Information processing device, information processing method, and program

Abstract

To provide an information processing apparatus capable of efficiently improving benefits enjoyed by a hydrogen station.SOLUTION: An information processing apparatus 10 comprises a prediction unit 140 and a price determination unit 210. The prediction unit 140 predicts a demand for hydrogen at a hydrogen station using a demand prediction model being a learned model generated in advance by machine learning, for receiving at least a customer's behavior pattern as input and outputting the predicted demand for hydrogen. The price determination unit 210 determines a hydrogen price at the hydrogen station based on the predicted demand for hydrogen.SELECTED DRAWING: Figure 14

Description

[Technical field]

[0001] The present invention relates to an information processing device, an information processing method, and a program, and more particularly to an information processing device, an information processing method, and a program for predicting demand for hydrogen at a hydrogen station. [Background technology]

[0002] Patent Document 1 discloses a control method for a hydrogen station reservation system that aims to smoothly fill hydrogen fuel. The method in Patent Document 1 allows a user to input reservation information for reserving the date and time for filling hydrogen fuel at a hydrogen station, and creates a hydrogen filling reservation table in which the input reservation information can be registered. The method in Patent Document 1 also calculates the required amount of hydrogen fuel on a target date that is a preset number of days after the date of reading the reservation information from the user, using the hydrogen filling reservation table in which the reservation information from the user is registered. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2016-183768 A Summary of the Invention [Problem to be solved by the invention]

[0004] In the technology disclosed in Patent Document 1, unless a user makes a reservation, it is not possible to know whether or not the user will visit the hydrogen station, so it is not possible to accurately predict the demand for hydrogen. In addition, the price of hydrogen at the hydrogen station is not determined based on future demand, so it may not match the demand for hydrogen. Therefore, it may be difficult to increase the profits of the hydrogen station.

[0005] The present invention provides an information processing device, an information processing method, and a program that can efficiently improve the benefits enjoyed by hydrogen stations. [Means for solving the problem]

[0006] The information processing device of the present invention has a prediction unit that predicts the demand for hydrogen at a hydrogen station using a demand prediction model, which is a learned model generated in advance by machine learning, and which takes at least customer behavior patterns as input and outputs predicted demand for hydrogen, and a determination unit that determines the price of hydrogen at the hydrogen station based on the predicted demand for hydrogen.

[0007] In addition, the information processing method of the present invention predicts the demand for hydrogen at a hydrogen station using a demand prediction model, which is a learned model generated in advance by machine learning, and which takes at least customer behavior patterns as input and outputs predicted demand for hydrogen, and determines the price of hydrogen at the hydrogen station based on the predicted demand for hydrogen.

[0008] In addition, the program of the present invention causes a computer to execute the steps of predicting the demand for hydrogen at a hydrogen station using a demand prediction model, which is a learned model generated in advance by machine learning, and which takes at least customer behavior patterns as input and outputs predicted demand for hydrogen, and determining the price of hydrogen at the hydrogen station based on the predicted demand for hydrogen.

[0009] Because the present invention is configured as described above, it is possible to adjust the demand for hydrogen at a hydrogen station, which makes it possible to operate the hydrogen station efficiently. Furthermore, because the present invention is configured as described above, it is possible to improve the profits of the hydrogen station. The present invention makes it possible to efficiently improve the profits enjoyed by the hydrogen station.

[0010] In addition, preferably, the determination unit determines the price of hydrogen based on the ratio of a predicted demand, which is the predicted demand for hydrogen, to a supplyable amount, which is the amount of hydrogen that can be supplied, at a certain time at the hydrogen station. By configuring the present invention in this manner, it becomes possible to adjust the demand for hydrogen and the price of hydrogen in accordance with the objectives of the hydrogen station, based on the predicted demand and available supply.

[0011] In addition, preferably, when a first ratio, which is the ratio of the predicted demand to the supplyable amount, is greater than or equal to a predetermined first threshold and is less than or equal to a predetermined second threshold which is greater than the first threshold, the determination unit determines the price of hydrogen between a predetermined first price and a predetermined second price, with the price increasing as the first ratio increases. By configuring the present invention in this way, it becomes possible to improve profits according to the predicted demand amount and the available supply amount.

[0012] Also, preferably, when the first ratio is lower than the first threshold value, the determination unit determines the price of hydrogen to be lower than the first price. By being configured in this manner, the present invention makes it possible to promote demand for hydrogen when the predicted demand amount is too low compared to the supplyable amount.

[0013] Also, preferably, when the first ratio exceeds the second threshold value, the determination unit determines the price of hydrogen to be higher than the second price. By being configured in this manner, the present invention makes it possible to suppress the demand for hydrogen when the predicted demand amount is excessively large compared to the supplyable amount.

[0014] In addition, preferably, the demand prediction model takes as input the behavioral pattern and the price of hydrogen at the hydrogen station, and the determination unit determines the price of hydrogen based on the price of hydrogen input into the demand prediction model and the predicted demand, which is the demand for hydrogen obtained using the demand prediction model. By being configured in this manner, the present invention can more reliably improve profits.

[0015] Preferably, the hydrogen supply system further comprises a notification unit that notifies the customer of the determined price of hydrogen. By being configured in this manner, the present invention can improve convenience for customers.

[0016] In a preferred embodiment, the system further includes a learning unit that performs machine learning on the demand forecasting model, and the learning unit continuously learns the demand forecasting model in accordance with the difference between the demand predicted by the prediction unit and the actual demand. By configuring the present invention in this manner, it becomes possible to predict changes in hydrogen demand depending on the determined hydrogen price, thereby making it possible to further improve the accuracy of hydrogen demand prediction. Effect of the Invention

[0017] According to the present invention, it is possible to provide an information processing device, an information processing method, and a program that can efficiently improve the benefits enjoyed by a hydrogen station. [Brief description of the drawings]

[0018] [Figure 1] FIG. 1 illustrates an information processing system according to a first embodiment. [Diagram 2] FIG. 2 is a diagram illustrating a hardware configuration of the information processing device according to the first embodiment. [Diagram 3] 1 is a block diagram showing a configuration of an information processing device according to a first embodiment; [Figure 4] FIG. 2 is a diagram illustrating an example of input data input to the demand forecasting model according to the first embodiment. [Diagram 5] FIG. 4 is a diagram illustrating an example of feature amounts in input data according to the first embodiment. [Figure 6] FIG. 2 is a diagram illustrating a customer behavior pattern according to the first embodiment. [Figure 7] FIG. 2 is a diagram illustrating a customer behavior pattern according to the first embodiment. [Figure 8] FIG. 2 is a diagram illustrating a customer behavior pattern according to the first embodiment. [Figure 9] FIG. 2 is a diagram illustrating a customer behavior pattern according to the first embodiment. [Figure 10] FIG. 4 is a diagram illustrating an example of output data output from the demand forecasting model according to the first embodiment. [Figure 11] 4 is a diagram illustrating a demand forecast obtained by a demand forecasting unit according to the first embodiment. FIG. [Figure 12] 4 is a flowchart showing an information processing method executed by the information processing device according to the first embodiment. [Figure 13] 4 is a flowchart showing an information processing method executed by the information processing device according to the first embodiment. [Figure 14] FIG. 11 is a block diagram showing a configuration of an information processing device according to a second embodiment. [Figure 15] FIG. 11 is a diagram for explaining an example of a price determination method performed by a price determination unit according to the second embodiment. [Figure 16] FIG. 11 is a diagram illustrating an example of actual hydrogen demand when the hydrogen price is determined by the price determination unit according to the second embodiment. [Figure 17] FIG. 11 is a diagram illustrating a hydrogen price notification according to the second embodiment. [Figure 18] 11 is a flowchart showing an information processing method executed by an information processing device according to a second embodiment. [Figure 19] FIG. 11 is a block diagram showing a configuration of an information processing device according to a third embodiment. [Figure 20] FIG. 13 is a diagram illustrating an example of feature amounts in input data according to the third embodiment. [Figure 21] 13 is a flowchart showing an information processing method executed by an information processing device according to a third embodiment. [Figure 22] FIG. 13 is a block diagram showing a configuration of an information processing device according to a fourth embodiment. [Figure 23] 13 is a flowchart showing an information processing method executed by an information processing device according to a fourth embodiment. [Figure 24] FIG. 13 is a diagram for explaining a price determination method performed by a price determination unit according to the fourth embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0019] (Embodiment 1) Hereinafter, an embodiment of the present invention will be described with reference to the drawings. For clarity of explanation, the following description and drawings are omitted and simplified as appropriate. In addition, in each drawing, the same elements are given the same reference numerals, and duplicate explanations are omitted as necessary.

[0020] FIG. 1 is a diagram showing an information processing system 1 according to a first embodiment. The information processing system 1 includes a plurality of vehicles 2 and an information processing device 10. The vehicles 2 are vehicles (e.g., fuel cell vehicles) that use hydrogen as fuel. The information processing device 10 is, for example, a computer such as a server. The information processing device 10 and the vehicles 2 can be communicatively connected via a network 1a, such as wirelessly. The vehicle 2 may have a hardware configuration of the information processing device 10, which will be described later with reference to FIG. 2.

[0021] The information processing device 10 predicts the demand for hydrogen at a hydrogen station that supplies hydrogen to the vehicle 2. Specifically, the information processing device 10 predicts the demand for hydrogen by a machine learning algorithm such as deep learning, a neural network, or a recurrent neural network. The information processing device 10 may be realized by one or more computers. The information processing device 10 may also be realized by a cloud system. Therefore, the information processing device 10 is not limited to being realized by one physical device.

[0022] 2 is a diagram showing a hardware configuration of the information processing device 10 according to the first embodiment. The information processing device 10 has, as its main hardware components, a CPU 12 (Central Processing Unit), a ROM 14 (Read Only Memory), a RAM 16 (Random Access Memory), and an interface unit 18 (IF; Interface). The CPU 12, the ROM 14, the RAM 16, and the interface unit 18 are connected to each other via a data bus or the like.

[0023] The CPU 12 has a function as a calculation device (processing device or processor) that performs control processing and calculation processing. The calculation device may be realized by a dedicated device for machine learning, such as an NPU (Neural network Processing Unit) or a GPU (Graphics Processing Unit). The ROM 14 has a function as a storage that stores a control program and a calculation program executed by the CPU 12 (calculation device). The RAM 16 has a function as a memory that temporarily stores processing data and the like. The interface unit 18 has a function as a communication device that inputs and outputs signals to and from the outside via wired or wireless communication. The interface unit 18 also has a function as a user interface that accepts a data input operation by a user and performs processing to display information to the user. The interface unit 18 may display the results of the demand forecast.

[0024] 3 is a block diagram showing a configuration of the information processing device 10 according to the first embodiment. The information processing device 10 according to the first embodiment includes a learning unit 100, a trained model storage unit 122, an input data acquisition unit 124 (acquisition unit), a prediction unit 140, a notification unit 150, and a learning continuation processing unit 160. The learning unit 100 includes a teacher data acquisition unit 102 and a demand prediction model learning unit 104. The prediction unit 140 includes a demand prediction unit 142 and a supplyable amount determination unit 144.

[0025] These components can be realized, for example, by the CPU 12 (arithmetic unit) executing a program stored in the ROM 14 (storage unit). Each component may be realized by recording a necessary program in an arbitrary non-volatile recording medium and installing it as necessary. Each component is not limited to being realized by software as described above, but may be realized by hardware such as some circuit element. One or more of the above components may be realized by physically separate hardware. For example, the learning unit 100 may be realized by hardware separate from the other components. The same applies to other embodiments described later.

[0026] The learning unit 100 learns a demand prediction model for predicting hydrogen demand at hydrogen stations using the above-mentioned machine learning algorithm. In other words, the learning unit 100 performs machine learning on the demand prediction model. The learning unit 100 performs machine learning to predict hydrogen demand using at least customer behavior patterns. Therefore, the demand prediction model receives input data including at least customer behavior pattern information indicating customer behavior patterns, and outputs demand (predicted demand amount of hydrogen (demand forecast amount)) for each hydrogen station. The demand forecast amount indicates the demand amount of hydrogen after a predetermined period of time (e.g., one day later, two days later, one week later, one month later, etc.).

[0027] The teacher data acquisition unit 102 acquires teacher data, which is a set of input data and correct answer data. The input data includes customer behavior pattern information and area information. Here, the input data is time-series data whose feature values ​​change over time.

[0028] The customer behavior pattern information indicates the behavior pattern of each of a plurality of customers. Therefore, the customer behavior pattern information can be generated for each of a plurality of customers. The customer behavior pattern information can be acquired, for example, from the vehicle 2 owned by the customer via the network 1a. The customer behavior pattern information indicates, for example, the timing (frequency of filling) when the customer fills the vehicle 2 with hydrogen, the hydrogen stations visited by the customer, and the amount of hydrogen filled when filling. Details will be described later.

[0029] The area information is different from the behavioral patterns of the customers and indicates various information in the area. For example, the area information indicates the weather, information about hydrogen stations in the corresponding area, information about events in the corresponding area, etc. Details will be described later.

[0030] The correct answer data corresponds to the output data in the operation phase (inference phase, prediction phase). As described above, the output data indicates the demand for hydrogen for each hydrogen station after a predetermined period of time. Therefore, the correct answer data corresponds to the actual demand for hydrogen for each hydrogen station at a certain time.

[0031] The demand forecast model learning unit 104 performs learning processing of the demand forecast model using the acquired teacher data. The demand forecast model can be realized by a machine learning algorithm such as deep learning, neural network, or recurrent neural network. The demand forecast model learning unit 104 learns the demand forecast model so that input data is input and correct answer data is output. The demand forecast model learning unit 104 may generate a demand forecast model using teacher data in a certain period (e.g., several months) as learning data. Then, the demand forecast model learning unit 104 may adjust parameters (weights, etc.) of the demand forecast model using teacher data in a predetermined period (e.g., several weeks) after the period as evaluation data. The demand forecast model learning unit 104 may also extract important features from the input data by an autoencoder.

[0032] FIG. 4 is a diagram illustrating input data input to the demand forecasting model according to the first embodiment. As illustrated in FIG. 4, the input data is time-series data of a plurality of feature quantities. In the example of FIG. 4, the input data is shown with the horizontal axis representing the time axis and the vertical axis representing the time-series feature quantities. That is, each of the feature quantities x1, x2, x3, . . . , x N is time series data. N is the number of features. The features may be sampled, for example, every predetermined period Δt. In this case, t1, t2, t3, . . . , t k The time interval between the sampling periods is Δt. Δt may be, for example, 30 minutes, 1 hour, 6 hours, or 1 day (24 hours). This sampling period Δt may be set appropriately depending on the time-series granularity of the desired demand forecast. For example, the sampling period Δt when it is desired to obtain a demand forecast every few hours may be shorter than the sampling period Δt when it is desired to obtain a demand forecast every few days.

[0033] In addition, input data can be generated for each customer and region. For example, input data (customer behavior pattern information) U1, U2, and U3 for customer #1, customer #2, and customer #3 are generated, respectively. In addition, input data (region information) U1, U2, and U3 for region #1, region #2, and region #3 are generated, respectively. m+1 ,U m+2 ,U m+3 These input data U1 to U M The set of is input as input data to the demand forecasting model.

[0034] FIG. 5 is a diagram illustrating an example of feature quantities in input data according to the first embodiment. Note that the feature quantities illustrated in FIG. 5 are merely examples, and various other feature quantities are possible. Here, in FIG. 5, components x1 to x n indicates the feature value in the customer behavior pattern information. Also, the component x n+1 ~x N indicates the feature value in the regional information. In addition, in the customer behavior pattern information, x n+1 ~x N The value of may be 0. Similarly, in the area information, x1 to x n The value of may be 0.

[0035] 5, a component x1 indicates the position (vehicle position) of the vehicle 2 of the corresponding customer at the corresponding time (sampling time) for the feature amount included in the customer behavior pattern information. Also, a component x2 indicates the remaining amount of hydrogen in the vehicle 2 of the corresponding customer at the corresponding time (sampling time). The remaining amount of hydrogen may be the filling rate (State Of Charge: SOC).

[0036] Furthermore, component x3 indicates the hydrogen station that the corresponding customer visited to fill the vehicle 2 with hydrogen at the corresponding time (sampling time). The component value of x3 is determined in advance for each hydrogen station, for example, "hydrogen station A: x3=1", "hydrogen station B: x3=2", etc. If the customer has not visited the hydrogen station at the corresponding time (sampling time), the component value of x3 may be 0.

[0037] Furthermore, component x4 indicates the amount of hydrogen filled into the vehicle 2 of the corresponding customer. The filled amount may be the increased filling rate when hydrogen is filled. If the customer has not filled hydrogen into the vehicle 2 at the corresponding time (sampling time), the component value of x4 may be 0. Furthermore, component x5 indicates reservation information regarding the corresponding customer. The reservation information indicates whether or not the customer has made a reservation in advance to visit the hydrogen station and fill up with hydrogen for the corresponding time (sampling time). The component value of x5 is predetermined depending on whether or not there is a reservation, for example, "with reservation: x5=1" and "without reservation: x5=0".

[0038] Furthermore, component x6 indicates the frequency of visits by the corresponding customer to each hydrogen station. Furthermore, component x7 indicates the frequency with which the corresponding customer fills up the vehicle 2 with hydrogen. Furthermore, component x8 indicates seasonal fluctuations in the behavioral pattern of the corresponding customer. Note that, as will be described later, x6 to x8 do not have to be time-series data, and can be derived from customer behavioral pattern information. Therefore, x6 to x8 do not have to be included as feature quantities.

[0039] Regarding the feature quantity included in the regional information, in the example shown in Fig. 5, the component x n+1 indicates the weather in the corresponding area at the corresponding time (sampling time). n+1 The component value of is, for example, "Clear: x n+1 =1", "Rainy weather: x n+1 = 2" are predefined for each weather (sunny, rainy, etc.). n+2 indicates the temperature of the corresponding area at the corresponding time (sampling time). Also, the component x n+3 indicates the operation status of a hydrogen station installed in a corresponding area at a corresponding time (sampling time). The operation status indicates, for example, whether a corresponding hydrogen station is in operation for each day of the week and each time period. In addition, the component x n+4indicates event information for the corresponding area. The event information may indicate the type of event to be held at the corresponding time (sampling time) and the scale of the event (such as the number of people that can be accommodated).

[0040] 6 to 9 are diagrams illustrating customer behavior patterns according to the first embodiment. The customer behavior patterns illustrated in FIGS. 6 to 9 are shown as graphs with the horizontal axis representing time and the vertical axis representing the hydrogen filling rate (remaining hydrogen amount) of the corresponding customer's vehicle 2. Therefore, the customer behavior patterns are time-series data. Note that FIGS. 6 to 9 show the hydrogen filling rate over time. Therefore, the filling rate over time in FIGS. 6 to 9 corresponds to the feature "remaining hydrogen amount" illustrated in FIG. 5. Note that the customer behavior patterns may also show the location of the corresponding vehicle 2 over time. In this case, the location of the vehicle 2 over time corresponds to the feature "vehicle location" illustrated in FIG. 5.

[0041] FIG. 6 illustrates an example of a customer behavior pattern of customer #1. In the customer behavior pattern illustrated in FIG. 6, the hydrogen filling rate of customer #1's vehicle 2 drops to 20% about two weeks after the hydrogen filling rate is 90%. Then, the first time the filling rate drops to 20% (time t11), customer #1 visits hydrogen station A and fills up the vehicle from 20% to 90%, i.e., with hydrogen for a filling rate of 70%. At this time, customer #1 makes a reservation to visit hydrogen station A and fill up with hydrogen.

[0042] Furthermore, when the filling rate drops to 20% for the second time (time t12), customer #1 visits hydrogen station B and fills up from 20% to 90%, i.e., a filling rate of 70%. At this time, customer #1 does not make a reservation to visit hydrogen station B to fill up with hydrogen. When the filling rate drops to 20% for the third time (time t13), customer #1 visits hydrogen station A to fill up with hydrogen from 20% to 90%, i.e., a filling rate of 70%. At this time, customer #1 does not make a reservation to visit hydrogen station A to fill up with hydrogen.

[0043] In the customer behavior pattern illustrated in Fig. 6, visiting hydrogen stations A, B, and A at times t11, t12, and t13, respectively, corresponds to the feature "visited hydrogen stations" illustrated in Fig. 5. Also, filling with hydrogen to a filling rate of 70% at times t11, t12, and t13, respectively, corresponds to the feature "amount filled per trip" illustrated in Fig. 5. Also, "reservation: yes," "reservation: no," and "reservation: no" at times t11, t12, and t13, respectively, correspond to the feature "reservation information" illustrated in Fig. 5.

[0044] Furthermore, the fact that customer #1 visited hydrogen station A at time t11 and time t13, and visited hydrogen station B at time t12 corresponds to the feature "visit frequency per hydrogen station" illustrated in Fig. 5. Furthermore, the fact that hydrogen was filled up every two weeks corresponds to the feature "filling frequency" illustrated in Fig. 5.

[0045] FIG. 7 illustrates a customer behavior pattern of customer #1 in a season different from that in FIG. 6. FIG. 6 corresponds to a customer behavior pattern in summer, and FIG. 7 corresponds to a customer behavior pattern in winter. In summer, customer #1 fills hydrogen into vehicle 2 when the filling rate drops to 20%, but in winter, customer #1 fills hydrogen into vehicle 2 when the filling rate drops to 40%. That is, customer #1 fills hydrogen into vehicle 2 before the filling rate drops more in winter than in summer. On the other hand, customer #1 fills hydrogen into vehicle 2 every two weeks in summer, but in winter, customer #1 fills hydrogen into vehicle 2 every three weeks. That is, customer #1 fills hydrogen less frequently in winter than in summer. In this way, the behavior pattern differs depending on the season, which corresponds to the "seasonal variation" of the feature illustrated in FIG. 5.

[0046] FIG. 8 illustrates the customer behavior pattern of customer #2. FIG. 9 illustrates the customer behavior pattern of customer #3. Note that the time axis is the same in FIG. 8 and FIG. 9. As illustrated in FIG. 8, customer #2 fills hydrogen into vehicle 2 once a month. Also, customer #2 fills hydrogen into vehicle 2 when the filling rate drops to 20%. On the other hand, as illustrated in FIG. 9, customer #3 fills hydrogen into vehicle 2 once every two weeks, but may not fill hydrogen into vehicle 2 for two months. Also, customer #3 fills hydrogen into vehicle 2 when the filling rate drops to 40%. In other words, customer #3 normally fills hydrogen more frequently than customer #2. Also, customer #3 fills hydrogen into vehicle 2 before the filling rate drops more than customer #2. Also, customer #2 fills hydrogen into vehicle 2 at roughly the same intervals, whereas customer #3 has periods of low hydrogen consumption, so the intervals at which hydrogen is filled into vehicle 2 are not constant. In this way, behavioral patterns may vary depending on the customer.

[0047] Fig. 10 is a diagram illustrating an example of output data output from the demand forecasting model according to the first embodiment. As illustrated in Fig. 10, the demand forecasting model outputs the predicted hydrogen demand at each hydrogen station after a predetermined period of time. In the example of Fig. 10, for hydrogen station A, the demand forecasting model outputs the hydrogen demand after period T1, the hydrogen demand after period T2, the hydrogen demand after period T3, and the hydrogen demand after period T4. The same is true for hydrogen station B and hydrogen station C.

[0048] Here, in the teacher data used in the learning stage, the correct answer data may correspond to the output data exemplified in FIG. 10. Therefore, the correct answer data may be, for example, the last point in time on the time series of the customer behavior pattern information input for hydrogen station A (t in FIG. 4). k (corresponding to the above) to after a period T1, a period T2, a period T3, and a period T4.

[0049] In the learning stage, information prior to the predicted timing of hydrogen demand (a point in time after a specified period, such as after period T1) may be used as input data for customer behavior pattern information. In the operation stage, past information in a time series may be used as input data for customer behavior pattern information. This is because it is practically difficult to obtain future customer behavior pattern information in the operation stage. In the case of reservation information in customer behavior pattern information, if there is a reservation at the predicted timing, information up to the predicted timing (future information) may be used as input data.

[0050] On the other hand, for regional information, information up to the prediction target time may also be used as input data. In other words, in the operation stage, for regional information, future information may also be used as input data. Here, in the example of FIG. 5, "weather" and "temperature" can be obtained from a weather forecast. Furthermore, "hydrogen station operation status" can be obtained from the operation schedule of the hydrogen station. Furthermore, "event information" can be obtained from the event schedule.

[0051] When learning a demand forecast for a period T1 after time T0, the demand forecast model learning unit 104 may input data for a period ΔT going back from T0, and may learn a demand forecast model using the actual hydrogen demand from time T0 to the period T1 as correct answer data. Note that ΔT is the time from t1 to t on the time axis in FIG. 4. k where ΔT>Δt. For example, when the sampling period is Δt=30 minutes, ΔT=6 hours may be set and input data for the past six hours from T0 may be input to the demand forecasting model. When the sampling period is Δt=24 hours, ΔT=1 month may be set and input data for the past one month from T0 may be input to the demand forecasting model. When the sampling period is Δt=24 hours, ΔT=1 year may be set and input data for the past one year from T0 may be input to the demand forecasting model.

[0052] When the demand prediction model learning unit 104 finishes learning the demand prediction model, it outputs the learned demand prediction model to the learned model storage unit 122. As a result, the learned model storage unit 122 stores the demand prediction model, which is a learned model generated in advance by machine learning. The demand prediction model, which is a learned model, receives input data, which is time-series data including features as exemplified in Fig. 4 and Fig. 5, and outputs the predicted demand for hydrogen for each hydrogen station as exemplified in Fig. 10.

[0053] Furthermore, the learning unit 100 may continue to perform learning on the demand prediction model according to the difference between the demand predicted by the prediction unit 140 (to be described later) and the actual demand. This will be described in detail later.

[0054] The input data acquisition unit 124 acquires the above-mentioned input data in the operation stage. Here, the input data acquisition unit 124 acquires at least a customer behavior pattern (customer behavior pattern information) as input data. The input data acquisition unit 124 acquires the customer behavior pattern information, which is input data, from each vehicle 2 via the network 1a using the interface unit 18. The input data acquisition unit 124 also acquires regional information as input data. The input data acquisition unit 124 also acquires customer behavior pattern information, which is time-series data, for example, from the present time point to a time point going back a predetermined period of time. The input data acquisition unit 124 also acquires regional information, which is time-series data, for example, from a time point going back a predetermined period of time from the present time to a time point that can be acquired in the future.

[0055] The prediction unit 140 predicts the demand for hydrogen at at least one hydrogen station using the demand prediction model stored in the learned model storage unit 122. In other words, the prediction unit 140 predicts the demand for hydrogen at at least one hydrogen station using a demand prediction model that receives as input at least customer behavior pattern information and outputs a predicted demand for hydrogen.

[0056] The demand prediction unit 142 inputs the input data acquired by the input data acquisition unit 124 into the demand prediction model stored in the learned model storage unit 122. As a result, the demand prediction model outputs a predicted demand for hydrogen for each hydrogen station, as shown in Fig. 10. As a result, the demand prediction unit 142 predicts the demand for hydrogen for each hydrogen station.

[0057] In this way, the demand prediction unit 142 (prediction unit 140) is configured to predict the demand for hydrogen at at least one hydrogen station using a demand prediction model that inputs at least customer behavior pattern information and outputs predicted hydrogen demand. This allows the information processing device 10 according to the first embodiment to accurately predict the demand for hydrogen at a hydrogen station. In other words, since it is configured to predict the demand for hydrogen using customer behavior patterns, it is possible to predict the demand for hydrogen even if a customer does not visit a hydrogen station and make a reservation to fill up with hydrogen. Therefore, the information processing device 10 according to the first embodiment is able to accurately predict the demand for hydrogen.

[0058] In addition, from the feature quantities "remaining hydrogen amount," "visited hydrogen stations," and "amount filled per time" illustrated in FIG. 5, it can be said that the customer behavior pattern information, which is time-series data, indicates the timing when the customer fills the vehicle with hydrogen. In other words, the timing when the component value of the feature quantity "remaining hydrogen amount" increases and the component values ​​of the feature quantities "visited hydrogen stations" and "amount filled per time" change corresponds to the timing when the customer fills the vehicle with hydrogen. Therefore, the demand prediction unit 142 predicts the demand for hydrogen using the timing when the customer fills the vehicle with hydrogen, which is indicated in the customer behavior pattern information. Since the demand prediction unit 142 (prediction unit 140) predicts the demand for hydrogen as described above, it is possible to improve the prediction accuracy of the demand for hydrogen. In other words, the timing when the customer fills the vehicle with hydrogen often occurs at roughly the same cycle. Therefore, the demand prediction model is adjusted so as to predict that the demand will be high at the timing corresponding to that cycle, thereby making it possible to improve the prediction accuracy.

[0059] As illustrated in FIG. 5, the customer behavior pattern information includes the vehicle location. Therefore, the demand prediction unit 142 predicts the demand for hydrogen using the customer's vehicle location indicated by the customer behavior pattern information. As illustrated in FIG. 5, the customer behavior pattern information includes the remaining amount of hydrogen. Therefore, the demand prediction unit 142 predicts the demand for hydrogen using the remaining amount of hydrogen in the customer's vehicle 2 indicated by the customer behavior pattern information. The demand prediction unit 142 (prediction unit 140) predicts the demand for hydrogen as described above, so that the prediction accuracy of the demand for hydrogen can be improved. That is, for example, when the remaining amount of hydrogen in the customer's vehicle 2 becomes low enough to require refueling (a filling rate of 20% in the examples of FIGS. 6 and 8, and a filling rate of 40% in the examples of FIGS. 7 and 9), the customer is likely to visit a hydrogen station. Furthermore, the customer is likely to visit a hydrogen station close to the vehicle location at that time. Therefore, the prediction accuracy can be improved by adjusting the demand prediction model so as to predict that the demand for hydrogen stations close to the vehicle location at that time will be high at that time.

[0060] As shown in FIG. 5, the customer behavior pattern information includes reservation information. Therefore, the demand prediction unit 142 predicts the demand for hydrogen using the reservation information from customers, which is indicated in the customer behavior pattern information. Since the demand prediction unit 142 (prediction unit 140) predicts the demand for hydrogen as described above, it is possible to improve the prediction accuracy of the demand for hydrogen. In other words, there is a very high possibility that a customer will visit a hydrogen station at a time corresponding to the reservation information. Therefore, by adjusting the demand prediction model so as to predict that demand will be high at that time, it is possible to improve the prediction accuracy.

[0061] FIG. 11 is a diagram illustrating a demand forecast obtained by the demand forecasting unit 142 according to the first embodiment. FIG. 11 illustrates a demand forecast for hydrogen station A. FIG. 11 also illustrates a graph in which the horizontal axis represents time and the vertical axis represents predicted demand. Here, the demand forecast model outputs predicted demand at multiple timings (after T1, after T2, after T3, after T4, ...) as illustrated in FIG. 10. Therefore, the graph illustrated in FIG. 11 can be generated by plotting the predicted demand at these multiple timings.

[0062] In the demand forecast illustrated in FIG. 11, the demand increases at a timing corresponding to Ta. The demand decreases at a timing corresponding to Tb. The demand increases at a timing corresponding to Tc. Note that Ta, Tb, and Tc may indicate a time, a time period, or a date. Whether each timing indicates a time, a time period, or a date may depend on which timing of the demand forecast is performed. For example, if the demand forecasting model is configured to perform a demand forecast for each time period in a day, the above timing may indicate a time period. Also, if the demand forecasting model is configured to perform a demand forecast for each date in a week or a month, the above timing may indicate a date.

[0063] The supplyable amount determination unit 144 determines the amount of hydrogen that can be supplied (supplyable amount) according to the timing based on the predicted demand. Specifically, the supplyable amount determination unit 144 determines the supplyable amount so that the supplyable amount is increased at a timing when the demand is predicted to be high. On the other hand, the supplyable amount determination unit 144 determines the supplyable amount so that the supplyable amount is decreased at a timing when the demand is predicted to be low. In the example of FIG. 11, the supplyable amount determination unit 144 determines the supplyable amount at each timing for hydrogen station A such that the supplyable amount at time Ta is greater than the supplyable amount at time Tb. Similarly, the supplyable amount determination unit 144 determines the supplyable amount at each timing for hydrogen station A such that the supplyable amount at time Tc is greater than the supplyable amount at time Tb.

[0064] In this way, the supplyable amount determination unit 144 (prediction unit 140) determines the amount of hydrogen that can be supplied (supplyable amount) according to the timing based on the predicted demand, which makes it possible to stabilize the profits of the hydrogen station. In other words, by increasing the supplyable amount at times when demand is predicted to be high, it is possible to reduce opportunity losses such as the hydrogen station being unable to supply hydrogen when a customer visits the hydrogen station to fill up the vehicle 2 with hydrogen. In addition, by reducing the supplyable amount at times when demand is predicted to be low, it is possible to reduce losses due to over-preparation. This makes it possible to stabilize the profits of the hydrogen station.

[0065] Furthermore, the available supply amount determination unit 144 may determine the amount of hydrogen preparation in accordance with the timing of ordering hydrogen based on the predicted demand for hydrogen. Specifically, the available supply amount determination unit 144 determines the amount of hydrogen preparation corresponding to the demand for a period corresponding to the frequency of ordering for each hydrogen station. For example, when hydrogen is ordered once a week for hydrogen station A, the available supply amount determination unit 144 determines the amount of hydrogen preparation corresponding to the predicted demand for hydrogen for one week for hydrogen station A. For example, the amount of hydrogen preparation may be determined by adding up the predicted demand amounts at each timing at which the demand is predicted within one week. In this way, the available supply amount determination unit 144 (prediction unit 140) determines the amount of hydrogen preparation in accordance with the timing of ordering hydrogen based on the predicted demand for hydrogen, thereby further suppressing the opportunity loss or overpreparation described above.

[0066] Furthermore, the supplyable amount determination unit 144 may determine the timing for preparing high-pressure hydrogen gas based on the predicted demand. Specifically, when hydrogen demand is predicted for each time period of a day, the supplyable amount determination unit 144 determines the timing for preparing high-pressure gas so that high-pressure gas (high-pressure hydrogen) is prepared a predetermined time (e.g., one hour) before the time period when demand is high. The "predetermined time" may be set appropriately according to the time required to pressurize hydrogen. Even if hydrogen is prepared at a hydrogen station, hydrogen cannot be supplied to the vehicle 2 unless the hydrogen is pressurized. Therefore, by determining the timing for preparing high-pressure hydrogen gas based on the predicted demand, it is possible to suppress opportunity losses such as not being able to supply hydrogen to the vehicle 2 when a customer visits the hydrogen station.

[0067] The notification unit 150 notifies the customer of when and from which hydrogen station hydrogen can be supplied according to the predicted demand. Using the interface unit 18, the notification unit 150 transmits a notification (supply availability notification) to the customer's device via the network 1a indicating the hydrogen stations that can supply hydrogen and the timing (time period) when hydrogen can be supplied from the hydrogen stations.

[0068] Specifically, the notification unit 150 determines the timing when it is predicted that there will be a demand for hydrogen for each hydrogen station. For example, the notification unit 150 determines the timing when it is predicted that the demand for hydrogen will be equal to or greater than a predetermined value for each hydrogen station. Then, the notification unit 150 sets the timing when it is predicted that there will be a demand for hydrogen for each hydrogen station as the timing when hydrogen can be supplied. Then, the notification unit 150 generates a supply availability notification according to this hydrogen station and timing. The notification unit 150 transmits the generated supply availability notification.

[0069] For example, the notification unit 150 may transmit a supply availability notification to the customer's vehicle 2. This causes the supply availability notification to be displayed on the vehicle 2. In this case, the notification unit 150 may display the supply availability notification using a navigation system mounted on the vehicle 2. For example, when a hydrogen station capable of supplying hydrogen is displayed on the screen of the navigation system, the notification unit 150 may display the timing when hydrogen can be supplied at that hydrogen station.

[0070] Also, for example, the notification unit 150 may transmit a supply availability notification to a terminal (such as a smartphone) owned by the customer. In this case, the notification unit 150 may perform the same process for the navigation system of the vehicle 2 as described above for the navigation system that can be realized on the customer's terminal. Alternatively, the notification unit 150 may cause the terminal to display a list that associates hydrogen stations with the timing when hydrogen can be supplied.

[0071] Alternatively, the notification unit 150 may display a supply availability notification on the website of the hydrogen station. In this case, the notification unit 150 may display a map on the website and display the times when hydrogen can be supplied at the hydrogen stations displayed on the map. Alternatively, the notification unit 150 may display on the website a list that associates hydrogen stations with the times when hydrogen can be supplied.

[0072] The notification unit 150 notifies the customer of when and at which hydrogen station hydrogen can be supplied according to the predicted demand, thereby improving convenience for the customer. Furthermore, the hydrogen station also increases the possibility of being able to supply the prepared hydrogen. Therefore, by notifying the customer as described above, it becomes possible to more reliably adjust the demand and supply of hydrogen.

[0073] Furthermore, the notification unit 150 may also notify the price of hydrogen when notifying the hydrogen stations that can supply hydrogen and the timing when hydrogen can be supplied. This allows the customer to simultaneously know the price of hydrogen and the timing when hydrogen can be filled into the vehicle 2, improving convenience for the customer.

[0074] The continuing learning processing unit 160 performs a process for continuing the learning of the demand prediction model. Specifically, the continuing learning processing unit 160 acquires an actual value (actual demand amount) corresponding to the predicted value of the demand. Then, the continuing learning processing unit 160 performs a continuing process of the learning by the learning unit 100 (continuing learning process) according to the difference between the predicted value and the actual value of the demand. More specifically, the continuing learning processing unit 160 performs the continuing learning process when the difference between the predicted value and the actual value of the demand is equal to or greater than a predetermined threshold. That is, when the difference between the predicted value and the actual value of the demand becomes large, it is a case where the accuracy of the demand forecast by the demand prediction model may have decreased. Therefore, in this case, it is preferable to re-learn the demand prediction model. The above threshold can be appropriately determined according to the accuracy of the demand required.

[0075] The continuous learning process may be performed, for example, as follows. The continuous learning processing unit 160 acquires, as input data, customer behavior pattern information (and area information) up to the time when the difference between the predicted value and the actual value of the demand becomes equal to or greater than a predetermined threshold. The continuous learning processing unit 160 also acquires, as supervised data, the actual value of the demand obtained up to that time. Here, since more time has passed at this time than in the learning stage of the demand forecasting model, the amount of input data acquired at this time is greater than the amount of input data used in the learning stage of the demand forecasting model. The continuous learning processing unit 160 then performs processing to re-learn the demand forecasting model using the acquired set of input data and supervised data as teacher data. In this way, the learning unit 100 re-learns the demand forecasting model.

[0076] The learning continuation process does not have to be executed immediately when the difference between the demand forecast value and the actual value becomes equal to or greater than the threshold value. For example, the learning continuation process may be executed when the difference between the demand forecast value and the actual value becomes equal to or greater than the threshold value a predetermined number of times or more.

[0077] In this way, the information processing device 10 may continue to perform learning for the machine learning algorithm according to the difference between the demand predicted by the prediction unit 140 and the actual demand. With such a configuration, the demand prediction model is adjusted according to the actual operation, so that it is possible to further improve the prediction accuracy of the demand.

[0078] 12 and 13 are flowcharts showing an information processing method executed by the information processing device 10 according to the first embodiment. The flowcharts shown in Fig. 12 and 13 correspond to a demand prediction method for predicting demand for hydrogen.

[0079] 12 shows the process in the learning stage of the demand forecasting model. As described above, the teacher data acquisition unit 102 acquires teacher data, which is a set of input data and correct answer data (step S102). As described above, the demand forecasting model learning unit 104 performs a learning process of the demand forecasting model using the acquired teacher data (step S104).

[0080] 13 shows the process in the operation stage of the demand forecasting model. As described above, the input data acquisition unit 124 acquires input data (step S112). As described above, the input data includes at least a customer behavior pattern (customer behavior pattern information).

[0081] As described above, the demand prediction unit 142 inputs input data into the demand prediction model, which is a trained model, and obtains the predicted hydrogen demand for each hydrogen station (step S114). As described above, the supplyable amount determination unit 144 determines the supplyable amount based on the predicted demand (step S116). As described above, the notification unit 150 notifies the customer of the hydrogen stations that can supply hydrogen and the time periods (timing) (step S118).

[0082] The continuing learning processing unit 160 judges whether the difference between the predicted value and the actual value of the demand is equal to or greater than a predetermined threshold value (step S120). If it is judged that the difference between the predicted value and the actual value of the demand is equal to or greater than the threshold value (YES in S120), the continuing learning processing unit 160 performs the continuing learning process as described above (step S122). On the other hand, if it is not judged that the difference between the predicted value and the actual value of the demand is equal to or greater than the threshold value (NO in S120), the process of S122 is not performed. Then, the processes of S112 to S122 may be repeated.

[0083] (Embodiment 2) Next, a second embodiment will be described. The second embodiment differs from the first embodiment in that the hydrogen price at the hydrogen station is determined according to the demand for hydrogen. The configuration of the information processing system 1 according to the second embodiment is substantially similar to the configuration of the information processing system 1 according to the first embodiment shown in FIG. 1, and therefore a description thereof will be omitted. Also, the hardware configuration of the information processing device 10 according to the second embodiment is substantially similar to the hardware configuration of the information processing device 10 according to the first embodiment shown in FIG. 2, and therefore a description thereof will be omitted.

[0084] Fig. 14 is a block diagram showing a configuration of the information processing device 10 according to the second embodiment. The information processing device 10 according to the second embodiment has substantially the same components as those of the information processing device 10 according to the first embodiment shown in Fig. 3. Furthermore, the information processing device 10 according to the second embodiment has a price determination unit 210 (determination unit) and a notification unit 250. In the information processing device 10 according to the second embodiment, the functions of the components of the information processing device 10 shown in Fig. 3 are substantially similar to those of the first embodiment unless otherwise specified, and therefore descriptions thereof will be omitted as appropriate.

[0085] As described above, the prediction unit 140 predicts the time-series demand for hydrogen at the hydrogen station. In other words, the prediction unit 140 predicts the time-series change in the demand for hydrogen at the hydrogen station. The prediction unit 140 can also predict the time-series change in the demand for hydrogen at the hydrogen station at least in a day after a predetermined time (e.g., one week or one month).

[0086] The price determination unit 210 determines the price of hydrogen (hydrogen price) for each hydrogen station based on the demand for hydrogen predicted by the prediction unit 140. That is, the price determination unit 210 determines the hydrogen price for each hydrogen station according to the predicted time-series demand for hydrogen. Specifically, the price determination unit 210 determines the hydrogen price at the hydrogen station such that the hydrogen price is high during the time period when the predicted demand for hydrogen is large. Also, the price determination unit 210 determines the hydrogen price after a predetermined time from the present (for example, one week later or one month later).

[0087] FIG. 15 is a diagram for explaining an example of the price determination method by the price determination unit 210 according to Embodiment 2. FIG. 15 illustrates a method for determining the hydrogen price for the demand prediction exemplified in FIG. 11. FIG. 15 illustrates the relationship between the predicted demand for hydrogen and the hydrogen price at hydrogen station A. In FIG. 15, the predicted demand is indicated by a thin broken line. Also, in FIG. 15, the hydrogen price is indicated by a thick solid line. Also, for the predicted demand, two threshold values ThA and ThB are set in advance. Here, ThA < ThB. Also, three hydrogen prices PrA, PrB, and PrC are set in advance. Here, PrA < PrB < PrC.

[0088] The price determination unit 210 determines the hydrogen price to be the lowest hydrogen price PrA during the time period when the predicted demand is below the threshold value ThA. Also, the price determination unit 210 determines the hydrogen price to be the second lowest PrB during the time period when the predicted demand is equal to or greater than the threshold value ThA and equal to or less than the threshold value ThB. Also, the price determination unit 210 determines the hydrogen price to be the highest hydrogen price PrC during the time period when the predicted demand exceeds the threshold value ThB. Thereby, the hydrogen price becomes high at the timings corresponding to Ta and Tc where the predicted demand is large, and the hydrogen price becomes low at the timing corresponding to Tb where the predicted demand is small.

[0089] In the example of FIG. 15, the number of thresholds for the predicted demand is two, and the number of hydrogen prices is three, but the present invention is not limited to this configuration. The number of hydrogen prices is arbitrary, and the number of thresholds for the predicted demand may be appropriately set according to the number of hydrogen prices. In addition, in the example of FIG. 15, the hydrogen price changes stepwise and discontinuously according to the predicted demand, but the present invention is not limited to this configuration. The price determination unit 210 may determine the hydrogen price so that the hydrogen price changes continuously according to the predicted demand. In other words, the price determination unit 210 may determine the hydrogen price so that the hydrogen price increases continuously as the predicted demand increases, and the hydrogen price decreases continuously as the predicted demand decreases. In this case, the relationship between the predicted demand and the hydrogen price may be determined by a predetermined function.

[0090] Furthermore, the price determination unit 210 may update the method of determining the hydrogen price each time so as to improve (maximize) profits. For example, the price determination unit 210 may update the method of determining the hydrogen price according to the predicted demand volume by machine learning such as a reinforcement learning algorithm so as to improve (maximize) profits. For example, the price determination unit 210 may update the set value of the hydrogen price (PrA, etc.) and the threshold value of the predicted demand volume (ThA, etc.) by a reinforcement learning algorithm or the like. Furthermore, the price determination unit 210 may update a function indicating the relationship between the predicted demand volume and the hydrogen price by a reinforcement learning algorithm or the like. This also applies to other embodiments described later.

[0091] With such a configuration, the information processing device 10 according to the second embodiment is able to adjust the price of hydrogen at a hydrogen station in accordance with the demand for hydrogen. Generally, when the price of a product rises, the demand for that product tends to decrease, and when the price of a product falls, the demand for that product tends to increase. Therefore, by adjusting the price of hydrogen in accordance with the demand for hydrogen, it is possible to level out the demand for hydrogen at a hydrogen station. In other words, it is possible to adjust the demand for hydrogen at a hydrogen station. This makes it possible to reduce the difference between the amount of work in the off-season and the amount of work in the busy season. This makes it possible to operate the hydrogen station efficiently. This makes it possible to efficiently improve the profits (merits) enjoyed by the hydrogen station.

[0092] FIG. 16 is a diagram illustrating actual hydrogen demand when the hydrogen price is determined by the price determination unit 210 according to the second embodiment. In FIG. 16, the predicted demand is indicated by a thin dashed line. Also, in FIG. 16, the actual demand is indicated by a thick solid line. The predicted demand corresponds to that shown in FIG. 11 and FIG. 15. As illustrated in FIG. 15, the hydrogen price is high when the predicted demand is high, and the hydrogen price is low when the predicted demand is low. As a result, as illustrated in FIG. 16, the fluctuation in the actual demand is smaller than the fluctuation in the predicted demand. In other words, the hydrogen demand at the hydrogen station is leveled out by the hydrogen price being determined by the price determination unit 210.

[0093] Furthermore, the information processing device 10 according to the second embodiment can increase the hydrogen price when the predicted demand at the hydrogen station is high, thereby improving the profits of the hydrogen station. On the other hand, by lowering the hydrogen price when the predicted demand at the hydrogen station is low, it is possible to increase the demand at that time. Therefore, it is possible to efficiently improve the profits (benefits) enjoyed by the hydrogen station.

[0094] The notification unit 250 notifies the customer of the hydrogen price determined by the price determination unit 210. The notification unit 250 uses the interface unit 18 to transmit a notification (hydrogen price notification) indicating the hydrogen price at the hydrogen station to the customer's device via the network 1a.

[0095] FIG. 17 is a diagram illustrating a hydrogen price notification according to the second embodiment. FIG. 17 illustrates a hydrogen price notification for hydrogen station A. The hydrogen price notification illustrated in FIG. 17 includes a "normal hydrogen price," a "today's hydrogen price," and a "tomorrow's hydrogen price." In this way, by notifying the hydrogen prices for multiple days, the customer can decide which day to visit the hydrogen station based on the hydrogen price and their own time availability. In particular, the customer can visit the hydrogen station on a day when the hydrogen price is low. Therefore, it is possible to improve convenience for the customer.

[0096] Furthermore, the notification unit 250 may, for example, transmit a hydrogen price notification to the customer's vehicle 2. As a result, the hydrogen price notification is displayed on the vehicle 2. In this case, the notification unit 250 may display the hydrogen price using a navigation system installed in the vehicle 2. For example, the notification unit 250 may display the hydrogen price for the hydrogen station displayed on the screen of the navigation system.

[0097] Also, for example, the notification unit 250 may transmit the hydrogen price notification to a terminal (such as a smartphone) owned by the customer. In this case, the notification unit 250 may perform the same process for the navigation system of the vehicle 2 as described above for the navigation system that can be realized on the customer's terminal.

[0098] Alternatively, the notification unit 250 may display the hydrogen price notification on the website of the hydrogen station. In this case, the notification unit 250 may display a map on the website and display the hydrogen prices of the hydrogen stations displayed on the map. Alternatively, the notification unit 250 may display a list associating hydrogen stations with hydrogen prices on the website.

[0099] Also, for example, the notification unit 250 may notify a customer of the hydrogen price of a hydrogen station that the customer visits frequently. In this case, the notification unit 250 may use customer behavior pattern information to determine the frequency of the customer's visit. Also, for example, the notification unit 250 may notify a customer of the hydrogen price of a hydrogen station that is registered in advance for each customer. For example, if customer #1 has registered hydrogen station A in the system, the notification unit 250 may notify customer #1 of the hydrogen price of hydrogen station A.

[0100] The notification unit 250 may also notify the customer of the hydrogen price at the necessary timing for each customer. For example, when the hydrogen price at a hydrogen station near the customer is changed, the notification unit 250 may notify the customer of the hydrogen price at that hydrogen station. For example, when the hydrogen price at a hydrogen station previously registered by the customer is changed, the notification unit 250 may notify the customer of the hydrogen price at that hydrogen station. The notification unit 250 may also notify a customer of a vehicle 2 with a low amount of hydrogen remaining of the hydrogen price at the hydrogen station.

[0101] The notification unit 250 notifies customers of the determined hydrogen price, thereby improving convenience for customers. In particular, by having customers visit hydrogen stations when the hydrogen price is low, the cost of purchasing hydrogen can be reduced. Furthermore, the notification unit 250 notifies each customer of the hydrogen price at the necessary timing, thereby further improving convenience for customers. Furthermore, it becomes possible for hydrogen stations to more reliably equalize the demand for hydrogen.

[0102] FIG. 18 is a flowchart showing an information processing method executed by the information processing device 10 according to the second embodiment. The flowchart shown in FIG. 18 corresponds to a price determination method for determining the hydrogen price of a hydrogen station. As in S112 of FIG. 13, the input data acquisition unit 124 acquires input data (step S212). As in S114 of FIG. 13, the demand prediction unit 142 inputs the input data into the demand prediction model, which is a trained model, to obtain a predicted hydrogen demand amount for each hydrogen station (step S214). As in S114 of FIG. 13, the supplyable amount determination unit 144 determines a supplyable amount based on the predicted demand (step S216).

[0103] As described above, the price determination unit 210 determines the hydrogen price of the hydrogen station according to the predicted demand for hydrogen (step S218). Specifically, the price determination unit 210 determines the hydrogen price so that the hydrogen price is higher when the predicted demand is high. As described above, the notification unit 250 notifies the customer of the hydrogen price of the hydrogen station (step S220).

[0104] The continuing learning processing unit 160 judges whether or not the difference between the predicted value and the actual value of the demand is equal to or greater than a predetermined threshold value, as in S120 of FIG. 13 (step S221). If it is judged that the difference between the predicted value and the actual value of the demand is equal to or greater than the threshold value (YES in S221), the continuing learning processing unit 160 performs continuing learning processing, as in S122 of FIG. 13 (step S222). On the other hand, if it is not judged that the difference between the predicted value and the actual value of the demand is equal to or greater than the threshold value (NO in S221), the processing of S222 is not performed. In this way, it becomes possible to predict the demand for hydrogen that has changed due to the determined hydrogen price, and therefore it is possible to further improve the prediction accuracy of the demand for hydrogen. Then, the processing of S212 to S222 may be repeated.

[0105] (Embodiment 3) Next, a third embodiment will be described. The third embodiment differs from the other embodiments in that a demand forecasting model is used in which the hydrogen price at the hydrogen station is input when determining the hydrogen price at the hydrogen station. Note that the configuration of the information processing system 1 according to the third embodiment is substantially similar to the configuration of the information processing system 1 according to the first embodiment shown in FIG. 1, and therefore a description thereof will be omitted. Also, the hardware configuration of the information processing device 10 according to the third embodiment is substantially similar to the hardware configuration of the information processing device 10 according to the first embodiment shown in FIG. 2, and therefore a description thereof will be omitted.

[0106] 19 is a block diagram showing a configuration of an information processing device 10 according to the third embodiment. The information processing device 10 according to the third embodiment includes a learning unit 100, a trained model storage unit 122, an input data acquisition unit 124, a prediction unit 140, and a learning continuation processing unit 160, similarly to the first embodiment and the like. Here, the prediction unit 140 includes a demand prediction unit 142, but does not necessarily include a supplyable amount determination unit 144. The information processing device 10 according to the third embodiment also includes a supplyable amount acquisition unit 302, a price determination unit 310 (determination unit), and a notification unit 250.

[0107] 3 are substantially similar to those in the first embodiment unless otherwise specified, and therefore the description thereof will be omitted as appropriate. Also, the function of the notification unit 250 is substantially similar to that in the second embodiment unless otherwise specified, and therefore the description thereof will be omitted as appropriate.

[0108] FIG. 20 is a diagram illustrating an example of feature quantities in input data according to the third embodiment. The customer behavior pattern information may be substantially the same as that according to the first embodiment. Regarding the feature quantities included in the area information, the component x n+5indicates the hydrogen price at the hydrogen station in the corresponding region. In the input data, the hydrogen price can be set arbitrarily. Thus, in the third embodiment, the demand prediction model takes the hydrogen price at the hydrogen station as input. Then, the learning unit 100 learns a demand prediction model that includes the hydrogen price at the hydrogen station as input data. Therefore, the demand prediction model according to the third embodiment can predict an under-demand when the hydrogen price in the input data is high, and predict an over-demand when the hydrogen price is low.

[0109] The available supply amount acquisition unit 302 acquires the available supply amount. That is, in the third embodiment, the available supply amount is not determined according to the predicted demand amount but is set in advance. That is, the available supply amount cannot always be determined according to the predicted demand amount. For example, the available supply amount may be set in advance according to an agreement between a hydrogen station and a wholesaler that provides hydrogen to the hydrogen station.

[0110] The price determination unit 310 determines the hydrogen price according to the hydrogen price input to the demand prediction model and the predicted demand obtained using the demand prediction model. At this time, the price determination unit 310 determines the hydrogen price so as to improve (maximize) the profit at the hydrogen station. The price determination unit 310 also determines the hydrogen price after a predetermined time from the present (for example, one week or one month). Details will be described later.

[0111] FIG. 21 is a flowchart showing an information processing method executed by the information processing device 10 according to the third embodiment. The flowchart shown in FIG. 21 corresponds to a price determination method for determining the hydrogen price of a hydrogen station. Moreover, the flowchart shown in FIG. 21 can be executed for each hydrogen station. The supplyable amount acquisition unit 302 acquires the supplyable amount (step S302). As in S112 of FIG. 13, the input data acquisition unit 124 acquires input data (step S310). Here, the hydrogen price does not need to be set in the input data acquired in S310. Then, the price determination unit 310 sets the hydrogen price in the input data (step S312).

[0112] As in S114 of FIG. 13, the demand forecasting unit 142 inputs input data into the demand forecasting model, which is a trained model, and obtains the predicted hydrogen demand (predicted demand) for each hydrogen station (step S314). The price determination unit 310 determines whether the predicted demand obtained in S314 exceeds the available supply amount obtained in S302 (step S316). If the predicted demand amount exceeds the available supply amount (YES in S316), the predicted demand amount is excessive compared to the available supply amount because the hydrogen price was set too low. Therefore, in this case, the price determination unit 310 sets the hydrogen price higher than the hydrogen price set in S312 (step S318). Then, the processes of S314 and S316 are executed again. Note that any method may be used to determine how much higher the hydrogen price is set. For example, the hydrogen price may be set higher by a predetermined price (e.g., 5 yen / kg) than the hydrogen price set in S312.

[0113] On the other hand, if the predicted demand is equal to or less than the available supply (NO in S316), the price determination unit 310 calculates the sales amount at that time (step S320). Specifically, the price determination unit 310 multiplies the set hydrogen price by the predicted demand to calculate the sales amount of hydrogen for the corresponding hydrogen station.

[0114] Next, the price determination unit 310 changes the hydrogen price and performs the processes of S314 to S320 (step S322). Specifically, the price determination unit 310 repeats the processes of S314 to S320 until the processes for all hydrogen prices that can be set are completed. In this way, the sales amount is calculated for each hydrogen price that can be set. Note that, when changing the hydrogen price in the process of S322, it is also possible not to change it to the hydrogen price when the process of S316 is judged as YES.

[0115] Next, the price determination unit 310 determines the hydrogen price according to the calculated sales amount (step S330). Specifically, the price determination unit 310 may determine the hydrogen price at the hydrogen station to be the hydrogen price that maximizes the sales amount. Alternatively, the price determination unit 310 may determine the hydrogen price at the hydrogen station to be the hydrogen price that maximizes the value obtained by subtracting the purchase price (cost) of the supplyable amount from the sales amount. In this way, the hydrogen price that maximizes the revenue (profit) of the hydrogen station is determined.

[0116] As described above, the notification unit 250 notifies the customer of the hydrogen price of the hydrogen station (step S332). The information processing device 10 according to the third embodiment may perform a learning continuation process (S122 in FIG. 13). In the above example, the hydrogen price is set high when the predicted demand exceeds the available supply (YES in S316, S318), but this is not limited to the above configuration. The processes of S316 and S318 may be omitted. In this case, the process of S302 may be omitted.

[0117] As described above, the information processing device 10 according to the third embodiment is configured to determine the hydrogen price according to the hydrogen price input to the demand forecasting model and the predicted demand obtained using the demand forecasting model. This allows the information processing device 10 according to the third embodiment to determine the hydrogen price that maximizes the profit at the hydrogen station. This makes it possible to more reliably improve profits.

[0118] (Embodiment 4) Next, a fourth embodiment will be described. The fourth embodiment differs from the other embodiments in that the hydrogen price at a hydrogen station is determined according to the predicted demand and the available supply. The configuration of the information processing system 1 according to the fourth embodiment is substantially similar to the configuration of the information processing system 1 according to the first embodiment shown in FIG. 1, and therefore a description thereof will be omitted. The hardware configuration of the information processing device 10 according to the fourth embodiment is substantially similar to the hardware configuration of the information processing device 10 according to the first embodiment shown in FIG. 2, and therefore a description thereof will be omitted.

[0119] 22 is a block diagram showing a configuration of an information processing device 10 according to the fourth embodiment. The information processing device 10 according to the fourth embodiment includes a learning unit 100, a trained model storage unit 122, an input data acquisition unit 124, a prediction unit 140, and a learning continuation processing unit 160, similarly to the third embodiment. Here, the prediction unit 140 includes a demand prediction unit 142, but does not necessarily include the available supply amount determination unit 144. In addition, the information processing device 10 according to the fourth embodiment includes a available supply amount acquisition unit 302, a price determination unit 410 (determination unit), and a notification unit 250.

[0120] 3 are substantially similar to those in the first embodiment unless otherwise specified, and therefore descriptions thereof will be omitted where appropriate. In addition, the functions of the supplyable amount acquisition unit 302 and the notification unit 250 are substantially similar to those in the second and third embodiments unless otherwise specified, and therefore descriptions thereof will be omitted where appropriate.

[0121] The price determination unit 410 determines the hydrogen price based on the ratio of the predicted demand to the supplyable amount at a certain time of the hydrogen station. Specifically, the price determination unit 410 determines the hydrogen price based on a hydrogen demand ratio (first ratio), which is the ratio of the predicted demand to the supplyable amount. With this configuration, as will be described later, it becomes possible to adjust the demand for hydrogen and the hydrogen price according to the hydrogen station's objectives (profit improvement, demand promotion or demand suppression, etc.) based on the predicted demand and supplyable amount. In other words, it becomes possible to more reliably improve the profits of the hydrogen station (profit improvement, demand promotion or demand suppression, etc.) according to the predicted demand and supplyable amount.

[0122] More specifically, when the hydrogen demand ratio is equal to or greater than a first threshold and equal to or less than a second threshold greater than the first threshold, the price determination unit 410 determines the price of hydrogen such that the higher the hydrogen demand ratio, the higher the price. At this time, the price determination unit 410 determines the hydrogen price between the first price and the second price such that the higher the hydrogen demand ratio, the higher the price. Details will be described later. The first and second thresholds, and the first and second prices are predetermined values. These predetermined values ​​may be updated as appropriate by machine learning such as a reinforcement learning algorithm.

[0123] Furthermore, the price determination unit 410 may determine the hydrogen price to be lower than the first price when the hydrogen demand ratio (first ratio) is below the first threshold. Furthermore, the price determination unit 410 may determine the hydrogen price to be higher than the second price when the hydrogen demand ratio (first ratio) is above the second threshold. Details will be described later.

[0124] FIG. 23 is a flowchart showing an information processing method executed by the information processing device 10 according to the fourth embodiment. The flowchart shown in FIG. 23 corresponds to a price determination method for determining the hydrogen price of a hydrogen station. The flowchart shown in FIG. 23 can be executed for each hydrogen station. The available supply amount acquisition unit 302 acquires the available supply amount y (step S402). As in S112 of FIG. 13, the input data acquisition unit 124 acquires input data (step S412). As in S114 of FIG. 13, the demand forecasting unit 142 inputs the input data into the demand forecasting model, which is a trained model, and acquires the predicted hydrogen demand amount x for each hydrogen station (step S414).

[0125] The price determination unit 410 calculates the hydrogen demand ratio (first ratio) (step S418). Specifically, the price determination unit 410 calculates the hydrogen demand ratio z by dividing the hydrogen demand amount x by the available supply amount y. In other words, z=x / y.

[0126] Then, the price determination unit 410 determines whether the hydrogen demand ratio z is equal to or greater than a threshold value Th1 (first threshold value) and equal to or less than a threshold value Th2 (second threshold value) (step S420). The threshold value Th1 is a predetermined value less than 1. For example, Th1=0.15. The threshold value Th2 is a predetermined value greater than the threshold value Th1. For example, Th2=0.9. In other words, the price determination unit 410 determines whether the ratio of the hydrogen demand amount x to the supplyable amount y (hydrogen demand ratio) is equal to or greater than 15% and equal to or less than 90%.

[0127] If the hydrogen demand ratio z is equal to or greater than the threshold value Th1 and equal to or less than the threshold value Th2 (YES in S420), the price determination unit 410 determines the hydrogen price according to the value of the hydrogen demand ratio z (step S422). Specifically, the price determination unit 410 determines the price of hydrogen such that the higher the hydrogen demand ratio z, the higher the price.

[0128] FIG. 24 is a diagram for explaining a price determination method by a price determination unit 410 according to Embodiment 4. FIG. 24 illustrates the relationship between the hydrogen demand ratio z and the hydrogen price. In the example of FIG. 24, when the hydrogen demand ratio z is equal to or greater than a threshold Th1 and equal to or less than a threshold Th2, the price determination unit 410 determines the hydrogen price between a price Pr1 (first price) and a price Pr2 (second price) such that the higher the hydrogen demand ratio z, the higher the price. Note that Pr1 and Pr2 are predetermined values, and Pr1 < Pr2. For example, Pr1 = 1000 yen / kg and Pr2 = 1200 yen / kg. In the example of FIG. 24, the price determination unit 410 determines the hydrogen price to be 1000 yen / kg when the hydrogen demand ratio z = 0.15. Also, the price determination unit 410 determines the hydrogen price to be 1200 yen / kg when the hydrogen demand ratio z = 0.9. By thus determining the hydrogen price such that the higher the hydrogen demand ratio z, the higher the price, it is possible to improve the profit according to the predicted demand volume and the supplyable volume.

[0129] Note that in the example of FIG. 24, the relationship between the hydrogen demand ratio z and the hydrogen price is linear, but the relationship between the hydrogen demand ratio z and the hydrogen price does not have to be linear. The hydrogen price may increase discontinuously in steps as the hydrogen demand ratio z increases. Also, the hydrogen price may increase exponentially as the hydrogen demand ratio z increases. Also, the hydrogen price may increase logarithmically as the hydrogen demand ratio z increases.

[0130] On the other hand, when the hydrogen demand ratio z is not equal to or greater than the threshold Th1 and equal to or less than the threshold Th2 (NO in S420), the price determination unit 410 determines whether the hydrogen demand ratio z is less than the threshold Th1 (first threshold) (step S424). That is, the price determination unit 410 determines whether the ratio of the hydrogen demand volume x to the supplyable volume y (hydrogen demand ratio) is less than 15%.

[0131] If the hydrogen demand ratio z is below the threshold value Th1 (YES in S424), it can be said that the predicted demand x is too small compared to the available supply y. In other words, if things continue as they are, there is a risk that hydrogen will be oversupplied and that excess hydrogen will remain unsold. Therefore, in this case, in order to promote demand, the price determination unit 410 sets the hydrogen price low (step S426). Specifically, the price determination unit 410 determines the hydrogen price to be lower than Pr1. For example, the price determination unit 410 determines the hydrogen price to be 700 yen / kg. This makes it possible to promote demand for hydrogen when the predicted demand x is too small compared to the available supply y.

[0132] On the other hand, when the hydrogen demand ratio z is not below the threshold value Th1 (NO in S424), the hydrogen demand ratio z exceeds the threshold value Th2. In other words, the ratio of the hydrogen demand amount x to the supplyable amount y (hydrogen demand ratio) exceeds 90%. In this case, it can be said that the predicted demand amount x is too high compared to the supplyable amount y. In other words, if this continues, there is a risk of a supply shortage. Therefore, in this case, in order to suppress the demand, the price determination unit 410 sets the hydrogen price high (step S428). Specifically, the price determination unit 410 determines the hydrogen price to be higher than Pr2. For example, the price determination unit 410 determines the hydrogen price to be 1,500 yen / kg. This makes it possible to suppress the demand for hydrogen when the predicted demand amount x is excessively large compared to the supplyable amount y.

[0133] As described above, the notification unit 250 notifies the customer of the hydrogen price of the hydrogen station (step S432). Note that the information processing device 10 according to the fourth embodiment may perform a learning continuation process (S122 in FIG. 13).

[0134] (Modification) The present invention is not limited to the above-described embodiment, and can be modified as appropriate without departing from the spirit of the present invention. For example, the order of the steps in the above-described flowchart can be changed as appropriate. Furthermore, one or more steps in the above-described flowchart can be omitted as appropriate. For example, the processes of S216 and S220 to S222 in FIG. 18 may be omitted. In FIG. 21 and FIG. 23, the processes of S332 and S432 may be omitted, respectively.

[0135] The program includes a set of instructions (or software code) that, when loaded into a computer, causes the computer to perform one or more functions described in the embodiments. The program may be stored in a non-transitory computer-readable medium or a tangible storage medium. By way of example and not limitation, the computer-readable medium or tangible storage medium includes random-access memory (RAM), read-only memory (ROM), flash memory, solid-state drive (SSD) or other memory technology, CD-ROM, digital versatile disk (DVD), Blu-ray® disk or other optical disk storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage device. The program may be transmitted on a transitory computer-readable medium or communication medium. By way of example and not limitation, the transitory computer-readable medium or communication medium includes electrical, optical, acoustic, or other forms of propagated signals. [Explanation of symbols]

[0136] 1. Information Processing Systems 2 Vehicles 10. Information processing device 100 Learning Department 102 Teacher Data Acquisition Department 104 Demand forecast model learning section 122 Trained model storage 124 Input data acquisition unit 140 Prediction Department 142 Demand Forecasting Department 144 Available supply amount determination section 150 Notification Department 160 Learning Continuation Processing Unit 210,310,410 Pricing Department 250 Notification Department 302 Available supply amount acquisition unit

Claims

1. a prediction unit that predicts demand for hydrogen at hydrogen stations using a demand prediction model, which is a trained model generated in advance by machine learning, and which inputs customer behavior pattern information that indicates the behavior patterns of each of a plurality of customers, the customer behavior pattern information including at least time series data on the remaining amount of hydrogen in the customer's vehicle, time series data on the amount of hydrogen filled in one filling of the customer's vehicle, the filling frequency with which hydrogen is filled into the customer's vehicle, and seasonal fluctuations in the customer's behavior pattern, and outputs a predicted demand for hydrogen; A determination unit that determines a price of hydrogen at the hydrogen station based on the predicted demand for hydrogen; An information processing device having the above configuration.

2. The determination unit determines the price of hydrogen based on a ratio of a predicted demand amount, which is a predicted demand for hydrogen, to a supplyable amount, which is an amount of hydrogen that can be supplied, at a certain time point of the hydrogen station. The information processing device according to claim 1 .

3. the determination unit determines the price of hydrogen between a predetermined first price and a predetermined second price when a first ratio, which is a ratio of the predicted demand amount to the supplyable amount, is equal to or greater than a predetermined first threshold value and is equal to or less than a predetermined second threshold value that is greater than the first threshold value, such that the price increases as the first ratio increases, within a range from a predetermined first price to a predetermined second price. The information processing device according to claim 2 .

4. The determination unit determines the price of hydrogen to be lower than the first price when the first ratio is lower than the first threshold value. The information processing device according to claim 3 .

5. The determination unit determines the price of hydrogen to be higher than the second price when the first ratio exceeds the second threshold value.

5. The information processing device according to claim 3.

6. The demand forecast model receives the customer behavior pattern information and the price of hydrogen at the hydrogen station as inputs, The determination unit determines a price of hydrogen according to the price of hydrogen input to the demand forecast model and a predicted demand amount, which is the demand for hydrogen obtained using the demand forecast model. The information processing device according to claim 1 .

7. a notification unit that notifies customers of the determined hydrogen price; The information processing device according to claim 1 , further comprising:

8. A learning unit that performs machine learning on the demand forecasting model; and The learning unit continues to learn the demand forecasting model according to a difference between the demand predicted by the prediction unit and an actual demand. The information processing device according to claim 1 .

9. predicting the demand for hydrogen at a hydrogen station using a demand prediction model which is a trained model generated in advance by machine learning, the demand prediction model receiving customer behavior pattern information which indicates the behavior patterns of each of a plurality of customers, the customer behavior pattern information including at least time series data on the remaining hydrogen in the customer's vehicle, time series data on the amount of hydrogen filled into the customer's vehicle in one filling, the frequency with which hydrogen is filled into the customer's vehicle, and seasonal fluctuations in the customer's behavior pattern, and outputting a predicted demand for hydrogen; determining a price of hydrogen at the hydrogen station based on the predicted demand for hydrogen; A computer-implemented information processing method.

10. a step of predicting the demand for hydrogen at a hydrogen station using a demand prediction model, which is a trained model generated in advance by machine learning, and which inputs customer behavior pattern information indicating the behavior patterns of each of a plurality of customers, the customer behavior pattern information including at least time series data on the remaining amount of hydrogen in the customer's vehicle, time series data on the amount of hydrogen filled in one filling of the customer's vehicle, the filling frequency with which hydrogen is filled in the customer's vehicle, and seasonal fluctuations in the customer's behavior pattern, and outputs a predicted demand for hydrogen; determining a price of hydrogen at the hydrogen station based on the predicted demand for hydrogen; A program that causes a computer to execute the following.

Images