Electric vehicle power consumption forecasting analysis and improvement suggestion system using LLM

Abstract

In predictive analysis and improvement suggestions for electric vehicle power consumption, this system predicts power consumption with greater accuracy for a specified period in the near future and presents drivers with measures to improve power consumption. [Solution] The LLM-based power consumption forecast analysis and improvement suggestion system for electric vehicles (100) includes an estimation unit (202) that estimates the power consumption of the battery (150) to be consumed during driving within a predetermined period toward the future or within the period until the next charging, based on various information relating to the driver and the electric vehicle, as well as various information including past driving history of electric vehicles of the same type, and also estimates improvement measures using the LLM to bring the estimated power consumption closer to the minimum value, and a suggestion unit (203) that presents the estimated improvement measures.

Description

Technical Field

[0001] The present invention is applicable to electric vehicles using batteries, including, for example, battery electric vehicles (so-called BEV: Battery Electrical Vehicle), etc. For example, until the use of electric vehicles is completely finished today, for a period of several days or a week in the future, until this weekend or until the next charging, etc., in the usage period of electric vehicles until the relatively near future, it relates to the technical field of a system that pre-reads and analyzes the power consumption of the battery and presents an improvement in the power consumption.

Background Art

[0002] As a technology related to this type of system, as a cruising range notification device in an electric vehicle navigation device, a device has been developed that displays how far a vehicle can travel according to the remaining battery level of the in-vehicle battery and notifies whether it is possible to return home or to a predetermined charging facility (see Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, according to the above-mentioned background art, in the navigation device, simply displaying information regarding the remaining cruising range and the reachability to the destination, for example, during the end of the day with multiple destinations or for several days in the future, and further, until this weekend, it is technically difficult to present when and where to charge and further measures and countermeasures to avoid insufficient battery remaining.

[0005] The present invention aims to provide an LLM-based system for predicting and suggesting improvements to the power consumption of electric vehicles, which can predict power consumption with greater accuracy not only for the current drive but also for a predetermined period in the near future, and can suggest measures to improve power consumption. [Means for solving the problem]

[0006] One aspect of the LLM-based power consumption forecast analysis and improvement suggestion system for electric vehicles according to the present invention solves the above problem by using an LLM to estimate the power consumption of the battery to be consumed during driving in an electric vehicle equipped with a battery during a predetermined period from the present to the future or during the period until the electric vehicle is next charged, based on various information relating to the driver, the electric vehicle, and the driving, including (I) the driver of the electric vehicle and the electric vehicle, (Ia) the current situation and driving conditions and (Ib) past driving history, and (II) past driving history of an electric vehicle of the same type as the electric vehicle, and also includes an estimation unit that uses an LLM to estimate improvement measures to bring the estimated power consumption closer to the minimum value, and a presentation unit that presents the estimated improvement measures to the driver. [Effects of the Invention]

[0007] According to one aspect of the system according to the present invention, based on the various states and conditions of the user and electric vehicle on a daily basis, high-dimensional learning including characters enables the prediction of power consumption for a predetermined period in the near future (not just for the current drive) with more accurate prediction accuracy, and enables the suggestion of measures to improve power consumption.

[0008] The effects and benefits of the present invention will be further clarified by the embodiments of the invention described below. [Brief explanation of the drawing]

[0009] [Figure 1] This is a block diagram showing the overall configuration of the system according to the embodiment. [Figure 2]This is a flowchart showing an example of processing in the system according to the embodiment. [Modes for carrying out the invention]

[0010] First, with reference to Figure 1, the overall configuration of the electric vehicle power consumption predictive analysis and improvement suggestion system (hereinafter simply referred to as the "analysis and improvement suggestion system") using LLM (Large Language Models) according to the embodiment will be described.

[0011] The LLM used for analysis and estimation or for linguistic presentation in this embodiment may be either a so-called single-modal LLM or a multimodal LLM. This embodiment uses various information related to the driver, electric vehicle, and driving, including the current situation, driving conditions, and past driving history of the electric vehicle, as raw data for LLM training. Furthermore, this embodiment uses various information related to the driver, electric vehicle, and driving, including past driving history of electric vehicles of the same type as the electric vehicle, as raw data for LLM training. The system according to this embodiment is constructed as a system that performs LLM analysis or AI analysis based on this various information.

[0012] Specifically, as will be explained in detail below, the system will be designed to accurately estimate power consumption over a predetermined period in the near future, based on the daily conditions and circumstances of the user or driver and the electric vehicle, and to present power consumption improvement measures in a way that is easily accepted by the user or driver.

[0013] Furthermore, such AI learning or LLM learning can employ not only traditional AI learning systems such as supervised learning, unsupervised learning, or reinforcement learning, but also new technologies such as generative AI or LLM that have recently been put into practical use, are currently under development, or will be developed in the future. For example, the AI ​​learning or LLM learning described here may be constructed using a neural network that performs efficient learning through representation learning, transfer learning, feature selection, fine tuning or hyperparameter tuning, ensemble learning, etc.

[0014] As shown in Figure 1, the analysis and improvement suggestion system according to the embodiment comprises an on-board unit 101 mounted on an electric vehicle 100 and a server unit 200. The on-board unit 101 and the server unit 200 are connected to a communication network 10, such as the Internet or a dedicated network line. The communication network 10 also contains multiple or many other electric vehicles 100. The communication network 10 also contains an external related knowledge collection unit 301 that collects information obtained outside the electric vehicle 100 that can be used to perform fine-tuning or hyper-tuning in the analysis and improvement suggestion system to impart domain knowledge (i.e., "external related knowledge"). Furthermore, the communication network 10 also contains a user data collection unit 306 that collects data specific to the driver or user related to the electric vehicle 100 (i.e., "user data"). These external related knowledge collection units 301 and user data collection units 306 may be provided, at least in part, within the server unit 200 or within the facility where the server unit 200 is located, or within the vehicle-mounted unit 101 or within the vehicle.

[0015] The server unit 200 is connected to a database (DB) 300 that stores various data, including data used in the analysis and improvement suggestion system. The DB 300 may be connected to the server unit 200 or the vehicle unit 101 via a communication network 10. The server unit 200 is composed of various computer-equipped devices and various computer devices that perform centralized or distributed processing. In other words, the analysis and improvement suggestion system is constructed as a system that performs centralized or distributed processing using the large-scale data in the DB 300.

[0016] In Figure 1, the electric vehicle 100 is equipped with a battery 150 and is configured as a BEV, for example. The electric vehicle 100 may also be a so-called HEV (Hybrid Electric Vehicle), PHEV (Plugin HEV), FCEV (Fuel Cell EV), etc., which use a battery.

[0017] The in-vehicle unit 101 comprises a sensor unit 102 including various sensors positioned in predetermined locations inside the vehicle, a processing unit 103 including a computer, a communication unit 104 including a modem configured to communicate with the outside of the vehicle via a communication network 10, an interface unit 106 configured to exchange voice and images with the user or driver inside the vehicle, and a user data collection unit 116.

[0018] One of its detection functions is to detect the remaining charge data of the battery 150 and pass it to the processing unit 103. The sensor unit 102 is also configured to detect the current driving status of the electric vehicle 100 and various information 102a related to the driver of the electric vehicle 100, and pass it to the processing unit 103 as CAN (Controlled Are Network) data, etc.

[0019] The processing unit 103 has a CPU, a memory, etc. that control the sensor unit 102, the communication unit 104, and the interface unit 106, and transmits various information related to the driver and driving of the electric vehicle 100, including the planned route of the electric vehicle 100, the current driving state of the electric vehicle 100, and the past driving history, from the communication unit 104 to the server unit 200 side in data of a predetermined format. Further, via the communication unit 104, improvement measures and the like indicated by improvement measure data and the like received after processing there from the server unit 200 side are configured to be presented to the user or the driver through the interface unit 106.

[0020] The communication unit 104 transmits data necessary for consumption power prediction analysis and improvement presentation collected in the electric vehicle 100 to the server device 200 via the communication network 10 under the control of the processing unit 103. Further, it is configured to receive, via the communication network 10, the results of the consumption power prediction analysis of the electric vehicle 100 and data related to improvement presentation generated by using the LLM in the server unit 200.

[0021] The interface unit 106 is configured to be able to input, by voice input or a predetermined operation on an image, the destination of the electric vehicle 100, the conditions when selecting the planned route to the destination, etc. Regarding the selection of the planned route here (i.e., the navigation function), all or part of it may be configured to be executed by the processing unit 103, or part or the main part of it may be executed by the processing unit 202 on the server unit 200 side (in other words, the in-vehicle unit 101 side mainly serves as a browser function). The interface unit 106 is further configured to be able to output, in a predetermined format on voice output or an image, the results of the consumption power prediction analysis and data related to improvement presentation obtained from the server unit 200 side.

[0022] The user data collection unit 107 is configured to collect data unique to the driver or user related to the electric vehicle 100 (i.e., "user data") inside the electric vehicle 100 separately from the sensor unit 102. Examples of user data include attribute data unique to the user, such as gender, age, driving experience, accident history, hobbies, driving habits, fatigue level, medical history, and pre-existing conditions.

[0023] The user data collection unit 306 collects user data via a PC, smartphone, dedicated app, etc. (not shown) owned or related to the user or driver, whether it is accommodated in the communication network 10 or not. The user data collection unit 306 passes the collected user data to the server unit 200 via the communication network 10 to contribute to the LLM learning there. Thus, the user data collection unit 306 can widely collect user data by collecting user data outside the electric vehicle 100, and the aforementioned user data collection unit 107 can collect user data inside the electric vehicle 100. The user data will be used for the processing related to the LLM in the processing unit 202, at least in part in a form that has been texturized or verbalized.

[0024] In FIG. 1, the server device 200 includes a communication unit 201 including a modem or the like capable of communicating with each electric vehicle 100, the external related knowledge collection unit 301, and the user data collection unit 306 via the communication network 10, a processing unit 202 including a computer capable of executing estimation processing of power consumption by the LLM, which will be described in detail later, and a presentation unit 203 capable of generating presentation data for presenting improvement measures according to the estimation result by the processing unit 202.

[0025] The communication unit 201, under the control of the processing unit 202, receives data necessary for predictive power consumption analysis and improvement suggestions collected by the electric vehicle 100 via the communication network 10. The communication unit 201, under the control of the processing unit 202, also receives external related knowledge collected by the external related knowledge collection unit 301 and user data collected by the user data collection unit 306 via the communication network 10 as part of the data necessary for predictive power consumption analysis and improvement suggestions. The communication unit 201 is further configured to transmit the results of the predictive power consumption analysis and improvement suggestions, which have been processed and generated by the processing unit 202 and the suggestion unit 203, to the electric vehicle 100 that was the subject of the analysis via the communication network 10.

[0026] The processing unit 202 is configured to estimate the power consumption of the battery 150 consumed during driving of the electric vehicle 100 that is the subject of this analysis, over a predetermined period from the present to the future (for example, three days, five days, one week, ten days, one month, etc.) or until the next time the electric vehicle 100 is charged, using LLM based on information related to the driver of the electric vehicle 100, the electric vehicle 100, and driving, including (I) (Ia) the current situation and driving conditions and (Ib) past driving history of the electric vehicle 100 and the electric vehicle 100, and (II) past driving history of electric vehicles 100 of the same type as the electric vehicle 100. It is also configured to estimate improvement measures using LLM to bring the estimated power consumption closer to the minimum value.

[0027] The presentation unit 203 is configured to generate presentation data in a predetermined format corresponding to the interface unit 106 installed inside the electric vehicle 100, indicating that the improvement measures thus estimated should be presented to the driver or user of the electric vehicle 100, and to pass this data to the communication unit 201. In this embodiment, the "presentation unit" is configured to include the presentation unit 203 on the server unit 200 side and the interface unit 106 on the in-vehicle unit 101 side, and the in-vehicle unit 101 side mainly handles the browser function with respect to the presentation function.

[0028] DB300 is configured to include a large-scale and high-speed data input / output storage device that stores various data received by the server unit 200 via the communication network 10, in particular various data necessary for estimation processing using LLM, data related to the estimation results or intermediate results generated by the processing unit 202, and presentation data generated by the presentation unit 203.

[0029] Next, with reference to the block diagram in Figure 1 and the flowchart in Figure 2, an example of processing in the analysis and improvement suggestion system according to this embodiment (in particular, processing performed using LLM in the processing unit 202 of the server unit 200) will be explained.

[0030] In Figure 2, first, user data such as driving history, habits, and preferences of the driver or user operating the electric vehicle 100 is collected by the user data collection unit 107 and the user data collection unit 306 (step S1).

[0031] Next, the collected user data is passed to the processing unit 202 on the server unit 200 via the communication network 10. The processing unit 202 analyzes the user data (step S2) and determines whether or not there is a driving history of a similar user (step S3). The analysis here may include classification processing by appropriate categorization, characterization processing, or language processing. A "similar user" here typically refers to a user who has driven the same route with the "same type of electric vehicle (BEV)". Furthermore, "same type of BEV" can include not only BEVs that are completely the same type or model as the user's own BEV, but also BEVs that have pre-set common or similar specifications. For example, if the powertrain is the same or the navigation number is the same for both vehicles, they can be treated as the same type here. Furthermore, minor differences between the two vehicles may be corrected by LLM so that they can be treated as the same type of vehicle or a vehicle of the same type of user. As described above, the determination of whether or not a user is similar in step S3 is based on predetermined criteria that are identical or similar to the electric vehicle 100 being driven and identical or similar to the driving route.

[0032] If the determination shows no similar user driving history (Step S3: No), the processing unit 202 then performs a prediction of the electric vehicle 100's driving schedule (Step S4). More specifically, the processing unit 202 performs predictions of the day of the week (i.e., the day of the week when the vehicle will be driven next or in the near future), the time of day (i.e., the time of day when the vehicle will be driven), the weather (i.e., the weather at the location where the vehicle will be driven on that day or at that time), and the route congestion (i.e., congestion on the route and at the time when the vehicle will be driven) using AI or LLM estimation. Furthermore, the processing unit 202 generates measures to improve power consumption in the predicted driving schedule using LLM (Step S5).

[0033] On the other hand, if the determination results in a driving history of a similar user (Step S3: Yes), the past driving history of that similar user is obtained from the DB300 or the user data collection unit 107 or user data collection unit 301 of the other electric vehicle 100 (Step S6). Subsequently, the power consumption improvement measures in the aforementioned operating schedule are generated by the processing unit 202 in the LLM, in a form adopted as reference information for others and used as raw data for the LLM (Step S5).

[0034] Prior to the processing using the LLM in step S5, for example, in parallel with, before, or concurrent with step S1, external related knowledge may be acquired by the external related knowledge acquisition unit 301. This allows the processing unit 202 to perform fine-tuning or hyper-tuning to impart domain knowledge in step S5, etc. By using an LLM that has been fine-tuned with domain knowledge related to electric vehicles, such as BEV domain knowledge, it becomes possible to generate improvement measures by comparing them with the best driving settings.

[0035] For example, in step S5, the estimation of "improvement measures" using LLM may involve the processing unit 202 extracting data corresponding to "minimum power consumption" from the past driving history acquired in step S6, and then estimating improvement measures based on this past driving history. In this case, the LLM may identify the difference between the driving method (driving time, driving route, driving method, etc.) corresponding to the extracted minimum power consumption and the current driving method, and measures to reduce this difference may be estimated as improvement measures. For these reasons, converting all the relationship graphs and numerical data between elements and power consumption accumulated for each road link into text using LLM and then vectorizing them can improve the overall efficiency or accuracy of the processing.

[0036] Thus, in this embodiment, the LLM executed in step S5, etc., uses a large amount of text data, including various information related to the driver of the electric vehicle 100 and the electric vehicle 100, such as the current situation, driving status, and past driving history of the electric vehicle 100. Furthermore, this embodiment uses various information related to the driver of the electric vehicle 100 and the electric vehicle 100, including past driving history of electric vehicles of the same type as the electric vehicle 100. The various information appropriately includes personal information, traffic laws and common sense information related to traffic laws (e.g., driving on the left, speed limits in residential areas, etc.), and general common sense information (e.g., it is dark at night, traffic congestion is likely during commuting hours and holidays, the presence of landmarks near the planned route, etc.), and uses a large amount of information that has been transcribed or verbalized.

[0037] However, a multimodal LLM capable of AI learning based not only on verbalized information but also on non-verbalized information may be employed in steps S5, etc. That is, the data used in processing in step S5, etc., includes text data, but is not limited to text data.

[0038] In the processes described above, such as step S5, a large amount of text data is used to fine-tune a large-scale language model using LLM. As a result, it can be adapted to various natural language processing (NLP) tasks such as text classification, sentiment analysis, information extraction, text summarization, text generation, and question answering.

[0039] In either step S4 or S6, improvements such as power saving measures, calculation of the next charging timing, and measures to avoid battery depletion are generated using LLM based on the large amount of available text data as described above (step S5).

[0040] In step S7, the presentation unit 203 generates presentation data indicating that the improvement measures generated by LLM are to be presented as a notification to the user of the electric vehicle 100, and the interface unit 106 on the in-vehicle unit 101 outputs the "improvement measures" as audio or image. In addition, the processing unit 202 and the presentation unit 203 also output presentation data indicating that the "basis" for the improvement measures is to be presented, as well as output as audio or image via the interface unit 106. It is also preferable to use LLM to generate the presentation data for outputting the improvement measures etc. as audio or image in the presentation unit 203. That is, here too, presentations using AI voice or AI image may be performed for the user or driver, adapted to various natural language processing tasks such as text classification, sentiment analysis, information extraction, text summarization, text generation, and question answering.

[0041] Next, it is determined whether or not there is feedback (step S8: Yes), and if there is (step S8: Yes), the process returns to step S4 and the subsequent steps are repeated, and an "improvement plan" updated by AI learning is presented (step S7).

[0042] The "improvement measures" notified or presented here should preferably be notified or presented along with their "basis" from the perspective of convincing the user or driver, or in other words, from the perspective of getting the driver to follow the improvement measures.

[0043] For example, the system might suggest that users charge their vehicles once at night during the next week. This would save costs by charging during off-peak hours when electricity rates are lower, and ensure a sufficient battery level of 150. The benefits include reduced charging costs due to charging during off-peak hours, and the ability to secure enough battery power for the next week's driving, thus avoiding power shortages. In this case, along with the suggestion, the system would provide justification to the user, such as, "Based on your driving history over the past week and your future plans, it is predicted that battery consumption will exceed the current level. Charging at night will allow you to reduce costs and replenish power."

[0044] For example, a company might suggest that users avoid frequent acceleration and deceleration during rush hour and drive at a constant speed. As a result, improving driving habits could reduce power consumption by approximately 15%, extend the car's range, and reduce the number of charging cycles. In this case, along with the suggestion of improvement, the company might provide or notify users that, for example, "analysis of past rush hour driving data has revealed that frequent acceleration and deceleration are the cause of increased power consumption. By changing driving methods during future rush hour, energy consumption can be effectively reduced."

[0045] For example, the system might suggest that users "prepare their vehicles in advance and moderately reduce their use of the air conditioner in preparation for the drop in temperature over the next few days." This is expected to reduce excess power consumption from the air conditioner and heating system, thereby improving the range of the 150 battery. In this case, along with the suggestion of improvement, the system would provide or notify the user that, for example, "based on weather forecasts and past power consumption data during low temperatures, it is predicted that if the use of the air conditioner is not adjusted, power consumption will increase significantly, potentially resulting in insufficient driving range."

[0046] After processing related to the presentation of such improvement measures, if the result of the determination in step S8 is that there is no feedback (step S8: No), the series of processes is terminated.

[0047] As explained in detail above, according to this embodiment, rather than simply regarding the current journey to the destination, the user's driving history of the electric vehicle 100, the periodicity and characteristics of the driving, or the driving history of other electric vehicles 100 that are driving ahead on the same route or have driven in the past, the driving history of other electric vehicles 100 that are driving on the same route or surrounding routes or have driven in the past (Step S6), and furthermore, the user's usual driving habits, customs, tendencies, and preferences (Step S1), and furthermore, based on the driving schedule (Step S4) including the days of the week, time of day, weather conditions, and route congestion that the user is estimated to drive in the near future, the system presents the user with measures to improve power saving, the timing of the next charge, or measures and responses to avoid running out of battery power before charging (Step S5), either in advance or in real time (Step S7).

[0048] In addition, in this embodiment, based on user feedback (step S8), improvement measures that take personal preferences into even greater consideration can be presented for subsequent uses (steps S4 to S7). In this way, personal preferences can be reflected by utilizing LLM-specific feedback mechanisms such as RLHF (i.e., reinforcement learning from human feedback). Therefore, it becomes possible to collect feedback for each element that affects power consumption and generate improvement measures that are more tailored to the individual.

[0049] Unlike previous technologies or background technologies, which only use numerical data, this embodiment allows for the extraction of relevant features from textual data such as papers and related literature by using LLM, enabling the calculation or estimation of power consumption at a higher level. Furthermore, this embodiment allows for the simultaneous provision of results, justifications, and improvement measures to the user in natural language or voice, thereby increasing the user's understanding and satisfaction with the response.

[0050] Note The following additional information is disclosed regarding the embodiments described above.

[0051] [Note 1] The analysis and improvement suggestion system described in Appendix 1 of the present invention is characterized by comprising: an estimation unit that estimates the power consumption of a battery in an electric vehicle equipped with a battery during driving within a predetermined period from the present to the future or within the period until the next time the electric vehicle is charged, based on various information relating to the driver, the electric vehicle, and the driving, including (I) the driver of the electric vehicle and the electric vehicle, (Ia) the current situation and driving conditions and (Ib) past driving history, and (II) past driving history of an electric vehicle of the same type as the electric vehicle, and estimates improvement measures to bring the estimated power consumption closer to the minimum value using LLM; and a suggestion unit that presents the estimated improvement measures to the driver.

[0052] According to the analysis and improvement suggestion system described in Appendix 1, by using LLM (Limited Literacy Modeling), based on the various conditions and circumstances of the user and electric vehicle on a daily basis, it is possible to predict power consumption for a predetermined period in the near future with greater accuracy through high-dimensional learning that includes or incorporates text. The factors that have the greatest impact on power consumption and improvement measures can also be explained in human-understandable natural language. Furthermore, it is possible to provide the collected driving history as information that can be useful to other vehicles. Such suggestions for improvement measures also lead to teaching users how to drive in a way that saves power as they use their electric vehicles on a daily basis.

[0053] [Note 2] The analysis and improvement suggestion system described in Appendix 2 of the present invention is a power consumption predictive analysis and improvement suggestion system described in Appendix 1, characterized in that the estimation unit estimates the improvement measures by identifying the difference between the current driving state and the lowest power consumption in the past driving history using the LLM.

[0054] According to the analysis and improvement suggestion system described in Appendix 2 of the present invention, by identifying the difference between the current driving state and the past driving history with the lowest power consumption using LLM, it becomes possible to estimate improvement measures that bring power consumption closer to the minimum value with relatively efficient efficiency.

[0055] [Note 3] The analysis and improvement suggestion system described in Appendix 3 of the present invention is a power consumption predictive analysis and improvement suggestion system described in Appendix 1 or 2, characterized in that the estimation unit acquires external relevant knowledge other than information relating to the current driving state and the past driving history, and estimates the improvement measures taking into account domain knowledge relating to the electric vehicle using a finely tuned LLM using the acquired external relevant knowledge.

[0056] According to the analysis and improvement suggestion system described in Appendix 3 of the present invention, improvement measures are estimated not only based on information related to the current driving state and past driving history, but also by considering domain knowledge with an LLM that has been fine-tuned using external relevant knowledge, thus enabling more accurate estimation.

[0057] [Note 4] The analysis and improvement suggestion system described in Appendix 4 of the present invention is a power consumption predictive analysis and improvement suggestion system described in any one of Appendix 1 to 3, characterized in that the estimation unit acquires information relating to the personal preferences of the electric vehicle driver and estimates the improvement measures in an LLM that reflect the personal preferences using a mechanism that provides the acquired information relating to personal preferences as dedicated feedback to the LLM.

[0058] According to the analysis and improvement suggestion system described in Appendix 4 of the present invention, improvement measures are estimated using LLM not only based on information relating to the current driving state and past driving history, but also in a way that reflects individual preferences. Therefore, it becomes possible to present highly accurate improvement measures appropriate to the driver or user in a more convincing manner.

[0059] [Note 5] The analysis and improvement suggestion system described in Appendix 5 of the present invention is a power consumption predictive analysis and improvement suggestion system described in any one of Appendix 1 to 4, characterized in that the suggestion unit presents information indicating the basis for the estimation of the improvement measures together with the improvement measures.

[0060] According to the analysis and improvement suggestion system described in Appendix 5 of the present invention, not only improvement measures but also the rationale for those measures are presented, making it possible to present highly accurate improvement measures in a way that is easier for drivers or users to understand.

[0061] [Note 6] The analysis and improvement suggestion method described in Appendix 6 of the present invention is characterized by comprising the steps of: estimating the power consumption of the battery to be consumed on a planned route when traveling to a destination in an electric vehicle equipped with a battery, using LLM based on information related to the driver and driving of the electric vehicle, including the current driving state and past driving history of the electric vehicle, and the past driving history of an electric vehicle of the same type as the electric vehicle, and using LLM to estimate improvement measures to bring the estimated power consumption closer to the minimum value; and presenting the estimated improvement measures to the driver.

[0062] According to the analysis and improvement suggestion method described in Appendix 6 of the present invention, similar to the analysis and improvement suggestion system described in Appendix 1, by using LLM, it is possible to predict power consumption with greater accuracy through high-dimensional learning that includes or also includes characters, and to explain the factors that have the greatest impact on power consumption and improvement measures in human-understandable natural language.

[0063] The present invention may be modified as appropriate, without contradicting the gist or spirit of the invention as can be inferred from the claims and the entire specification, and such modifications, along with the analysis and improvement suggestion system and method, are also included in the technical concept of the present invention. [Explanation of Symbols]

[0064] Electric vehicles (BEVs)...100 Vehicle-mounted unit...101 Sensor unit...102 Processing section...103 Interface...106 User Data Collection Department...107 Battery... 150 Server section...200 Processing section...202 Presentation section...203 DB...300 External Knowledge Acquisition Department...301 User Data Collection Department...306

Claims

1. An estimation unit for an electric vehicle equipped with a battery estimates the power consumption of the battery during driving within a predetermined period from the present to the future, or within the period until the next time the electric vehicle is charged, based on various information relating to the driver, the electric vehicle, and the driving, including (I) the driver of the electric vehicle and the electric vehicle, (Ia) the current situation and driving conditions and (Ib) past driving history, and (II) past driving history of an electric vehicle of the same type as the electric vehicle, and estimates improvement measures to bring the estimated power consumption closer to the minimum value, using LLM. A presentation unit that presents the estimated improvement measures to the driver. A system for predictively analyzing and suggesting improvements to the power consumption of electric vehicles using LLM, characterized by having the following features.

2. The estimation unit estimates the improvement measures by identifying the difference between the current driving state and the past driving history with the lowest power consumption using the LLM. A power consumption forecast analysis and improvement suggestion system for electric vehicles using LLM according to the description in 1, characterized in that it is a system for predicting and analyzing the power consumption of electric vehicles.

3. The estimation unit acquires external relevant knowledge other than the information relating to the current driving state and the past driving history, and uses the acquired external relevant knowledge to fine-tune the LLM and estimate the improvement measures taking into account the domain knowledge relating to the electric vehicle. A power consumption forecast analysis and improvement suggestion system for electric vehicles using LLM according to claim 1 or 2, characterized in that it is a system for predicting power consumption and suggesting improvements.

4. The estimation unit acquires information relating to the personal preferences of the electric vehicle driver and uses a mechanism to provide LLM-specific feedback of the acquired personal preference information to estimate the improvement measures in LLM in a manner that reflects the personal preferences. A power consumption forecast analysis and improvement suggestion system for electric vehicles using LLM as described in any one of claims 1 to 3.

5. The display unit presents information indicating the basis for the estimation of the improvement measures, along with the improvement measures themselves. A power consumption forecast analysis and improvement suggestion system for electric vehicles using LLM as described in any one of claims 1 to 4.

6. In an electric vehicle equipped with a battery, the power consumption of the battery consumed during driving within a predetermined period from the present to the future, or within the period until the next charging of the electric vehicle, is estimated by LLM based on various information relating to the driver, the electric vehicle, and the driving, including (I) the driver of the electric vehicle and the electric vehicle, (Ia) the current situation and driving conditions and (Ib) past driving history, and (II) past driving history of an electric vehicle of the same type as the electric vehicle, and the LLM is also estimated to be an improvement measure to bring the estimated power consumption closer to the minimum value. A step of presenting the estimated improvement measures to the driver. A method for predictively analyzing and suggesting improvements to the power consumption of an electric vehicle using LLM, characterized by comprising the following features.

Images