operations. As mentioned above, the controllability and responsiveness of generators and storage batteries differ depending on the operational purpose of the generators and storage batteries. Since the patterns are different, a plurality of virtual avatar groups are configured to represent these differences.

The virtual avatar group may be configured using the state transition model shown in FIGS. 2 and 9, or may be configured using an autoregressive model based on past performance.

In the first embodiment of the present invention, the transition probability of the state of SoC and operation at each discretized time is determined by an exploratory method, and based on the results when the control power is not activated and the results after the control power is activated showed how to adjust Regarding C & I resources, when using the state transition model, at each discretized time, if the relationship between the power consumption level at the next time at each discretized power consumption level is modeled, the above-described implementation It can be handled in the same way as the EV case of form 1. If you have variable renewable energy power generation facilities, you can treat them in the same way as a model without renewable energy by creating a model for net load that excludes the amount of power generated by them for self-consumption.

In the case of the first embodiment as well, it is shown that the state transition is modeled at each time, but the model may assume that the tendency of the transition changes depending on not only the time but also the day of the week, weather conditions, and the like. By doing so, it is possible to enhance the accuracy of prediction of control power and the effect of correction under specific weather conditions and days of the week. In particular, when there is self-consumption of renewable energy power generation equipment as described above, the net load model is strongly affected by weather conditions, so it is effective to create a model that responds to weather conditions.

If the facility used for load regulation is a generator, the effect of the generator's regulating power can be known almost accurately if the load is known when it is not used as regulating power. Therefore, if a load model such as the one described above, that is, a baseline model can be created, it becomes possible to predict controllability and its stochastic variation.

Response delays to commands and the presence and magnitude of steady-state deviations depend on the performance of resources such as generators and storage batteries. You may make it calculate a control power effect also considering an error element like.

In addition, when there are many C&I resources that use storage batteries, it is particularly useful for peak shaving, arbitrage operation considering price differences in time zones, and improving the self-consumption rate of variable renewable power generation facilities. It is assumed to be susceptible to influences and individual differences among individual C&I resource providers. In the case of storage batteries operated for such purposes, the SoC state at each discretized time is divided into multiple stages, with the upper limit of the total capacity determined based on the contract information such as the amount of control power supplied by the storage battery and the period. It should be discretized, and the control power itself supplied by the load control equipment may be predicted in the same way as the group of virtual avatars in the case of the EV described above. In this case, the correction of the avatar group parameters 822, 823, 824 and the avatar group meta-parameter 825 based on the track record when the control power is exercised can be performed in the same manner as the method described in the first embodiment.

If the behavior of the real resource changes depending on the weather, the day of the week, etc., the correspondence between the virtual avatar group and the real resource may be modified according to the weather. For this purpose, the behavior of each resource is evaluated in advance to determine which virtual avatar group it is closest to, depending on the weather conditions and the day of the week. should be determined. In addition, a means may be provided to disclose the prediction results before supply of adjustment power by the virtual avatar group to the operator of the real resource, and to register which virtual resource the operator himself/herself operates close to.

FIG. 11 is a diagram showing an example of the hardware configuration of the coordinating power management devices 100 and 800. As shown in FIG. A computer 5000 that realizes the coordinating power management devices 100 and 800 includes a processor 5300 represented by a CPU (Central Processing Unit), a memory 5400 such as a RAM (Random Access Memory), and an input device 5600 (for example, a keyboard, mouse, touch panel, etc.). , and an output device 5700 (eg, a video graphics card connected to an external display monitor) are interconnected through memory controller 5500 . In the computer 5000, a program for realizing the coordinating power management devices 100 and 800 is read from a ROM (Read Only Memory) or an external storage device 5800 such as an SSD or HDD via an I/O (Input/Output) controller 5200. , and executed by cooperation of processor 5300 and memory 5400, coordination power management devices 100 and 800 are realized. Alternatively, a program for realizing coordination power management devices 100 and 800 may be acquired from an external computer through communication via network interface 5100 .

In addition, the present invention is not limited to the above-described embodiments, and includes various modifications. For example, the above-described embodiments have been described in detail in order to explain the present invention in an easy-to-understand manner, and are not necessarily limited to those having all the configurations described. Also, as long as there is no contradiction, it is possible to replace part of the configuration of one embodiment with the configuration of another embodiment, and to add the configuration of another embodiment to the configuration of one embodiment. Moreover, it is possible to add, delete, replace, integrate, or distribute a part of the configuration of each embodiment. Also, the configurations and processes shown in the embodiments can be appropriately distributed or integrated based on processing efficiency or implementation efficiency.

100, 800: coordination power management device, 110, 810: behavior simulation unit, 111, 112, 113, 811, 812, 813: virtual avatar group, 130, 830: coordination power trial calculation unit, 140, 840: coordination power control command Section 150, 850: Result Collection Section 160, 860: Model Correction Section

Claims (11)

An adjustability management device for managing the adjustability of distributed resources,
a behavior simulating unit that globally simulates the time distribution of the operating states of the distributed resources;
a controllability estimation unit that estimates controllability that can be supplied by the distributed resource based on the time distribution of the operating state of the distributed resource that is globally simulated by the behavior simulating unit ;
The behavior simulating unit
Using a state transition probability model as a model for simulating the time distribution of the operating state of the distributed resource as a whole, and using the state transition probability model for each of a plurality of groups according to a use case of the distributed resource, the distributed resource calculates the adjustment power that can be supplied for each of the multiple groups
An adjustability management device characterized by: The adjustability trial calculation unit
a first step of calculating the maximum amount of flexibility that the distributed resource can supply at each supply start time under conditions of no supply of flexibility;
Based on the maximum amount at each supply start time calculated in the first step, predict the suppliable amount that evaluates the influence of the decrease in the suppliable controllability over time when the controllability is actually supplied. a second step of
and calculating the period and the controllability in which the controllability can be supplied with a certain probability or more after each supply start time based on the prediction result of the second step . The adjustability management device according to . A control power control command unit that distributes the control power request to each distributed resource and outputs a control power supply command according to the operation state of each distributed resource at the time when the control power request is issued to the distributed resource. The adjustability management device according to claim 1 , further comprising: a result collection unit that collects results of operation statuses of the distributed resources;
a model correction unit that corrects the state transition probability model based on a difference between the operation state results collected by the result collection unit and the operation state of the distributed resource simulated by the state transition probability model; The adjustability management device according to claim 3 , characterized in that: The model correction unit
The state transition is performed based on the results of the operation state of each distributed resource after the control power is activated collected by the result collection unit and the difference between the operation state of each distributed resource after the control power is supplied in the state transition probability model. The adjustability management device according to claim 4 , wherein the state transition probability of the stochastic model is corrected. The model correction unit
The adjustability management device according to claim 5 , wherein the probabilities of mutual transitions between operation states in the state transition probability model are corrected. Having a plurality of types of the state transition probability models,
presenting the behavior of the distributed resource in each of the plurality of types of state transition probability models;
The behavior simulating unit
6. The coordination power management device according to claim 5 , wherein the time distribution of the operation states of the distributed resources is simulated based on the state transition probability model selected according to the presentation. The state transition probability model is
It is characterized by a transition model in which the state of the next time is determined only by the state of the previous time, a transition model related to a combination of multiple operational states at consecutive times, or a transition model with different multiple time resolutions. The adjustability management device according to claim 1 . The coordination power management device according to claim 1, wherein the distributed resource is an electric vehicle. The coordination power management device according to claim 1, wherein the distributed resource is a C&I resource. An adjustability management method executed by an adjustability management device that manages the adjustability of distributed resources,
The adjustability management device
simulating the time distribution of the operating state of the distributed resources globally;
Based on the overall simulated time distribution of the operating state of the distributed resource, trial calculation of the adjustment power that the distributed resource can supply,
Using a state transition probability model as a model for simulating the time distribution of the operating state of the distributed resource as a whole, and using the state transition probability model for each of a plurality of groups according to a use case of the distributed resource, the distributed resource calculates the adjustment power that can be supplied for each of the multiple groups
A control power management method characterized by executing each process.

Images