Prediction device, prediction method, and program

Abstract

The purpose of the present disclosure is to more easily predict the remaining capacity of a battery of a mobile body such as an EV.　The present disclosure is a prediction device that has a charge determination unit that determines that a battery has been charged when a prescribed travel distance that is based on information about travel distance included in usage / performance data is higher than a threshold value, a correction value calculation unit that, when the charge determination unit has determined that the battery has been charged, calculates a correction value that is a prescribed amount of charge on the basis of information included in the usage / performance data that indicates an amount of charge or information included in the usage / performance data that indicates a duration of charge, a prediction unit that uses a trained machine learning model to predict a value for the remaining capacity of the battery at a return time relative to scheduled usage data, a correction unit that uses the correction value to correct the remaining capacity of the battery at the return time predicted by the prediction unit, and an output unit that outputs the corrected remaining capacity of the battery.

Description

[Technical Field]

[0001] The present disclosure relates to a technology for predicting the remaining battery charge value of a vehicle such as an electric vehicle (EV). [Background technology]

[0002] In recent years, there has been a trend to utilize EV batteries for supply and demand control to match the supply and demand of electricity. Users who use EV batteries for supply and demand control need to know the remaining battery capacity (discharge capacity) or the available battery capacity (charge capacity) several days to several hours before the planned start time of use.

[0003] In response to this, a technology has been disclosed that predicts the remaining battery charge based on the past usage history of an EV battery (Non-Patent Document 1). The technology disclosed in Non-Patent Document 1 builds a model based on EV charging information and traffic information measured on expressways, and based on the model, calculates the distance and average speed from the initial position (starting point) to the current position using real-time location information acquired while the EV is traveling, and estimates the current remaining battery charge by calculating the amount of power consumed during that time. [Prior art documents] [Non-patent literature]

[0004] [Non-Patent Document 1] EV charging navigation system that prevents electric vehicles from running out of battery on highways (https: / / www.global.toshiba / content / dam / toshiba / migration / corp / techReviewAssets / tech / review / 2019 / 05 / 74_05pdf / f02.pdf.) Summary of the Invention [Problem to be solved by the invention]

[0005] However, in order to know the remaining battery power using the technology disclosed in Non-Patent Document 1, Building such a system requires a lot of costs, as it requires the installation of communication devices in the EV and a server to obtain information from the EV in real time. Moreover, the server needs to obtain a large amount of information from the EV in real time, such as charging information, traffic information, vehicle type, and current location, which makes the processing complicated and difficult to implement.

[0006] The present invention has been made in view of the above points, and aims to predict the remaining battery charge of a vehicle such as an EV more easily than conventional techniques. [Means for solving the problem]

[0007] In order to solve the above problem, the invention of claim 1 is a prediction device for predicting a remaining battery charge value of a vehicle, comprising: an acquisition unit that acquires usage history data including information on the mileage of the vehicle and information indicating the charge amount of the battery or information indicating the charging time of the battery as a past usage history of the vehicle, and acquires usage plan data including information on the mileage of the vehicle as a future usage plan of the vehicle; a charging determination unit that determines that the battery has been charged if a predetermined mileage based on the information on the mileage included in the usage history data is higher than a threshold; and a charging determination unit that determines that the battery has been charged if it is determined that the battery has been charged. The prediction device includes: a correction value calculation unit that calculates a correction value as a predetermined charge amount based on information indicating the charge amount or information indicating the charging time included in the data; a prediction unit that uses the planned use data as input and predicts the remaining battery amount at the time of return for the planned use data acquired by the acquisition unit using a trained machine learning model that outputs the remaining battery amount at the time of return of the vehicle; a correction execution unit that uses the correction value calculated by the correction value calculation unit to execute a correction to the remaining battery amount at the time of return predicted by the prediction unit; and an output unit that outputs the remaining battery amount after correction by the correction execution unit. [Effects of the Invention]

[0008] As described above, the present invention has the effect of making it possible to predict the remaining battery charge of a vehicle more easily than with the prior art. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is a schematic diagram illustrating a usage situation of a prediction device according to an embodiment. [Figure 2] FIG. 2 is a diagram illustrating the electrical hardware configuration of a prediction device and a server. [Figure 3] FIG. 10 is a functional configuration diagram of the prediction device in a learning phase. [Figure 4] FIG. 2 is a functional configuration diagram of a prediction device in a prediction (estimation) phase. [Figure 5] 10 is a conceptual diagram showing a usage history data management table constituting a usage history data management DB. FIG. [Figure 6] 10 is a conceptual diagram showing a planned use data management table constituting a planned use data management DB. FIG. [Figure 7] 10 is a flowchart illustrating a process or operation performed by the prediction device in a learning phase. [Figure 8] 10 is a flowchart illustrating a process or operation performed by the prediction device in an estimation phase. [Figure 9] 10 is a flowchart illustrating a process or operation performed by the prediction device in an estimation phase. [Figure 10] FIG. 10 is a diagram for explaining prediction of remaining battery capacity. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0011] [System configuration of the embodiment] First, an overview of the entire communication system of this embodiment will be described with reference to Fig. 1. Fig. 1 is a schematic diagram showing the usage status of a prediction device according to this embodiment.

[0012] As shown in FIG. 1, the communication system 1 of this embodiment is constructed by a prediction device 3 and a server 5.

[0013] The prediction device 3 is composed of one or more computers. When the prediction device 3 is composed of multiple computers, it may be referred to as a "prediction device" or a "prediction system." The prediction device 3 is a device that predicts the remaining battery capacity (discharge capacity) of a vehicle such as an EV7. Note that the prediction device 3 may predict the available battery capacity (charge capacity) instead of the remaining battery capacity. In this case, the prediction device 3 can finally determine the remaining battery capacity by subtracting the predicted available capacity from the battery capacity.

[0014] The server 5 is configured by one or more computers. The server 5 stores map information, various parameters, and the like.

[0015] The prediction device 3 and the server 5 can communicate via a communication network such as the Internet. The communication network may be connected wirelessly or via a wired connection. In FIG. 1, the prediction device 3 is located in a company at the departure point A. For example, an employee of the company at the departure point A drives an EV 7 from the departure point A to a destination B, and after finishing work such as a meeting at destination B, drives the EV from destination B back to the departure point A. FIG. 1 also shows a case where there is a charging station α between the departure point A and destination B, and a charging station β at destination B. In the situation shown in FIG. 1, the user (employee) can stop at charging station α to charge the EV 7 on the way there and back, and can also charge the EV 7 at charging station β after arriving at destination B.

[0016] [Hardware configuration] <Hardware configuration of the prediction device> Next, the electrical hardware configuration of the prediction device 3 will be described with reference to Fig. 2. Fig. 2 is a diagram showing the electrical hardware configuration of the prediction device.

[0017] As shown in FIG. 2, the prediction device 3 is a computer and includes a CPU (Central Processing Unit) 301, a ROM (Read Only Memory) 302, a RAM (Random Access Memory) 303, an SSD (Solid State Drive) 304, an external device connection I / F (Interface) 305, a network I / F 306, a media I / F 309, and a bus line 310.

[0018] Of these, the CPU 301 controls the overall operation of the prediction device 3. The ROM 302 stores programs such as an IPL (Initial Program Loader) used to drive the CPU 301. The RAM 303 is used as a work area for the CPU 301.

[0019] The SSD 304 reads or writes various data under the control of the CPU 301. Note that instead of the SSD 304, a hard disk drive (HDD) may be used.

[0020] The external device connection I / F 305 is an interface for connecting various external devices, such as a display, a speaker, a keyboard, a mouse, a USB (Universal Serial Bus) memory, and a printer.

[0021] The network I / F 306 is an interface for performing data communication via the communication network 100 .

[0022] The media I / F 309 controls reading and writing (storing) of data from and to a recording medium 309m such as a flash memory, etc. The recording medium 309m includes a DVD (Digital Versatile Disc) and a Blu-ray Disc (registered trademark).

[0023] The bus line 310 is an address bus, a data bus, etc. for electrically connecting the components such as the CPU 301 shown in FIG.

[0024] The electrical hardware configuration of the server 5 is the same as that of the prediction device 3, and therefore a description thereof will be omitted.

[0025] [Functional configuration of the prediction device] Next, the functional configuration of the prediction device 3 according to this embodiment in the learning phase and estimation (prediction) phase will be described.

[0026] <Learning phase functional configuration> First, each function of the prediction device 3 in the learning phase will be described in detail with reference to Fig. 3. Fig. 3 is a functional configuration diagram of the prediction device in the learning phase.

[0027] As shown in Fig. 3, the prediction device 3 includes an acquisition unit 31 and a machine learning unit 33. These units each have a function realized by an instruction from a CPU 301 in Fig. 2 based on a program. In addition, a machine learning model 30a and a usage history data management DB (Data Base) 41 are built in the RAM 303 or the SSD 304.

[0028] (Usage history data management DB) Figure 5 is a conceptual diagram showing the usage history data management table that constitutes the usage history data management DB. The usage history data management table manages each usage history data (each record in Figure 5) according to the number of times the EV7 has been used. The usage history data is data created based on the usage schedule data (see Figure 6), the user (employee, etc.) scheduler, and vehicle operation management data, which will be described later.

[0029] The usage history data management table manages information on the past usage history of the EV 7. Specifically, the usage history data management table associates and manages the following information: the date of use of the EV 7 (which may be the date and time of use), a user ID for identifying the user, the destination, the one-way distance (which may be the round-trip distance or the driving distance), the user's work (task) at the destination, the amount of time the EV 7 was used (h), the amount of time the user stayed at the destination (h), the remaining battery power (kWh) at the time of departure of the EV 7, the remaining battery power (kWh) at the time of return of the EV 7, whether the EV 7 was charged between departure and return, the amount of charge (kWh) if it was charged along the way, and the charging location if it was charged along the way.

[0030] The remaining battery capacity value at the time of departure of the EV7, the remaining battery capacity value at the time of return of the EV7, whether or not charging occurred between departure and return, the amount of charge if charging occurred along the way, and the charging location if charging occurred along the way may be recorded manually by the user, etc. Also, whether or not charging occurred between departure and return, or the amount of charge if charging occurred along the way, and the remaining battery capacity value (kWh) at the time of return of the EV7 may be calculated automatically, rather than manually, from the difference between the estimated standard consumption estimated from the electricity cost (catalog value) and one-way distance and the actual SoC value.

[0031] (machine learning model) The machine learning model 30a is constructed using a known machine learning algorithm such as a neural network, and performs machine learning on each piece of usage history data (each data set), using regression analysis, SVM (Support Vector Machine), or the like.

[0032] (Each functional configuration) The acquisition unit 31 acquires each piece of usage history data (each record in FIG. 5) from the usage history data management DB. The acquisition unit 31 also serves as an input unit that inputs each piece of data. The usage history data management DB 41 may be included in the server 5 in FIG. 1. In this case, the acquisition unit 31 acquires each piece of usage history data from the server 5 via a communication network.

[0033] The machine learning unit 33 trains the machine learning model 30g based on the usage history data.

[0034] <Functional configuration of the estimation phase> Next, each function of the prediction device 3 in the estimation phase will be described in detail with reference to Fig. 4. Fig. 4 is a functional configuration diagram of the prediction device in the estimation phase.

[0035] As shown in Fig. 5, the prediction device 3 includes an acquisition unit 31, a charging determination unit 34, a correction value calculation unit 35, a prediction unit 7, a correction execution unit 38, and an output unit 39. Each of these units has a function realized by an instruction from the CPU 301 in Fig. 2 based on a program. In addition, the RAM 303 or the SSD 304 stores a trained machine learning model 30b, a usage history data management DB 41, and a usage schedule data management DB 42. Note that functional components similar to those in the learning phase are denoted by the same reference numerals, and description thereof will be omitted.

[0036] (Pre-trained machine learning model) The trained machine learning model 30b represents the state of the machine learning model 30a after it has been trained through the learning phase. The trained machine learning model 30b receives the planned use data as input and outputs the remaining battery level at the time of arrival, etc.

[0037] (Data Management Database for Planned Use) Figure 6 is a conceptual diagram showing the usage schedule data management table that constitutes the usage schedule data management DB. The usage schedule data management table manages each usage schedule data (each record in Figure 6) corresponding to the scheduled number of times the EV7 is used. The usage schedule data is data created based on the user (employee, etc.) scheduler and vehicle operation management data.

[0038] The usage plan data management table manages information about future planned uses of the EV 7. Specifically, the usage plan data management table manages and associates information indicating the planned date of use of the EV 7 (or the planned date and time of use), the planned user ID, the planned destination of arrival, the expected one-way distance (or the round-trip distance or the driving distance), the planned work (task) content of the user at the destination, the planned time (hours) for using the EV, and the planned time (hours) for the user to stay at the destination.

[0039] The one-way distance may be automatically calculated from the destination.

[0040] (Each functional configuration) The acquisition unit 31 in the learning phase acquires each piece of usage history data from the usage history data management DB 41, and also acquires each piece of usage schedule data (each record in FIG. 6) from the usage schedule data management DB 42. Each piece of usage history data contains data that includes information about battery charging (information indicating the amount of charge of the battery or information indicating the charging time of the battery) (see the top three records in FIG. 5), or data that does not include information about this charging (see the bottom three records in FIG. 5). Note that the usage schedule data management DB 42 may be included in the server 5 in FIG. 1. In this case, the acquisition unit 31 acquires each piece of usage schedule data from the server 5 via a communication network.

[0041] The charging determination unit 34 determines that the battery has been charged if a predetermined traveling distance based on information regarding the vehicle's traveling distance included in the usage history data (such as information indicating the destination, information indicating the traveling distance, or information indicating the traveling time) is higher than a threshold value.

[0042] Alternatively, as another process, the charging determination unit 34 extracts, from each usage history data, specific usage history data that includes information on mileage that is the same as or similar to information on mileage included in the usage schedule data (information indicating the destination, information indicating the mileage, or information indicating the mileage time), and if this specific usage history data includes information on charging, it determines that the battery has been charged.

[0043] For example, even if the EV usage time is the same, the judgment result will be affected by factors such as whether there are chargers installed along the driving route, whether there is a charging station between the departure point and the destination, and whether there is a charging station at the destination.

[0044] Furthermore, if the EV 7 is a managed vehicle such as a company car, it is very effective to refer to the usage history data because the EV 7 has traveled to the same destination for the same reason in the past. On the other hand, if the EV 7 is a personal vehicle, and usage schedule data and usage history data are managed by an application on the personal smartphone, the charging determination unit 34 may refer to this data to determine whether or not to charge. Regardless of whether the vehicle is a company car or a personal vehicle, this data is also referred to by the correction value calculation unit 35, which will be described later.

[0045] When the charging determination unit 34 determines that charging has been performed, the correction value calculation unit 35 calculates a correction value as a predetermined charging amount based on information indicating the charging amount or information indicating the charging time included in the usage history data.

[0046] Alternatively, when the charging determination unit 34 performs another process described above and determines that charging has occurred, the correction value calculation unit 35 calculates a correction value as a predetermined charging amount based on the information indicating the charging amount or the information indicating the charging time among the information regarding charging included in the specific usage history data.

[0047] For example, if there is multiple historical usage data for the same destination, the correction value calculation unit 35 calculates the correction value using a representative value such as the average or most frequent value of past charging amounts. Also, if there is no charging amount that can be referenced in the historical usage data, the correction value calculation unit 35 may use, as the correction value, data calculated from the difference between the estimated standard consumption amount estimated from the electricity cost (catalog value) and one-way distance and the actual value of SoC.

[0048] The prediction unit 37 uses the trained machine learning model 30b, which takes the planned usage data as input and outputs the remaining battery level value at the time of the vehicle's return, to predict the remaining battery level value at the time of return for the planned usage data acquired by the acquisition unit 31.

[0049] The prediction unit 37 utilizes an appropriate prediction method such as regression analysis or SVM. It is assumed that managed vehicles such as company cars have traveled to the same destination for the same purpose in the past, and that performance management is performed. Furthermore, if similar data can be acquired using an application on a smartphone or the like, the managed vehicles are not limited to company cars, but may also be cars used in services such as car sharing.

[0050] The correction execution unit 38 corrects the remaining battery charge value at the time of return predicted by the prediction unit, using the correction value calculated by the correction value calculation unit 35. After the acquisition unit 31 acquires a plurality of pieces of usage history data, the correction execution unit 38 extracts, from the plurality of pieces of usage history data, specific usage history data that includes information on a mileage that is the same as or similar to information on a mileage included in the usage schedule data, and uses the correction value related to this specific usage history data to correct the remaining battery charge value at the time of return predicted by the prediction unit.

[0051] For example, the correction value calculation unit 35 may determine the maximum and minimum values ​​of the correction value (charge amount), and the correction execution unit 38 may adjust the correction value assuming multiple cases. Specifically, if a user intending to use an EV battery for supply and demand control wants information on discharge capacity (remaining capacity) at the time of supply and demand control on the day of use, the correction value calculation unit 35 may underestimate the remaining capacity based on the relevant schedule (scheduled use data for the target vehicle), or if a user wants charging capacity (available capacity), the correction value calculation unit 35 may overestimate the available capacity based on the relevant schedule, thereby adjusting the correction value for each vehicle.

[0052] The output unit 39 outputs the remaining battery capacity value that has not been corrected by the prediction unit 37 or the remaining battery capacity value that has been corrected by the correction execution unit 38. Examples of output include displaying the remaining battery capacity value on a display connected to the external device connection I / F 305 in Fig. 2, or transmitting the remaining battery capacity value to an external device via the network I / F 306.

[0053] [Processing or operation of the prediction device] Next, the processing or operation of the prediction device 3 in the learning phase and estimation phase will be described with reference to FIGS.

[0054] <Processing or operation in the learning phase> FIG. 7 is a flowchart showing the processes or operations executed by the prediction device in the learning phase.

[0055] S11: The acquisition unit 31 acquires each piece of past usage history data (each record in FIG. 5) from the usage history data management DB 41 as learning data.

[0056] S12: The machine learning unit 33 trains the machine learning model 30g based on the learning data acquired by the acquisition unit 31.

[0057] This completes the description of the processing or operation in the learning phase.

[0058] <Processing or operation in the estimation phase> 8 and 9 are flowcharts showing processes or operations executed by the prediction device in the estimation phase. Note that the processes in Fig. 8 and 9 are examples executed by the functional units shown in Fig. 4, and are not limited to the following processes.

[0059] S21: The acquisition unit 31 acquires each piece of usage history data from the usage history data management DB 41, and acquires each piece of scheduled usage data from the usage schedule data management DB .

[0060] S22: For each piece of usage history data, the charging determination unit 34 determines that charging has occurred if information about the travel distance included in the usage history data (information indicating the destination, information indicating the travel distance (one-way distance, round-trip distance, etc.), or information indicating the travel time) is higher than a threshold determined in advance by machine learning or the like. Alternatively, the charging determination unit 34 determines that charging has occurred based on information about charging included in the usage history data (information indicating that charging has occurred, information indicating the charging time, information indicating the amount of charge, or information indicating the charging location). For example, the charging determination unit 34 determines that charging has occurred for three pieces of predetermined usage history data (usage dates: 2022 / 2 / 20, 2022 / 2 / 23, 2022 / 3 / 23) related to the top three records among the usage history data in FIG. 5 .

[0061] S24: If it is determined in step S22 that charging has been performed (S23; YES), the correction value calculation unit 35 calculates a correction value as the charging amount based on information on the charging amount (information indicating the charging amount or information indicating the charging time) included in the predetermined usage history data for each piece of predetermined usage history data determined by the charging determination unit 34 to have been charged. For example, when the correction value calculation unit 35 uses the charging amount as information on the charging amount, it calculates the correction value by multiplying this charging amount by 1.0, etc. When the charging time is used as information on the charging amount, the correction value calculation unit 35 calculates the correction value based on the charging time and the known information on the "charging amount per charging time."

[0062] S25: The prediction unit 37 uses the trained machine learning model 30b to predict the remaining battery level at the time of arrival for each piece of usage schedule data acquired by the acquisition unit 31.

[0063] S25: The correction execution unit 38 extracts, for each scheduled usage data related to the target of correction, specific usage history data from the specified usage history data that includes information on the mileage included in the scheduled usage data and information on the same or similar mileage, and adds the correction value related to this specific usage history data to the remaining battery capacity value at the time of return predicted by the prediction unit 37, thereby generating a corrected remaining battery capacity value. For example, for planned usage data related to the correction target (planned usage date: 2022 / 5 / 1), the correction execution unit 38 extracts specific actual usage data (usage date: 2022 / 2 / 20) from three predetermined actual usage data (usage dates: 2022 / 2 / 20, 2022 / 2 / 23, 2022 / 3 / 23) that have the same destination B and mileage (one-way distance 60 km), and generates a corrected remaining battery capacity value by adding the correction value (70 kWh) calculated by the correction value calculation unit 35 in step S24 for this specific actual usage data to the remaining battery capacity value at the time of return predicted by the prediction unit 37. Similarly, the correction execution unit 38 generates corrected remaining battery capacity values ​​at the time of return for other planned usage data related to the correction target (planned usage dates: 2022 / 5 / 2, 2022 / 5 / 3). Furthermore, the above-mentioned "similar information" indicates a case where the destination related to the usage history and the destination related to the usage plan are within a predetermined value (for example, 3 km), or a case where the one-way distance related to the usage history and the one-way distance related to the usage plan are within a predetermined value (for example, 3 km).

[0064] S27: On the other hand, if it is not determined in step S22 that charging has been performed (S23; NO), the prediction unit 37 performs the same process as in step S27, i.e., uses the trained machine learning model 30b to predict the remaining battery level at the time of return for each planned use data acquired by the acquisition unit 31.

[0065] S28: The output unit 39 outputs the result data including the remaining battery capacity value at the time of arrival predicted in step S27 or the corrected remaining battery capacity value at the time of arrival corrected in step S26.

[0066] FIG. 10 is a diagram for explaining remaining battery capacity prediction. For example, as shown in FIG. 10, if the remaining battery capacity at the time of departure is S0 and the predicted value of power consumption without correction is D1, the output unit 39 can calculate the remaining battery capacity as "S0-D1" (see dashed dotted line). On the other hand, if the predicted value of power consumption after correction is D'1, the output unit 39 can calculate the remaining battery capacity as "S0-D'1" (see solid line). Note that when SoC (%) is obtained as the remaining battery capacity, the remaining capacity in kWh can be obtained by multiplying the output unit battery capacity by SoC.

[0067] As a result, on a specific future day, the output unit 39 can predict the available charging / discharging capacity for, for example, the evening of that day by adding up the calculated predicted values ​​per EV for all vehicles.

[0068] This completes the description of the processing or operation in the estimation phase.

[0069] [Effects of the embodiment] As described above, according to this embodiment, remaining battery charge can be predicted using data that is relatively widely used by people (companies) (personal schedulers, vehicle operation management data, etc.), such as planned usage data and actual usage data, and data derived from catalog values ​​(electricity costs), etc., thereby achieving the effect of being able to predict remaining battery charge of vehicles such as EVs more easily than conventional technology.

[0070] In addition, by using the correction value, it is possible to predict the remaining battery charge more realistically, even in cases where the battery does not run out of power but charging is required depending on the situation or person (for example, when there is a charging station at the destination), which makes it easier for users who use EV batteries for supply and demand control to plan their use.

[0071] 〔supplement〕 The present invention is not limited to the above-described embodiment, and may have the following configurations or processes (operations). (1) The prediction device 3 can be realized by a computer and a program, but this program can also be recorded on a (non-transitory) recording medium or provided via a communication network. (2) In the above embodiment, a notebook computer is shown as an example of the prediction device 3, but this is not limited to this and may be, for example, a desktop computer, a tablet terminal, a smartphone, a smartwatch, a car navigation device, a refrigerator, a microwave oven, etc. (3) Each CPU 301 may be not only a single CPU but also multiple CPUs. (4) An EV (electric vehicle) is an example of a battery-powered vehicle. This includes electric bicycles, electric kick scooters, and electric motorcycles. [Explanation of symbols]

[0072] 1. Communication Systems 3 Prediction Device 5 Server 30a Machine Learning Models 30b Trained machine learning model 31 Acquisition section (input section) 33 Machine Learning Department 34 Charge determination section 35 Correction value calculation unit 37 Prediction Department 38 Correction execution unit 39 Output section 41 Usage history data management DB (example of usage history data management section) 42 Scheduled use data management DB (an example of the scheduled use data management section)

Claims

1. A prediction device for predicting a remaining battery charge value of a vehicle, an acquisition unit that acquires usage history data including information on the mileage of the vehicle and information indicating the charge amount of the battery or information indicating the charging time of the battery as a past usage history of the vehicle, and also acquires usage plan data including information on the mileage of the vehicle as a future usage plan of the vehicle; a charging determination unit that determines that the battery has been charged when a predetermined mileage based on information regarding the mileage included in the usage history data is higher than a threshold value; a correction value calculation unit that calculates a correction value as a predetermined charging amount based on information indicating the charging amount or information indicating the charging time included in the usage history data when the charging determination unit determines that charging has been performed; and a prediction unit that uses a trained machine learning model that receives the usage plan data as input and outputs the remaining amount of the battery at the time of the vehicle's return, and predicts the remaining amount of the battery at the time of the vehicle's return based on the usage plan data acquired by the acquisition unit; a correction execution unit that executes a correction to the remaining battery charge value at the time of arrival predicted by the prediction unit, using the correction value calculated by the correction value calculation unit; an output unit that outputs the remaining battery charge value corrected by the correction execution unit; A prediction device having the following.

2. the acquiring unit acquires a plurality of pieces of usage history data; the correction execution unit extracts, from the plurality of pieces of usage history data, specific usage history data that includes information on a mileage that is the same as or similar to the information on the mileage included in the usage schedule data, and uses the correction value related to the specific usage history data to correct the remaining battery charge value at the time of return predicted by the prediction unit. The prediction device according to claim 1 .

3. A prediction device for predicting a remaining battery charge value of a vehicle, an acquisition unit that acquires, as a past usage history of the vehicle, a plurality of usage history data including information on a mileage of the vehicle and information on charging including information indicating a charge amount of the battery or information indicating a charging time of the battery, or not including information on charging, and acquires, as a future usage schedule of the vehicle, usage plan data including information on a mileage of the vehicle; a charging determination unit that extracts, from the plurality of pieces of usage history data, specific usage history data that includes information on a mileage that is the same as or similar to the information on the mileage included in the usage schedule data, and determines that the battery has been charged if the specific usage history data includes information on the charging; a correction value calculation unit that calculates a correction value as a predetermined amount of charge based on information indicating the amount of charge or information indicating the charging time among information regarding the charging included in the specific usage history data when the charging determination unit determines that charging has been performed; a prediction unit that uses a trained machine learning model that receives the usage plan data as input and outputs the remaining amount of the battery at the time of the vehicle's return, and predicts the remaining amount of the battery at the time of the vehicle's return based on the usage plan data acquired by the acquisition unit; a correction execution unit that executes a correction to the remaining battery charge value at the time of arrival predicted by the prediction unit, using the correction value calculated by the correction value calculation unit; an output unit that outputs the remaining battery charge value corrected by the correction execution unit; A prediction device having the following.

4. The prediction device according to claim 1 , wherein the information relating to the travel distance of the vehicle is information indicating a destination, information indicating a travel distance, or information indicating a travel time.

5. A prediction method executed by a prediction device that predicts a remaining battery charge value of a vehicle, comprising: The prediction device includes: an acquisition process of acquiring usage history data including information on the mileage of the vehicle and information indicating the charge amount of the battery or information indicating the charging time of the battery as a past usage history of the vehicle, and acquiring usage plan data including information on the mileage of the vehicle as a future usage plan of the vehicle; a charging determination process for determining that the battery has been charged when a predetermined mileage based on information about the mileage included in the usage history data is higher than a threshold value; a correction value calculation process for calculating a correction value as a predetermined charging amount based on information indicating the charging amount or information indicating the charging time included in the usage history data when it is determined that charging has been performed by the charging determination process; a prediction process for predicting the remaining battery level at the time of return for the usage plan data acquired by the acquisition process, using a trained machine learning model that receives the usage plan data as input and outputs the remaining battery level at the time of return of the vehicle; a correction execution process for executing a correction to the remaining battery charge value at the time of return predicted by the prediction process, using the correction value calculated by the correction value calculation process; an output process for outputting the remaining battery charge value corrected by the correction execution process; A forecasting method that performs.

6. A program causing a computer to execute the method according to claim 5.

7. A prediction method executed by a prediction device that predicts a remaining battery charge value of a vehicle, comprising: The prediction device includes: an acquisition process for acquiring, as a past usage history of the vehicle, a plurality of usage history data including or excluding information on charging, the information including information on the mileage of the vehicle and information on charging, the information including information on the charge amount of the battery or information on the charging time of the battery, and acquiring, as a future usage schedule of the vehicle, usage plan data including information on the mileage of the vehicle; a charging determination process for extracting, from the plurality of pieces of usage history data, specific usage history data including information on a mileage that is the same as or similar to the information on the mileage included in the usage schedule data, and determining that the battery has been charged if the specific usage history data includes information on the charging; a correction value calculation process for calculating a correction value as a predetermined amount of charge based on information indicating the amount of charge or information indicating the charging time among information regarding the charge included in the specific usage history data when it is determined that charging has been performed by the charging determination process; a prediction process for predicting the remaining battery level at the time of return for the usage plan data acquired by the acquisition process, using a trained machine learning model that receives the usage plan data as input and outputs the remaining battery level at the time of return of the vehicle; a correction execution process for executing a correction to the remaining battery charge value at the time of return predicted by the prediction process, using the correction value calculated by the correction value calculation process; an output process for outputting the remaining battery charge value corrected by the correction execution process; A forecasting method that performs.

8. A program causing a computer to execute the method according to claim 7.

Images