JP2017046498A - Use estimation system for vehicle - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the use rate of an on-vehicle storage battery in an energy management utilizing the on-vehicle storage battery.SOLUTION: An area 10 has a plurality of buildings 11, and an EMS12 is structured in each of these buildings 11. In the area 10, an area management system 13 is structured so as to include the EMS 12 of each building 11. The area management system 13 has an area controller 16 for estimating use of a vehicle 15. The area controller 16 has: a base-point setting part that sets a plurality of base points for estimating the use of a vehicle 15; a model creation part for creating a Markov model; and a use estimation part for estimating the use state of a vehicle 15 at each of the base points. The use estimation part has a maximum estimation part that estimates the maximum likelihood by using the Markov model.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a vehicle use prediction system that predicts a use state of a vehicle by a user.

  For vehicles such as electric vehicles (EV) and plug-in hybrids (PHV), research and development on V2X (vehicle to X), which uses a storage battery not only for driving but also for supplying electric power to the outside, has been actively conducted. For example, an energy management method using V2X has been proposed in which when a user is at home, a storage battery of a user-owned vehicle is used as a storage battery for adjusting the power supply / demand balance at home. In this energy management method, the vehicle is used as energy storage for storing energy.

  However, in order to use the storage battery of the vehicle with V2X, the vehicle needs to be connected to a power grid such as a house. When the user goes out using the vehicle, the storage battery of the vehicle is set to V2X. Cannot be used in For this reason, in order to perform energy management using the storage battery of the vehicle, it is necessary to grasp the timing of whether or not the storage battery can be used from the departure time and the return time of the user, and to prepare a charge / discharge plan for the storage battery for one day, for example. .

  As a method for grasping the timing of whether or not a storage battery can be used, there is a method in which a user performs an input operation of a vehicle use plan in advance, but this method imposes a burden on the user. Therefore, in order to grasp the timing of whether or not the storage battery can be used, it is conceivable to employ a method for predicting the usage state of the vehicle.

  For example, Patent Document 1 describes a system for predicting the use state of a vehicle. In this system, past departure times and return times are tabulated in frequency distribution, and the simultaneous probability of departure time and return times is maximized. It becomes. Moreover, in this system, the stay period and absence period of one vehicle with respect to one house are estimated.

JP 2014-75861 A

  However, in a configuration in which energy management is performed for one vehicle and one house, there is room for improvement regarding effective use of the in-vehicle storage battery in energy management. For example, when a vehicle is used for a user's commuting, an opportunity to use the vehicle as energy storage in a house is lost in the user's working hours.

  The present invention has been made in view of the above problems, and an object thereof is to increase the utilization rate of the in-vehicle storage battery in the energy management using the in-vehicle storage battery.

  The technical means of the invention for achieving the object will be described below. The reference numerals in parentheses described in the scope of claims and this column disclosing technical means of the invention indicate the correspondence with the specific means described in the embodiment described in detail later. It is not intended to limit the technical scope of the invention.

The invention disclosed in order to solve the above problems
Regarding the base (11, 11a to 11c) where the vehicle can be parked, whether the vehicle is in a parked state where the vehicle is parked or a moving state where the vehicle is not parked, the usage state (γ) of the vehicle by the user A vehicle usage prediction system that predicts by modeling the stochastic process of
A base setting unit (25) for setting a plurality of bases for the vehicle;
For each of a plurality of bases set by the base setting unit, a prediction unit (30) that predicts whether the use state of the vehicle is in a parking state or a movement state;
It is characterized by having.

  According to the present invention, since a plurality of bases are set for the user's vehicle for prediction of the usage state of the vehicle, prediction of the presence or absence of the vehicle can be individually acquired for each of these bases. For this reason, by performing energy management using in-vehicle storage batteries at each of the plurality of bases, the period of not using in-vehicle storage batteries is shortened, and it is possible to appropriately obtain the opportunity to use in-vehicle storage batteries for effective use of energy. it can. In this case, if a plurality of bases are included in the same area, it is possible to suppress the peak of power consumption in the area. As described above, the utilization rate of the in-vehicle storage battery can be increased in the energy management using the in-vehicle storage battery.

The figure which shows the whole structure of the area management system in 1st Embodiment. The figure which shows the movement aspect of the vehicle in an area. The figure for demonstrating the state transition of a vehicle. The block diagram which shows the function of a regional controller. The flowchart which shows the procedure of a use prediction process. The figure which shows a Markov model. The figure which shows frequency distribution for arrival. The figure which shows the frequency distribution for departures when a vehicle belongs to a home. The figure which shows the frequency distribution for departures when a vehicle belongs to a work place. The figure which shows the frequency distribution which digitized the occurrence frequency of the event. The figure which shows the relationship between the occurrence frequency of an event, and a movement period. The figure which shows an affiliation model, a use model, and a continuation model. The figure which shows the result of real-time prediction. The figure for demonstrating the state transition of the vehicle in 2nd Embodiment. The figure which shows a Markov model. The figure which shows an affiliation model, a use model, and a continuation model.

  Hereinafter, a plurality of embodiments of the present invention will be described with reference to the drawings. In addition, the overlapping description may be abbreviate | omitted by attaching | subjecting the same code | symbol to the corresponding component in each embodiment. When only a part of the configuration is described in each embodiment, the configuration of the other embodiment described above can be applied to the other part of the configuration. In addition, not only combinations of configurations explicitly described in the description of each embodiment, but also the configurations of a plurality of embodiments can be partially combined even if they are not explicitly specified unless there is a problem with the combination. .

(First embodiment)
As shown in FIG. 1, the area 10 has a plurality of buildings 11 such as houses, buildings, factories, stores, and public facilities. The area 10 is composed of a municipality such as a municipality, a district owned by the municipality, a school district, or the like. In this area 10, commercial power is supplied to each building 11 from a transmission / distribution network included in one power system.

  Each building 11 is provided with an EMS 12 (Energy Management System) as an energy management system for performing energy management, and a regional management system 13 is constructed including these EMSs 12. The regional management system 13 is a system that performs energy management of the regional 10 and can also be referred to as a regional energy management system or a CEMS (Community Energy Management System). The area management system 13 can predict the usage state of the vehicle 15 used corresponding to the building 11 and manage the EMS 12 of each building 11 based on the prediction result. In addition, energy management of the area 10 using the area management system 13 is performed by an operator such as an aggregator. The area management system 13 corresponds to a vehicle use prediction system.

  The regional management system 13 has a regional controller 16 as a control device. The regional controller 16 includes a control unit such as a microcomputer, and the control unit includes a storage unit such as a memory. In the regional management system 13, the regional controller 16 and the EMS 12 can communicate bidirectionally with a communication device or the like.

  The building 11 is provided with electric equipment for constructing a power network in the building 11. The vehicle 15 is an electric vehicle (EV) that runs only by an electric motor, or a plug-in hybrid (PHV) that uses a motor and an internal combustion engine as a driving force, and has an in-vehicle storage battery. In a state where the vehicle 15 is parked in the parking lot of the building 11 and the electrical equipment of the building 11 and the on-vehicle storage battery are electrically connected by a dedicated cable, power can be exchanged between the building 11 and the vehicle 15. ing. The EMS 12 includes the implementation of V2X (vehicle to X), which is power supply from the vehicle 15 to the building 11, and performs energy management between one vehicle 15 and one building 11. In this case, the EMS 12 uses the vehicle 15 as an energy storage capable of storing electric power, and corresponds to a management unit that performs energy management for the building 11.

  When the vehicle 15 moves between the plurality of buildings 11, energy management by the EMS 12 is performed in each of the buildings 11 during the parking period of the vehicle 15. In this case, the vehicle 15 plays a role of transporting the electric power stored in the in-vehicle storage battery to the plurality of buildings 11.

  As shown in FIG. 2, in a region 10, a user's home 11 a and the user's work place 11 b are included in a plurality of buildings 11, and the user commutes using a vehicle 15 a owned by the user. doing. The vehicle 15a is parked in the home 11a before going to work, after returning home, or on a holiday. On the other hand, the vehicle 15a is parked at the work location 11b during the period in which the user is at the work location 11b after going to work, and the vehicle 15a is parked at either the home 11a or the work location 11b during work or return home. Not. In addition, in the home 11a and the work place 11b, the vehicle 15 is parked in the parking lot in each site or off site.

  In the regional management system 13, the vehicle 15 belongs to one of a plurality of bases, and the destination is switched as the vehicle 15 moves. As shown in FIG. 3, for example, when the vehicle 15a is parked at the home 11a and when the destination is the home 11a when the vehicle 15a is moved, the vehicle 15a belongs to the home 11a. ing. Further, when the vehicle 15a is parked at the work place 11b and when the destination is the work place 11b when the vehicle 15a moves, the vehicle 15a belongs to the work place 11b.

  The use state of the vehicle 15 is changed by parking or moving. In this case, the use state of the vehicle 15 is changed from one to the other among the parking state and the moving state. If the movement of the vehicle 15 is accompanied, the state transition of the vehicle 15 occurs not only when the affiliation of the vehicle 15 changes but also when the affiliation of the vehicle 15 does not change. Here, when the state transition of the vehicle 15 occurs without changing the affiliation destination, this state transition corresponds to the first transition, and when the state transition of the vehicle 15 occurs with the change of the affiliation destination, this state transition Corresponds to the second transition. In addition, the transition from the parking state to the moving state occurs regardless of whether the affiliation destination of the vehicle 15 is changed or not, while the transition from the moving state to the parking state is a case where the affiliation destination of the vehicle 15 is not changed. Only occurs.

  As shown in FIG. 3, for example, when the vehicle 15a departs from the home 11a and returns to the home 11a again without going to the work place 11b, the vehicle 15a is a first transition in which the affiliation destination is held at the home 11a. Transitions A1 and A2 are performed. On the other hand, when the vehicle 15a departs from the work place 11b and returns to the work place 11b without going to the home 11a, the vehicle 15a uses the transitions B1 and B2 as the first transition in which the affiliation destination is held at the work place 11b. Will do. Transitions A1 and B1 are movement transitions A1 and B1 indicating a state transition from the parking state to the movement state, and transitions A2 and B2 are parking transitions A2 and B2 indicating a state transition from the movement state to the parking state.

  In addition, when the vehicle 15a moves from the work place 11b to the home 11a, the vehicle 15a changes its affiliation from the work place 11b to the home 11a at the timing of leaving the work place 11b, and from the parked state to the moved state. Home transition A3 will be performed as two transitions. On the other hand, when the vehicle 15a moves from the home 11a to the work location 11b, the vehicle 15a changes from the home state 11a to the work location 11b at the timing of leaving the home 11a and the second state from the parked state to the moved state. Work location transition B3 will be performed as a transition.

  A state in which the vehicle 15a is parked at a base such as the home 11a or the work place 11b is referred to as a parking state. In addition, the state in which the vehicle 15a is traveling is referred to as a moving state, and the state in which the vehicle 15a is parked in a place other than the base is also referred to as a moving state.

  Next, functions of the regional controller 16 will be described with reference to the functional block diagram of FIG.

  As illustrated in FIG. 4, the regional controller 16 includes acquisition units 21 to 24 that acquire various types of information, a usage prediction unit 30 that predicts the usage state of the vehicle 15 based on the various types of information, and prediction results of the usage prediction unit 30. Is output to the outside. The EMS 12 of each building 11 is electrically connected to the output side of the regional output unit 50, and the regional output unit 50 corresponds to a prediction output unit that outputs a prediction result to the EMS 12. Various sensors are connected to the regional controller 16, and the acquisition units 21 to 24 can acquire various information based on detection signals of the various sensors. Various sensors include those directly connected to the regional controller 16 and those indirectly connected via the EMS 12 of the building 11.

  As the acquisition units 21 to 23, a time acquisition unit 21 that acquires the current time, a current state acquisition unit 22 that acquires the current use state of the vehicle 15, and a latest acquisition unit 23 that acquires the nearest parking place of the vehicle 15 Is connected to the input side of the usage predicting unit 30. The current state acquisition unit 22 acquires that the vehicle 15 is in a moving state and in a parking state. Moreover, when the vehicle 15 is in a parking state, the parking location is acquired.

  When the vehicle 15 is in a moving state, the latest acquisition unit 23 acquires a base parked before the vehicle 15 starts the current movement as the nearest parking place. Moreover, the latest acquisition part 23 acquires a continuation period about the present movement based on the time which departed from the latest parking place.

  The acquisition unit 24 is a distribution acquisition unit 24 that acquires a frequency distribution table showing the usage history of the vehicle 15. A base setting unit 25 that sets a plurality of bases for the vehicle 15 on the input side of the distribution acquisition unit 24. Is connected. The distribution acquisition unit 24 acquires a frequency distribution table for a plurality of bases set by the base setting unit 25. The site setting unit 25 sets any of the plurality of buildings 11 as a site by a user input operation.

  The frequency distribution table acquired by the distribution acquisition unit 24 includes data indicating the past use state of the vehicle 15 as a parking period, a moving period, a destination, a departure time for the destination, Information such as arrival time is included in a discretized state. The event such as parking or movement is information indicating the usage state of the vehicle 15, and the frequency distribution table records the duration of the event and the number of occurrences of the event.

  The distribution acquisition unit 24 is connected to the input side of the use prediction unit 30 via a model creation unit 26 that creates a probability model (quasi-Markov model) regarding the state transition of the vehicle 15. The model creation unit 26 creates a Markov model 35 (see FIG. 6) as a probability model by developing a quasi-Markov process as a probability process based on the frequency distribution table acquired by the distribution acquisition unit 24. In this case, the model creation unit 26 models the use prediction of the vehicle 15 with respect to both the event occurrence order and the event occurrence time. In the quasi-Markov process, the state transition probability varies with time. The Markov model 35 corresponds to a prediction model for predicting the usage state of the vehicle 15.

  Incidentally, both the information on the base set by the base setting unit 25 and the Markov model 35 created by the model creation unit 26 are input to the use prediction unit 30.

  The usage prediction unit 30 performs prediction for a predetermined prediction period. The prediction result of the usage prediction unit 30 is that the prediction period is discretized into T steps, and the discretization step width Δt in this case is an occurrence interval between events. The step width Δt is a value obtained by dividing the prediction period by T. For example, if the prediction period is 24 hours and the step width Δt is 30 minutes, T = 48.

  The usage prediction unit 30 estimates the maximum likelihood solution based on the transition calculation unit 31 that calculates the probability that the usage state of the vehicle 15 transitions as the state transition probability and the Markov model 35 created by the model creation unit 26. And an estimation unit 32.

  The regional controller 16 performs usage prediction processing for performing usage prediction of the vehicle 15. Here, the usage prediction process will be described with reference to FIGS.

  In FIG. 5, in step S11, offline processing is performed. In the offline processing, the base setting unit 25 sets the base, the distribution acquisition unit 24 acquires the frequency distribution table, and the model creation unit 26 creates the Markov model 35. The offline processing is processing that is performed at the time of initial setting or maintenance.

  Here, as shown in FIG. 6, the Markov model 35 includes model units 36 and 37 for predicting the duration of the predetermined state when the vehicle 15 is in the predetermined state. The model unit 36 is a parking model unit 36 for predicting a duration in which the vehicle 15 is in a parking state, and the model unit 37 is a movement model unit 37 for predicting a duration in which the vehicle 15 is in a moving state. is there. In the model units 36 and 37, the horizontal axis is time t, t = τ is the step of the current time, and t = τ + T is the future step after the prediction period has elapsed from the current time. In this embodiment, the future step of τ + T is 24 hours ahead. The vertical axis is the duration u, u = Δt is a step indicating the end of the duration, and u = T is a step indicating that the duration is the same as the prediction duration. In this case, the prediction dot 38 that predicts the usage state of the vehicle 15 has two elements, time and duration.

  The model units 36 and 37 are individually provided for each of a plurality of bases in the Markov model 35. In the present embodiment, model units 36a and 37a for home use and model units 36b and 37b for work place are provided as the model units 36 and 37, respectively.

  Returning to FIG. 5, after the offline process is completed, the process proceeds to step S <b> 12 to perform online process. In the online processing, acquisition of various types of information by the acquisition units 21 to 23, processing by the transition calculation unit 31, and processing by the maximum likelihood estimation unit 32 are performed in this order.

  Here, the process by the transition calculation part 31 is demonstrated. The transition calculation unit 31 calculates a frequency distribution related to the use of the vehicle 15 based on the frequency distribution table acquired by the distribution acquisition unit 24, and calculates an initial state probability and a state transition probability based on the frequency distribution. As the frequency distribution, the frequency distribution for arrival indicating the relationship between the arrival time of the vehicle 15 at the base and the parking period at the base, and the relationship between the departure time of the vehicle 15 from the base and the movement period after the departure. The starting frequency distribution shown is mentioned.

  In this embodiment, the base of the vehicle 15a is set to the home 11a and the work place 11b, and the frequency distribution for arrival as illustrated in FIG. 7 is obtained for each of the home 11a and the work place 11b. In this frequency distribution, the horizontal axis is the arrival time at the home 11a, and the vertical axis is the parking period in which the vehicle 15a is continuously parked after arrival.

  Regarding the frequency distribution for departure, there are a case where the destination of the vehicle 15 is one of the two bases and a case where the destination is the other base. That is, there are a case where the vehicle 15 belongs to one base and a case where the vehicle 15 belongs to the other base. For example, when the vehicle 15a belongs to the home 11a, as shown in FIG. 8, when the vehicle 15a leaves the home 11a and returns to the home 11a, the vehicle 15a moves from the work place 11b to the home 11a. May move. As shown in FIG. 9, when the vehicle 15a belongs to the work place 11b, the vehicle 15a moves from the home 11a to the work place 11b, and the vehicle 15a leaves the work place 11b again. There is a case of returning to the work place 11b.

  In each frequency distribution shown in FIGS. 8 and 9, the horizontal axis is the departure time, and the vertical axis is the movement period until the arrival at the destination. This travel period is the time required to arrive at the destination after departure. For this reason, not only the period during which the vehicle 15 actually travels, but also the period during which the vehicle 15 is parked at a place different from the destination, for example, is included in the movement period.

  In addition, each frequency distribution shown in FIGS. 7-9 samples the usage log | history of the vehicle 15 by making a user into a test subject. For example, as information on the subject, age, sex, main purpose of use of the vehicle 15, and the like are stored in the storage unit of the regional controller 16. The purpose of using the vehicle 15 includes commuting and shopping.

  In the frequency distribution, the frequency of occurrence of an event such as parking or movement is expressed by the shading of the dot display indicating the sampling result, and the initial state probability and the state transition probability are calculated using this frequency. A diagram in which the frequency is quantified is shown in FIG. For the case where the vehicle 15a moves from the home 11a to the work place 11b, the state of this travel time is obtained by dividing the occurrence frequency of a predetermined travel time (for example, 50 minutes) by the sum of the occurrence frequencies of all samplings. A transition probability is calculated. For example, when the movement occurrence frequency when moving from the home 11a to the work place 11b is displayed in two dimensions with the movement period, a distribution diagram as shown in FIG. 11 is obtained.

  Similarly to the state transition probability regarding the vehicle movement of the home 11a → the work place 11b, the state transition probability regarding the parking of the home 11a and the work place 11b, and the home 11a → the home 11a, the work place 11b → the work place 11b, the work place 11b → the home A state transition probability relating to the movement of 11a is calculated. The initial state probability is also calculated in the same manner as the state transition probability according to the state of the vehicle 15 at the current time.

  The initial state probability and the state transition probability are individually assigned to each predicted dot 38 of the Markov model 35, and the transition calculation unit 31 performs initial processing for each predicted dot 38 in online processing. Update probabilities and state transition probabilities. In this case, the initial state probability and the state transition probability change with time. In the model portions 36 and 37, the decrease in the duration with the passage of time is indicated by an arrow extending diagonally downward to the right from the prediction dot 38.

  Next, the process by the maximum likelihood estimation part 32 is demonstrated. The maximum likelihood estimation unit 32 uses the prediction problem regarding the usage state of the vehicle 15 as a “combination optimization problem” and estimates a maximum likelihood solution for the “combination optimization problem” using “dynamic programming”. In this case, the maximum likelihood estimator 32 divides the prediction problem of the vehicle 15 into a plurality of partial problems on the time axis in the Markov model 35, and uses the calculation results of these partial problems in a forward algorithm, thereby generating a plurality of movement patterns. The occurrence probability is calculated for each of. And based on each occurrence probability of these movement patterns, the maximum likelihood solution about the prediction problem of the vehicle 15 is estimated.

  The maximum likelihood estimation unit 32 estimates the occurrence probability J of the movement pattern based on an estimation expression such as the following expression (1), using information regarding the current time and information regarding the future. Information on the current time includes the current use state γτ of the vehicle 15, the start time t0 of the state γτ, and the current affiliation v0 of the vehicle 15. As information about the future, future times t1, t2,... Tk and future affiliations v1, v2,.

J = P (t1, t2... Tk, v1, v2... Vk | t0, γτ, v0) (1)
Regarding the use prediction of the vehicle 15a for the home 11a and the work place 11b, the case where the maximum likelihood estimation unit 32 estimates the maximum likelihood solution X1 will be described with reference to FIG. The maximum likelihood solution X1 is that the vehicle 15a is parked at the home 11a after a certain period of time has elapsed (for example, one hour later) after the vehicle 15a starts moving from the work place 11b to the home 11a. The prediction is shown.

  As shown in FIG. 6, the maximum likelihood solution X1 has a home transition A3 accompanied by a movement from the work place 11b to the home 11a and a parking transition A2 with the affiliation destination being the home 11a. The home transition A3 starts at a future time of time t = τ + Δt in the home movement model unit 37a from a dot in the duration u = Δt at the current time of time t = τ in the parking model unit 36b for work place. A transition to a dot in the continuation period u = Δt is shown. The parking transition A2 indicates a transition from the dot at which the home transition A3 is transitioned to a dot in the duration u = 2Δt at the future time of the time t = τ + 2Δt in the home parking model unit 36a.

  In the parking model unit 36a for home, the arrow extending diagonally to the lower right from the dot where the parking transition A2 has transitioned indicates that the maximum likelihood solution X1 transitions again to a dot in the duration u = Δt at time t = τ + 3Δt. Show. In this case, the maximum likelihood solution X1 predicts the occurrence of the state transition of the vehicle 15a again at time t = τ + 3Δt.

  Returning to FIG. 5, after the online processing is completed, the process proceeds to step S <b> 13 and it is determined whether or not the time of the discretization step width Δt has elapsed. In the present embodiment, it is determined whether or not 30 minutes have elapsed since the online processing was executed. If the time of the discretization step width Δt has not elapsed, the process waits as it is, and if the time of the discretization step width Δt has elapsed, the process proceeds to step S12 to perform online processing. In this case, online processing is repeatedly performed at intervals of the discretization step width Δt.

  The maximum likelihood solution obtained by executing the usage prediction process is not the Markov model 35, but the affiliation model 41 indicating the affiliation of the vehicle 15, the usage model 42 indicating the usage state of the vehicle 15, and the duration of the usage state of the vehicle 15. The continuation model 43 indicating that can be expressed. Here, regarding the use prediction of the vehicle 15a, a case where the maximum likelihood estimation unit 32 estimates the maximum likelihood solution X2 will be described with reference to FIG. In the maximum likelihood solution X2, the vehicle 15a moves from the home 11a to the work location 11b with the user's attendance and parks, and the vehicle 15a moves from the work location 11b to the home 11a with the user's return after work. It is predicted that the user will park and then go out with the vehicle 15a.

  As shown in FIG. 12, in the belonging model 41, the affiliation destination v of the vehicle 15 is the vertical axis, in the usage model 42, the usage state γ of the vehicle 15 is the vertical axis, and in the continuation model 43, the duration u is The vertical axis. In these models 41 to 43, time t is the horizontal axis.

  The affiliation model 41 indicates that the vehicle 15a belongs to one of the home 11a and the work place 11b with respect to the maximum likelihood solution X2. More specifically, it is expressed that the affiliation v0 of the vehicle 15a is the home 11a at the current time τ, and that the affiliation of the vehicle 15a is updated every time a state transition occurs. In this case, the affiliations v1 to v4 are set according to the future times t1 to t5. However, as the affiliation destinations v1 and v2 are both set to the work place 11b, the affiliation destination of the vehicle 15a does not change when the use state γ shifts from the movement state to the parking state.

  The maximum likelihood solution X2 of the usage model 42 includes a work place transition B3, a parking transition B2 at the work place 11b, a home transition A3, a parking transition A2 at the home 11a, and a movement transition A1. In the usage model 42, it is expressed that the usage state γ of the vehicle 15a is in the parking state at the current time τ, and that the state transition of the vehicle 15a occurs by repeating movement and parking in the future. In addition, the timing at which the state transition occurs is set to future times t1 to t5.

  The continuation model 43 indicates the remaining periods of the parking state and the movement state for the maximum likelihood solution X2, and the continuation periods with respect to future times t1 to t5 are indicated as u1 to u5. For example, the duration until the future time t2 is represented by u1.

  Here, the prediction result of the usage prediction unit 30 is output from the regional output unit 50 to each EMS 12 of a plurality of bases. In these EMS12, based on the prediction result regarding the use condition of the vehicle 15, the energy management using the vehicle-mounted storage battery of the vehicle 15 and the energy management which does not use a vehicle-mounted storage battery are implemented.

  For example, when the estimation result by the maximum likelihood estimator 32 is the maximum likelihood solution X2, in the period of future times t1 to t3 in which the vehicle 15a belongs to the work place 11b, energy using the vehicle 15a as energy storage Management is performed by the EMS 12 of the work place 11b. Further, in a period after the future time t3 when the vehicle 15a belongs to the home 11a, energy management using the vehicle 15a as the energy storage is performed by the EMS 12 of the home 11a.

  Next, a test performed for use prediction of the vehicle 15a will be described with reference to FIG. In this test, real-time prediction was performed with a prediction period of 24 hours. In FIG. 13, the maximum likelihood solution X3 estimated by the regional controller 16 is indicated by a broken line as a prediction result, and the observed value Y when the vehicle 15a is actually used is indicated by a solid line. FIG. 13 shows an affiliation model 41 and a use model 42. For the affiliation destination v of the affiliation model 41, the home 11a is indicated by L1, and the work location 11b is indicated by L2. Further, regarding the usage state γ of the usage model 42, the movement state is indicated by “1” and the parking state is indicated by “0”.

  As shown in FIG. 13, the use prediction of the vehicle 15a is implemented with the current time τ = 6: 00, 6:30, 7:00, 7:30 as the prediction timing. As described above, since the initial state probability and the state transition probability change with time, the maximum likelihood solution X at each prediction timing is different from each other. However, as the prediction timing shifts to 6:00, 6:30, 7:00, 7:30, the difference between the maximum likelihood solution X and the observed value Y for the future times 9:00, 12:00, etc. Is getting smaller.

  The operational effects of the first embodiment described so far will be described below.

  According to the first embodiment, since a plurality of buildings 11 are set as a plurality of bases for one vehicle 15, the presence or absence of the vehicle 15 can be individually predicted for each of the bases. Based on the prediction result, the energy management using the on-vehicle storage battery of the vehicle 15 is individually performed for each of the plurality of bases, so that the power usage for the power system that supplies power to each building 11 in the region 10 is targeted. The peak value of the total of can be reduced. In addition, individually performing energy management at a plurality of bases also reduces the burden on the power system in the region 10, so that it is possible to adjust the power supply / demand balance in units of regions.

  According to the first embodiment, since the use prediction of the vehicle 15 is performed with any one of a plurality of bases as the affiliation destination, the prediction result regarding the parking or movement of the vehicle 15 is clarified for each affiliation destination. Moreover, since the affiliation is changed with the movement of the vehicle between the bases, unlike the case where the vehicle 15 does not belong to any of the plurality of bases for the movement, the use prediction of the vehicle 15 is made to the plurality of bases. Can be done continuously.

  According to the first embodiment, the state transition of the vehicle 15 predicted by the use prediction unit 30 includes the first transition (A1, A2, B1, B2) that does not involve the change of the affiliation destination and the change of the affiliation destination. The accompanying second transition (A3, B3) is included. For this reason, the use prediction of the vehicle which considered the movement between several bases can be implemented.

  According to the first embodiment, since the second transition accompanied by the change of the affiliation indicates a state transition from the parking state to the moving state, the vehicle 15 is started at the timing when the vehicle 15 starts moving toward the base. Will be changed. In this case, the EMS 12 to which the vehicle 15 belongs can update the energy management using the in-vehicle storage battery according to the current situation at a timing before the vehicle 15 arrives at the destination.

  According to the first embodiment, when a user of the vehicle 15a is assumed, the home 11a and the work place 11b are set as the base of the vehicle 15a. For this reason, both the period in which the vehicle 15a is parked in the home 11a and the period in which the vehicle 15a is parked in the work place 11b can be effectively used as an opportunity to utilize the vehicle 15a as energy storage.

  According to the first embodiment, since the prediction result of the use prediction unit 30 is output to the EMS 12 of the building 11 set as the base of the vehicle 15, the use prediction of the vehicle 15 is effectively performed in the energy management related to the building 11. Available. In this case, it is possible to optimize the energy management by the EMS 12 by increasing the prediction accuracy for the use prediction of the vehicle 15.

  According to the first embodiment, since a plurality of bases are included in one region 10, opportunities for utilizing the in-vehicle storage battery in this region 10 are increased. For this reason, the power supply-demand balance can be adjusted for each region simply by the user going back and forth between a plurality of bases in the region 10.

  According to the first embodiment, since the maximum likelihood estimation unit 32 estimates the maximum likelihood solution based on the Markov model 35, it is possible to reduce the amount of calculation of a prediction problem called use prediction of the vehicle 15. In particular, since dynamic programming is used to estimate the maximum likelihood solution, the prediction time can be shortened.

(Second Embodiment)
In the first embodiment, only two bases to which the vehicle 15 can belong in the use prediction of the vehicle 15 are set. However, in the second embodiment, three bases are set. In the present embodiment, the description will focus on differences from the first embodiment.

  As shown in FIG. 14, in this embodiment, the user's home 11c is set in addition to the home 11a and the work place 11b as three bases for performing the use prediction of the user's vehicle 15a. For example, when the vehicle 15a departs from the home 11c and returns to the home 11c again without going to either the home 11a or the work place 11b, the vehicle 15a moves as the first transition in which the home is held in the home 11c. Transition C1 and parking transition C2 will be performed. Further, when the vehicle 15a moves from the home 11a and the work place 11b to the home 11c, the home 15c performs the home transition C3 and C4 as the second transition. Furthermore, the vehicle 15a performs home transition A4 as the second transition when moving from the home 11c to the home 11a, and performs work location transition B4 as the second transition when moving from the home 11c to the work location 11b. become.

  As shown in FIG. 15, in the Markov model 35, model units 36 and 37 are individually provided for each of the three bases. More specifically, model parts 36c and 37c for a home are added to the Markov model 35 of the present embodiment.

  Here, regarding the use prediction of the vehicle 15a, a case where the maximum likelihood estimation unit 32 estimates the maximum likelihood solution X4 will be described with reference to FIG. This maximum likelihood solution X4 indicates a prediction that the vehicle 15a moves from the work place 11b to the home 11a, parks at the home 11a, and then moves again toward the home 11c. The maximum likelihood solution X4 includes a home transition A3 accompanied by a movement from the work place 11b to the home 11a, a parking transition A2 held in the home 11a, and a home transition C3 accompanied by a movement from the home 11a to the home 11c. Have.

  Even when three bases are set, the maximum likelihood solution estimated by the maximum likelihood estimation unit 32 can be expressed by the belonging model 41, the use model 42, and the continuation model 43. Here, regarding the use prediction of the vehicle 15a, the case where the maximum likelihood estimation unit 32 estimates the maximum likelihood solution X5 will be described with reference to FIG. In the maximum likelihood solution X5, the vehicle 15a moves from the home 11a to the work place 11b and parks with the user's attendance, stops at the home 11c with the vehicle 15a after the work ends, and then the user takes the vehicle 15a from the home 11c with the vehicle 15a. It shows the prediction of going out.

  As shown in FIG. 16, in the affiliation model 41 of the present embodiment, a home 11c is added to the vertical axis representing the affiliation destination v of the vehicle 15a. In this case, the affiliation model 41 indicates that the vehicle 15a belongs to any one of the home 11a, the work place 11b, and the home 11c with respect to the maximum likelihood solution X5. In the affiliation model 41, the affiliation destination v of the vehicle 15a is the parents' home 11c in a period after the future time t3.

  The maximum likelihood solution X5 of the usage model 42 has a work location transition B3, a parking transition B2 at the work location 11b, a home transition C4, a parking transition C2 at the home 11c, and a movement transition C1. In the usage model 42, the usage state of the vehicle 15a transitions between the parking state and the movement state while the affiliation v of the vehicle 15a is held in the parents' home 11c in a period equal to or greater than the future time t3.

  According to the present embodiment, even when three bases are set for the use prediction of the vehicle 15, the use prediction of the vehicle 15 is appropriately set by assuming the first transition and the second transition for the third base. It can be carried out. Therefore, for example, even when the home 11a, the work place 11b, and the real home 11c are set as three bases, the use prediction of the vehicle 15a can be continuously performed for the home 11a, the work place 11b, and the real home 11c.

(Other embodiments)
Although a plurality of embodiments of the present invention have been described above, the present invention is not construed as being limited to these embodiments, and various embodiments and combinations can be made without departing from the scope of the present invention. Can be applied.

  In the modification 1, the 2nd transition accompanied by the change of the affiliation destination of the vehicle 15 may not be a state transition from the parking state to the moving state but a state transition from the moving state to the parking state. For example, when the vehicle 15a moves from the home 11a to the work place 11b, the timing at which the vehicle 15a belongs to the work place 11b from the home 11a is not the time when the vehicle 15a leaves the home 11a, but the vehicle 15a works. It is set as the timing which arrived at the ground 11b. In this case, regarding the work place transition B3 as the second transition, the affiliation destination of the vehicle 15a is not the work place 11b but the home 11a. Even in this case, use prediction of the vehicle 15 can be continuously performed for a plurality of bases.

  In the modified example 2, regarding the use prediction of the vehicle 15, when the vehicle 15 is in a moving state, the vehicle 15 may not belong to any of a plurality of bases. Even in this configuration, since the parking location and parking period of the vehicle 15 are predicted by the use predicting unit 30, the prediction result can be utilized for EMS energy management.

  In the modification 3, about the use prediction of the vehicle 15, you may predict separately the parking state with respect to a base, and the movement state between bases. For example, the base is not used as a reference when the usage prediction of the vehicle 15 is performed. In this configuration, when the vehicle 15a departs from the home 11a, the state transition from the parking state to the moving state is not distinguished depending on whether the destination is the home 11a or the work location 11b. That is, there is no distinction between the movement transition A1 when the affiliation is held at the home 11a and the work location transition B3 when the affiliation is changed from the home 11a to the work location 11b.

  In the modified example 4, at least one of the home 11a and the work place 11b may not be included in the plurality of bases. For example, for the vehicle 15 owned by the company of the user's office, the office and business partners are included in a plurality of bases. In this configuration, the vehicle 15 can be effectively used as energy storage during a period in which the vehicle 15 is parked at each of the work place and the business partner.

  In the modified example 5, when the base setting unit 25 sets any of the plurality of buildings 11 as the base, the predetermined number of bases may be more than three. In short, it is sufficient if the number of bases is plural.

  In the modified example 6, a plurality of bases may be set by performing a selection process by the base setting unit 25 instead of selecting a plurality of bases by a user input operation. For example, the base setting unit 25 sets two of the plurality of buildings 11 as bases based on the frequency distribution acquired by the distribution acquisition unit 24. Specifically, in the usage history of the vehicle 15, the building 11 having the largest total parking period and the second largest building 11 are set as two bases. According to this configuration, since the two bases are selected based on the length of the parking period, in each of these bases, an appropriate parking period for transferring power between the vehicle 15 and the building 11. Can be secured.

  In the modified example 7, the area controller 16 may have a function of setting the number of bases. Examples of this configuration include a configuration in which the use predicting unit 30 has a number setting unit that sets the number of sites, and a configuration in which the site setting unit 25 can set the number of sites. For example, in the configuration in which the usage prediction unit 30 includes a number setting unit, the number setting unit determines how many buildings 11 are set as the base based on the usage history of the vehicle 15. In this configuration, the ratio of the parking period to the total of the past parking period of the vehicle 15 is calculated for each building 11 where the vehicle 15 has been parked in the past, and the ratio has reached a predetermined value (for example, 20%). All buildings 11 are set as bases.

  In the modified example 8, the base setting unit 25 may set land or equipment as a base instead of setting the building 11 as a base. Even in this case, if a power transmission / reception device that enables transmission / reception of electric power to / from the vehicle 15 is provided in the land or facility, the energy management of the base and the vehicle 15 can be performed.

  In the modification 9, the energy management for the building 11 may be performed by the regional controller 16 instead of the EMS 12 of the building 11. Even in this case, energy management can be performed individually for each of the plurality of bases.

  In the modified example 10, the building 11 set as the base may not be included in the single area 10. For example, it is assumed that the home 11a and the work place 11b are included in different areas.

  In the modified example 11, in the Markov process, the prediction period for predicting the use state of the vehicle 15 is discretized, but the prediction period may not be discretized.

  In the twelfth modification, an independent incremental process, a Gaussian process, or the like may be employed instead of the Markov process as the stochastic process used for predicting the use of the vehicle 15.

  DESCRIPTION OF SYMBOLS 10 ... Area, 11 ... Building, 11a ... Home, 11b ... Work location, 12 ... EMS, 15, 15a ... Vehicle, 25 ... Base setting part, 26 ... Model creation part, 30 ... Use prediction part, 32 ... Maximum likelihood estimation 35, Markov model, 50 ... Regional output, A1, B1, C1 ... Moving transition, A2, B2, C2 ... Parking transition, A3, A4 ... Home transition, B3, B4 ... Work location transition, C3, C4 ... Home transition, v ... belonging, γ ... use status.

Claims (8)

Regarding the bases (11, 11a to 11c) where the vehicle can be parked, whether the vehicle is in a parked state where the vehicle is parked or a moving state where the vehicle is not parked is used by the user. A vehicle usage prediction system that predicts by modeling a stochastic process for (γ),
A site setting unit (25) for setting a plurality of sites for the vehicle;
For each of the plurality of bases set by the base setting unit, a prediction unit (30) that predicts whether the use state of the vehicle is in the parking state or the moving state;
A vehicle use prediction system comprising:   The prediction unit predicts a use state of the vehicle by using any one of the plurality of bases as an affiliation (v) of the vehicle, and predicts movement of the vehicle between the plurality of bases. The vehicle usage prediction system according to claim 1, wherein the affiliation destination of the vehicle is changed along with the operation. The predicting unit predicts that the state of use of the vehicle changes state from one of the parking state and the moving state to the other,
As the state transition predicted by the prediction unit, the first transition (A1, A2, B1, B2, C1, C2) in which the use state of the vehicle changes without changing the affiliation destination of the vehicle, The use of the vehicle according to claim 2, characterized in that there is a second transition (A3, A4, B3, B4, C3, C4) in which the use state of the vehicle transitions with a change of the vehicle affiliation. Prediction system.   The said prediction part changes the affiliation place of the said vehicle, when the use condition of the said vehicle changes from the said parking state to the said movement state as said 2nd transition. Vehicle usage prediction system.   A prediction output unit (50) that outputs a prediction result of the prediction unit to a management unit (12) that performs energy management for each of the plurality of bases based on the prediction result of the prediction unit; The vehicle use prediction system according to any one of claims 1 to 4, wherein the vehicle use prediction system is characterized. The plurality of bases set by the base setting unit include the user's home (11a) and the user's work place (11b),
The vehicle use prediction system according to claim 5, wherein the management unit is individually provided in each of the home and the work place.   The vehicle use prediction system according to any one of claims 1 to 6, wherein the plurality of bases set by the base setting unit are included in a single region (10). A model creation unit (26) for creating a prediction model (35) by modeling the stochastic process for each of the plurality of bases;
A maximum likelihood estimator (32) for estimating a maximum likelihood solution for the usage state of the vehicle based on the prediction model created by the model creator;
The vehicle use prediction system according to claim 1, wherein the vehicle use prediction system is provided.