Model utilization system, model utilization method and program

Abstract

To further efficiently utilize a digital twin model while guaranteeing credit enhancement.SOLUTION: A model utilization system according to an embodiment includes a distributed ledger processing unit, a token processing unit, and a presentation control unit. The distributed ledger processing unit processes a distributed ledger that stores a scenario to be used for analysis of a target system, a digital twin model to be used for the analysis, and data to be input to the digital twin model. The token processing unit calculates a token for an organization that has provided the digital twin model to be utilized and for an organization that has provided the data to be utilized. The presentation control unit retrieves, from the distributed ledger, a candidate for the scenario that satisfies a user request, a candidate for the digital twin model that satisfies the user request, and a candidate for the data that satisfies the user request, presents the retrieved candidates, and presents the token required to use a combination of the retrieved candidate for the digital twin model and the retrieved candidate for the data.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] An embodiment of the present invention relates to a model utilization system, a model utilization method, and a program. [Background technology]

[0002] The use of digital twins is advancing to reduce downtime and improve availability of urban infrastructure systems, etc. Distributed ledger / blockchain systems and digital currency / token economy systems are also emerging as mechanisms for preventing data tampering and streamlining procedures such as contracts and negotiations. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Special Publication No. 2021-523504 [Patent Document 2] Patent Publication No. 2021-87608 Summary of the Invention [Problem to be solved by the invention]

[0004] However, with conventional technology, it was difficult to utilize digital twin models more efficiently while ensuring credit enhancement. [Means for solving the problem]

[0005] A model utilization system according to an embodiment includes a distributed ledger processing unit, a token processing unit, and a presentation control unit. The distributed ledger processing unit processes a distributed ledger that stores scenarios used in analyzing a target system, digital twin models used in the analysis, and data to be input to the digital twin models. The token processing unit calculates tokens that represent incentives for the institution that provided the digital twin model to be utilized and the institution that provided the data to be utilized. The presentation control unit searches the distributed ledger for candidate scenarios that satisfy user requests, candidate digital twin models that satisfy user requests, and candidate data that satisfy user requests, and presents the searched candidate scenarios, candidate digital twin models, and candidate data, as well as tokens required for utilizing a combination of the searched candidate digital twin models and candidate data. [Brief explanation of the drawings]

[0006] [Figure 1A] A diagram showing example 1 of a system that can utilize the digital twin model. [Figure 1B] A diagram showing example 2 of a system that can utilize the digital twin model. [Figure 2] FIG. 1 is a diagram for explaining the mechanism of a model utilization system according to an embodiment. [Figure 3] FIG. 2 is a diagram showing an example of a functional configuration related to model utilization processing of the model utilization system of the embodiment. [Figure 4] FIG. 1 is a diagram illustrating an example of the functions of a distributed ledger (blockchain) according to an embodiment. [Figure 5] FIG. 1 is a diagram showing an example of a network configuration of a distributed ledger (blockchain) according to an embodiment. [Figure 6] FIG. 1 is a diagram showing an example of the form of a distributed ledger (blockchain) according to an embodiment. [Figure 7] FIG. 1 is a diagram showing Example 1 of a distributed ledger (blockchain) according to an embodiment. [Figure 8]FIG. 10 is a diagram showing a second example of a distributed ledger (blockchain) according to an embodiment. [Figure 9] FIG. 1 is a diagram for explaining the configuration related to IP distribution and NFT management of a model utilization system according to an embodiment. [Figure 10] 1A to 1C are diagrams for explaining an example of a method for extracting candidates for a scenario, a model, and data according to an embodiment. [Figure 11] 1 is a flowchart showing an example of a method for extracting candidates for a scenario, a model, and data according to an embodiment. [Figure 12] 10 is a flowchart illustrating an example of a token calculation method according to an embodiment. [Figure 13] FIG. 1 is a diagram showing an example of a method for utilizing a digital twin model using a distributed ledger according to an embodiment (in the case of a surrogate model). [Figure 14] 10A and 10B are diagrams illustrating an example of a UI that presents presentation information according to an embodiment. [Figure 15] FIG. 1 is a diagram illustrating an example of a hardware configuration of a model utilization system according to an embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0007] Hereinafter, embodiments of a model utilization system, a model utilization method, and a program will be described in detail with reference to the accompanying drawings.

[0008] In the case of infrastructure systems involving multiple organizations, such as public institutions, infrastructure operating companies, infrastructure development companies, and infrastructure insurance companies, the mechanisms for utilizing digital models related to digital twins and for data credit enhancement are issues. For example, while expectations are high for digital twins of cities, the methods and mechanisms for credit enhancement for the utilization of digital models and sensitive data by multiple organizations, including users, multiple companies, public institutions, and national and local governments, are an issue. Examples of credit enhancement include the following:

[0009] Authenticity: Prevents connections from fake users and devices (sensors) Integrity: Protecting data and digital models from tampering Reliability: Reliability of data, reliability of data accuracy due to malfunctions of sensor equipment or inappropriate sensing methods, and reliability of communication and control (including the Verification & Validation (V&V) level of simulations). Confidentiality: Protecting personal information and limiting the information that can be used or disclosed based on user permissions (including user authentication and protecting the intellectual property (IP) of digital models). Availability: Robustness that can withstand system failures such as partial communication outages, and flexibility in system reconfiguration and reconstruction through a distributed mechanism in the operation and maintenance of digital twins (including improved scalability).

[0010] Furthermore, examples of infrastructure systems include infrastructure structures, plants, power electronics, battery systems, elevator systems, distributed energy, power grids, water grids, transportation networks, and communication networks.

[0011] Figure 1A is a diagram showing Example 1 of a system that utilizes a digital twin model. The example in Figure 1A shows the case of scenario planning for the maintenance of infrastructure equipment such as elevators.

[0012] Maintenance planning is determined through sensing and monitoring of infrastructure equipment, and failure risk prediction and triage of the equipment. For example, if the infrastructure equipment is an elevator, maintenance planning includes determining the order in which field engineers should carry out recovery work when multiple failures occur.

[0013] In the case of an infrastructure equipment system such as an elevator, for example, the building owner, the BIM (Building Information Modeling) owner company, the elevator company, the Ministry of Land, Infrastructure, Transport and Tourism, local governments, and infrastructure insurance companies are all involved.

[0014] Figure 1B shows Example 2 of a system that utilizes a digital twin model. The example in Figure 1B shows the case of predictive maintenance (failure risk prediction) of power electronics equipment in railway vehicles.

[0015] Failure risk prediction (provision of maintenance timing) is based on life prediction obtained by inputting sensing and monitoring data into a surrogate model.

[0016] In the case of predictive maintenance of power electronics (hereinafter referred to as "power electronics") equipment for railway vehicles, for example, railway companies, device vendors, railway system companies, maintenance companies, and infrastructure insurance companies are involved.

[0017] Figure 2 is a diagram for explaining the mechanism of the model utilization system of the embodiment. In order to promote the utilization of digital twin models, the following three mechanisms are mainly required.

[0018] (1) Methods and mechanisms for ensuring credibility (reliability, confidentiality, availability) in the utilization of digital twin models and data by multiple organizations, such as multiple companies, public institutions, and national and local governments. (2) A mechanism to provide incentives to data and model providers (3) A mechanism for balancing various values (such as improved resilience for safety and security, carbon neutrality, circular economy, energy management, and economic efficiency) in customer requests (examples of user requests). Calculate each value, risk, and tolerance range, then extract candidate scenarios and candidate models that can be combined while ensuring credit enhancement, and reach a consensus among multiple institutions.

[0019] The model utilization system 1 of the embodiment has a function of extracting and presenting candidates for scenarios, models, and data (data sets) to be utilized. The model utilization system 1 of the embodiment also has a function of calculating and presenting model risk, model value, and tokens (digital currency) related to model utilization.

[0020] The model utilization system 1 of the embodiment extracts candidate scenarios, models, and data by utilizing not only the importance and relevance obtained from the utilization history of model data, but also the token calculation model and the value / risk calculation model. Furthermore, for the utilized scenarios, models, and data, not only utilization history data and data related to reliability, but also matrix data (or network data, etc.) indicating the relevance and importance of each scenario, model, and data are registered (updated) in the distributed ledger.

[0021] [Example of functional configuration] Fig. 3 is a diagram illustrating an example of a functional configuration related to model utilization processing of the model utilization system 1 of the embodiment. The example of the model utilization system 1 in Fig. 3 illustrates a digital twin model utilization system that applies distributed ledger technology.

[0022] The model utilization system 1 of the embodiment includes a distributed ledger processing unit 11, a presentation control unit 12, a calculation unit 13, and a token processing unit 14.

[0023] The distributed ledger processing unit 11 executes distributed ledger processing related to the utilization of the model (e.g., processing of blockchain transactions). For example, the distributed ledger processing unit 11 performs at least one of the following processes: authentication of users accessing the distributed ledger, access control of the distributed ledger according to user authority, consensus formation, transaction execution, fee determination, block registration in the distributed ledger, and distributed ledger mining.

[0024] The presentation control unit 12 extracts candidates for scenarios, models, and data to be utilized, and presents presentation information including the extracted scenarios, models, and data. For example, the presentation control unit 12 searches for candidates for digital twin models whose risks and values calculated by the calculation unit 13 satisfy user requirements (e.g., customer requirements).

[0025] Here, scenarios refer to scenarios that anticipate events that may occur in the future. For example, in the case of predictive maintenance, scenarios include operation scenarios and maintenance scenarios for infrastructure systems that are consistent with customer requirements, as well as hazard scenarios such as earthquakes and strong winds.

[0026] The presentation control unit 12 extracts important or related scenario candidates from multiple assumed scenarios based on a customer request and presents the extracted scenario candidates. Furthermore, the presentation control unit 12 finds models and datasets related to the assumed scenarios based on relevance information from among the scenarios, models, and data registered in the distributed ledger (blockchain), and presents the models and datasets as well.

[0027] The relevance information is, for example, a matrix with scenario ID, model ID, and data ID as dimensions and their relevance as elements. Alternatively, the relevance information is a network with each scenario ID, model ID, and data ID as a node, and links (with values related to relevance) between related nodes. Alternatively, the relevance information is, for example, a statistical / probability model related to scenario ID, model ID, and data ID. Alternatively, the relevance information is, for example, an AI (artificial intelligence) / machine learning model related to scenario ID, model ID, and data ID.

[0028] The model presented by the presentation control unit 12 may be a model automatically generated from a function candidate library.

[0029] The calculation unit 13 calculates the risk and value of the model. Based on the candidate scenarios, the calculation unit 13 calculates the risk calculated from the candidate model and the value indicating the usefulness of the candidate model. For example, the calculation unit 13 calculates each value, risk, and tolerance range in order to balance various values in customer requests (such as improving the resilience of infrastructure systems for safety and security, carbon neutrality, circular economy, and energy management). The calculation unit 13 enables the presentation control unit 12 described above to extract and present scenarios and candidate models that also have guaranteed credit enhancement.

[0030] The token processing unit 14 processes NFTs (Non-Fungible Tokens) and IP tokens related to the utilization of models. Here, NFTs are digitalized data of non-fungible tokens linked to model IDs and data IDs.

[0031] Furthermore, the model utilization system 1 of the embodiment includes various model bases and databases related to digital twins.

[0032] Here, model bases related to digital twins include, for example, temporal and spatial digital models, simulation models, control models, failure and lifespan prediction models, sensor and actuator models, and risk, value, and token calculation models. Each of these models has a digital ID or NFT linked to a distributed ledger. The digital model itself can also be stored on the model base of the blockchain system, or only the digital ID and NFT can be stored on the blockchain system, and the digital model of the model itself stored in another system can be referenced and read.

[0033] Figure 4 is a diagram showing an example of the functions of a distributed ledger (blockchain) according to an embodiment. The blockchain according to an embodiment incorporates an "electronic signature" mechanism using public key cryptography (public key / private key) encryption technology to guarantee the trustworthiness and accuracy of the parties attempting to transact, and to protect privacy. When sending and receiving secret messages using public key cryptography, the sender encrypts the message using a public key made public by the recipient. Since the encrypted message can only be decrypted using the recipient's private key, the contents cannot be deciphered or tampered with even if the message is intercepted by a third party.

[0034] In the case of a distributed ledger such as the blockchain of the embodiment, there is no central administrator, so all participants must monitor for fraudulent transactions and agree on the legitimacy of new transactions made within the network. However, consensus building in a distributed network is naturally more difficult than in a centralized network. Therefore, in a distributed network, a certain "consensus algorithm (program)," that is, rules for consensus building, are established to smoothly build consensus among participants and record new transactions while eliminating fraudulent transactions.

[0035] A smart contract (contract automation) is a mechanism that programs the contract performance conditions (agreements) in blockchain transactions of this embodiment and automatically establishes and fulfills the contract. A smart contract is a program that runs on a blockchain. In insurance terms, a contract is established when the conditions for contract establishment are met, and insurance payments are made when the conditions for insurance payment are met. A smart contract handles these procedures digitally. This mechanism is similar to the mechanism in a vending machine, where a sales contract is established the moment a user inserts coins and presses the button for a drink, and can streamline complex procedures such as contract negotiation, execution, and verification.

[0036] Figure 5 is a diagram showing an example of the network configuration of a distributed ledger (blockchain) according to an embodiment. The role of distributed ledger management in the blockchain according to an embodiment is played by a P2P (peer-to-peer) network. Unlike the conventional client-server model, nodes 2 communicate directly with each other on a one-to-many basis without having a specific server or client, enabling participating users (nodes 2) to share information and exchange transaction settlements. The P2P network is the key to realizing open, fair, and impartial direct transactions between participants, moving away from centralized transactions.

[0037] In the embodiment, each node 2 is managed and used by participants who use the model utilization system 1, such as data ownership management organizations, manufacturers, creators, and IP management companies.

[0038] FIG. 6 is a diagram showing an example of the form of a distributed ledger (blockchain) according to an embodiment. The form of the distributed ledger according to an embodiment is a consortium type. The administrator of the distributed ledger according to an embodiment is, for example, a plurality of companies. Note that the administrator may also include non-profit organizations and government agencies other than companies. The node participants and trading participants of the distributed ledger according to an embodiment are permissioned. Consensus formation of the distributed ledger according to an embodiment is a consensus among participants.

[0039] Consortium-based distributed ledgers are used, for example, by financial institutions.

[0040] A target system of the model utilization system 1 of the embodiment is, for example, an infrastructure system. For example, data input to the model includes at least one of the usage history of the infrastructure system, the load history of the infrastructure system, and monitoring data of the infrastructure system. Furthermore, for example, the utilized model is a surrogate model used for at least one of anomaly sign diagnosis of the infrastructure system, life expectancy prediction of the infrastructure system, and failure risk prediction of the infrastructure system.

[0041] FIG. 7 is a diagram illustrating Example 1 of a distributed ledger (blockchain) according to an embodiment. The example in FIG. 7 illustrates a case where the system to be analyzed is an infrastructure system. Specifically, the example in FIG. 7 illustrates a distributed ledger for use in managing the utilization of a simulation surrogate model for power electronics predictive maintenance.

[0042] Fig. 8 is a diagram illustrating Example 2 of a distributed ledger (blockchain) according to an embodiment. The example in Fig. 8 illustrates a distributed ledger in utilization management of an automatically generated model.

[0043] For example, a customer request from a maintenance company is to determine the timing of maintenance for the inverter ID-XX and battery module ID-YY, and a customer request from an infrastructure insurance company is to calculate the risk of failure for calculating insurance premiums for the inverter ID-XX and battery module ID-YY.

[0044] For example, the presented information includes a life prediction model ID for the inverter ID-XX and the battery module ID-YY, a cooling performance diagnosis model ID, a related surrogate model ID, and an input data ID. The presented information also includes, for example, a failure risk, anomaly cause candidates, and tokens for input data such as the operation history and load history of the inverter ID-XX and the battery module ID-YY.

[0045] As shown in Figure 8, the surrogate model to be utilized may be a model generated from a function candidate library based on a user request.

[0046] Next, we will explain the digital twin model IP (license & model) distribution and management function using NFT.

[0047] 9 is a diagram for explaining the configuration related to IP distribution and NFT management of the model utilization system 1 of the embodiment. The stored information of the distributed ledger (blockchain) of the model utilization system 1 of the embodiment is as follows. Model provider information Tag information (ID linking) · Scenario · Model · Data ID relevance information ·Model utilization history information ·Token information Model reliability, importance, risk, and value calculation information Transaction history information ·Owner information Other (e.g., information related to infrastructure failure risk calculations)

[0048] The tokens processed by the token processing unit 14 are traded on the IP distribution and NFT management platform.

[0049] Next, a method for extracting candidates for scenarios, models, and data to be utilized will be described in detail. Fig. 10 is a diagram for explaining an example of a method for extracting candidates for scenarios, models, and data according to an embodiment.

[0050] Digital twin models include, for example, simulation models, surrogate models, sensor models, BIM / CIM (Building / Construction Information Models), and control models.

[0051] The value calculation model is a model for calculating value indicators required by customers. For example, the value calculation model may be a carbon footprint calculation model, a predictive maintenance model, an energy consumption calculation model, a waste amount calculation model, or an economic efficiency calculation model.

[0052] Predictive maintenance models include life prediction models and failure probability prediction models.

[0053] A risk calculation model is a model that calculates the risk of not satisfying customer requirements and the tolerance range of value indicators. For example, when a customer requirement is to predict the risk of failure to calculate infrastructure insurance costs, or when a customer requirement is to formulate a predictive maintenance policy, the risk calculation model is a failure risk prediction model that calculates the probability of not satisfying the required lifespan and the loss cost.

[0054] The digital data provided by the data provider in FIG. 10 includes, for example, observation data, monitoring data, test measurement data, sensing data, simulation data, material property data, and learning data.

[0055] Next, a method for extracting candidates for scenarios, models, and data to be utilized will be described. Fig. 11 is a flowchart showing an example of the method for extracting candidates for scenarios, models, and data according to an embodiment.

[0056] First, the presentation control unit 12 searches for a combination of a model and a data set that are related to a hypothetical scenario based on the above-mentioned relevance information and the importance obtained from the utilization history of the model and data (step S1).

[0057] Here, the relevance information is calculated from the above-mentioned matrix, network, matrix, network, and relevance model that express the relevance of the statistical / probability model and AI / machine learning model. The presentation control unit 12 may calculate the relevance and importance from a relevance / importance calculation model that includes the importance of each scenario ID, model ID, and data ID in these relevance models.

[0058] Furthermore, as the scenario, model, and data are utilized, the presentation control unit 12 learns the relevance of the scenario, model, and data in the utilization, and updates the relevance calculation model and the importance calculation model. The presentation control unit 12 may also learn data related to the risk-value calculation results, usage history, and evaluation results from users.

[0059] Next, the token processing unit 14 calculates the token, and the calculation unit 13 calculates the risk and value (step S2).

[0060] Next, the presentation control unit 12 performs a tolerance determination of tokens and risk / value for the presented candidate scenarios, models, and data (step S3). That is, the presentation control unit 12 utilizes not only the importance and relevance obtained from the utilization history of the model / data, but also the token calculation model and value / risk calculation model to extract candidates for scenarios, models, and data so that they fall within the tolerance range.

[0061] If the tokens and risk / value for the presented candidate scenario, model, and data are outside the allowable range (step S3, No), return to step S1.

[0062] If the tokens and risk / value for the candidate scenarios, models, and data are within the allowable range (step S3, Yes), the presentation control unit 12 presents the presentation information (step S4). The presentation information includes, for example, candidates for scenarios to be utilized and candidates for model data (associations between model IDs and corresponding data IDs) that satisfy customer requirements. The presentation information may also include candidates for model data that are in high demand. The presentation information may also include candidates for data that will significantly contribute to improving the reliability of the model.

[0063] In addition, the presentation control unit 12 may register (update) in the distributed ledger not only utilization history data and data related to reliability for the utilized scenario ID, model ID, and dataset ID, but also data related to the relevance and importance matrix for each model ID and data ID, and data related to the calculation model.

[0064] Here, the presentation control unit 12 may set an assumed scenario ID in advance from among the scenario ID, model ID, and data ID, and extract and present candidates for the model ID and data set ID associated with that scenario ID.

[0065] Furthermore, the presentation control unit 12 presents candidates for models and data that have not yet been registered in the distributed ledger but that contribute to risk reduction when, for example, the value satisfies customer requirements but the risk is not within an acceptable range due to low model reliability, based on the value-risk calculation. In this case, the presentation control unit 12 may also present tokens that can be provided to the institution that provided the candidate model or data that contributes to risk reduction as an incentive.

[0066] Furthermore, the presentation control unit 12 may provisionally register models and data that have not yet been registered in the distributed ledger, and the calculation unit 13 may perform simulations such as provisional value calculations and risk calculations. That is, the presentation control unit 12 may include a provisional model / data registration function, and the calculation unit 13 may include a simulation function that utilizes the provisional model / data.

[0067] Next, we will explain the risk-value calculation method.

[0068] When the index is a probability distribution, such as when calculating an index using a value model while uncertainty is involved, the value refers to a representative value of the index, such as the expected value or average value of the probability distribution. The value model may be, for example, a value model index that outputs 1 when it is feasible to predict a physical quantity in response to a change in a model variable, or when it is feasible to predict a change in the probability of failure over time, and outputs 0 when it is not feasible.

[0069] On the other hand, risk refers to the probability of an event occurring in the tail of a probability distribution where the indicator is outside the acceptable range, the probability of an emergent and rare event occurring, or the probability of occurrence multiplied by the loss cost.

[0070] The calculation unit 13 may include a scenario simulation function when calculating risk and value.

[0071] In calculating the risk of a model, for example, the calculation unit 13 calculates an index expressed as a probability distribution by performing a simulation using a candidate model and candidate data to be input to the model based on a candidate scenario.The calculation unit 13 then calculates the risk based on the probability of an event occurring in the probability distribution where the index expressed as the probability distribution falls outside an acceptable range determined in accordance with a user request, or the value obtained by multiplying the probability of an event occurring in the probability distribution where the index expressed as the probability distribution falls outside the acceptable range by the loss cost.Furthermore, for example, the calculation unit 13 calculates the model reliability by an uncertainty assessment based on the index expressed as the probability distribution, and calculates a higher risk the lower the model reliability.

[0072] In calculating the value of a model, for example, the calculation unit 13 calculates an index represented by a probability distribution by performing a simulation using a model candidate and data candidates input to the model based on a scenario candidate. Then, the calculation unit 13 calculates the value based on at least one of the expected value and the average value of the index represented by the probability distribution. For example, the calculation unit 13 calculates model reliability by uncertainty assessment based on the index represented by the probability distribution, and calculates a higher value as the model reliability increases. For example, the calculation unit 13 calculates a higher first importance of the model candidate as the usage history of the model candidate increases, and calculates a higher second importance of the data candidate input to the model as the usage history of the data candidate input to the model increases, and calculates a higher value as the first and second importance increase.

[0073] Specifically, the inputs to the risk / value calculation method by the calculation unit 13 are scenario proposals (infrastructure system configuration proposal, system operation scenario proposal, maintenance scenario proposal, insurance scenario proposal, model / data selection proposal, etc.), and the outputs are the following (1) value calculation model, (2) risk calculation model, and (3) reliability calculation model.

[0074] For example, when extracting model candidates through a scenario simulation, the process is as follows.

[0075] (1) Value calculation model - Calculation of expected failure probability using life prediction simulation for calculating infrastructure insurance costs Calculating expected values of resilience indices through predictive maintenance simulation - Calculation of expected energy consumption through energy consumption prediction simulation for energy conservation Expected value calculation using carbon footprint calculation simulation to reduce CO2 emissions - Calculating expected values through waste volume calculation simulation for circular economy (waste reduction) - Calculation of expected value through economics simulation (system construction, operation, maintenance, insurance costs)

[0076] (2) Risk calculation model Calculation of the risk that the failure probability exceeds the acceptable range for calculating infrastructure insurance costs - Calculating the risk of failure during the life of an infrastructure system Calculation of the risk of exceeding the allowable energy consumption limit - Calculating the risk of exceeding the carbon footprint tolerance - Calculation of the risk of exceeding the allowable range of waste volume Calculation of allowable excess risk for infrastructure system construction costs, operation costs, maintenance costs and insurance costs

[0077] (3) Reliability calculation model - Model reliability calculation by evaluating the uncertainty of simulation models and surrogate models -Model reliability calculation by evaluating uncertainty in value and risk calculation

[0078] Here, we will explain the case where the digital twin model is a surrogate model, based on an example of its use in predictive infrastructure maintenance, including scenario simulation methods.

[0079] For example, we will explain the extraction of scenario, model, and data candidates for considering solutions for maintenance triage and infrastructure insurance, with the aim of reducing downtime and improving the resilience of urban elevator fleets. When predicting the degree of damage and risk of failure of elevators based on past earthquake history and anticipated future earthquakes, the maximum acceleration, maximum displacement, and duration of shaking of buildings and infrastructure structural responses to earthquakes become structural response indices for predicting the degree of damage and risk of failure of elevators.

[0080] To simulate this structural response index and predict the probability of failure, the company operating and maintaining the system will need a surrogate model for structural response simulation and failure probability prediction, in addition to the urban topography and ground model held by Institution A, the building model held by Institution B, the elevator structure held by Institution C, acceleration monitoring data for the building and ground held by Institution D, and image data capable of measuring building shaking held by Institution E.

[0081] This surrogate model may be prepared and saved in advance, or the presentation control unit 12 may automatically generate or update the surrogate model based on these data and models. The scenario simulation function in the calculation unit 13 generates a surrogate model that can simulate structural response to earthquakes and failure probability for scenarios such as target cities, proposed infrastructure structures, and load assumptions based on earthquake history, based on data and models held by multiple organizations.

[0082] Next, the calculation unit 13 executes a Monte Carlo simulation of the structural response based on the surrogate model, and calculates the failure probability from the structural response distribution such as maximum acceleration, maximum displacement, and shaking duration based on the surrogate model for predicting the occurrence probability of each failure mode constructed from past failure mode data and respective failure occurrence data.Since repair time and response costs differ for each failure mode, it becomes possible to consider solutions for maintenance triage and infrastructure insurance based on predicted data on changes over time in the failure probability of each failure mode.

[0083] In the risk and value calculation by the calculation unit 13, risk calculation, value calculation and model reliability calculation are performed for a group of scenarios assumed by the customer using a virtual scenario simulation function, for example, for customer requirements such as elevator failure probability, model reliability level and calculation time.

[0084] Next, we will explain how the digital twin model can be used as a control model.

[0085] For example, when a specific waveform that will have a long life is given using a surrogate model that estimates the structural response, the flow for identifying an observer is as follows: Consider the case where the state of the controlled object is estimated by the observer and the control input is calculated by the feedback controller.

[0086] A surrogate model for structural response prediction is used as the model of the controlled object used in the observer, and the state of the controlled object that cannot be detected from the state detected by the controlled object is estimated from the observer. The feedback controller calculates the control input using a specific waveform that will result in the target long life and the state of the controlled object estimated by the observer. In this way, by using a surrogate model for structural response prediction to make the waveform of the controlled object follow a given waveform that will result in a long life, it is possible to extend the life of the system.

[0087] That is, in the case of feedforward control, the model is a surrogate model used to calculate the control input of the feedforward control of the infrastructure system. In this case, the calculation unit 13 uses the surrogate model as an observer of the feedforward control to predict the state of the infrastructure system, performs a simulation to determine the control input, and calculates the risk and value of the surrogate model based on the results of the simulation.

[0088] In the case of feedback control, the model is a surrogate model used to calculate a control input for feedback control of the infrastructure system. In this case, the calculation unit 13 performs a simulation to determine the control input using the state of the infrastructure system and the surrogate model, and calculates the risk and value of the surrogate model based on the simulation results.

[0089] As another example, we will explain model predictive control when a calculation formula for how suitable a waveform is for a long life is given. In model predictive control, a model of the controlled object is used to generate candidate waveforms for the controlled object and the control input, and the optimal control input is calculated by calculating and optimizing the evaluation values of these waveforms. In this case, a surrogate model for predicting structural response is used as the model of the controlled object, and a calculation formula for how suitable a given waveform is for a long life is used to calculate the evaluation value, making it possible to control for a long life by utilizing the surrogate model of structural response.

[0090] In this way, the calculation unit 13 may perform a simulation when utilizing a control model. The simulation performed by the calculation unit 13 includes, for example, a simulation for identifying an observer using a surrogate model of structural response, a simulation for using a control model for calculation of feedforward control, and a simulation of model predictive control.

[0091] Next, a token calculation method for utilizing model data will be described. Fig. 12 is a flowchart showing an example of the token calculation method according to the embodiment. First, the token processing unit 14 acquires history data (step S11). For example, the history data includes the usage history of the model (for example, the request history of the digital twin model when using the digital twin model in a digital twin scenario).

[0092] Next, the token processing unit 14 evaluates the model based on the history data of step S11 (step S12). The model is evaluated based on, for example, the number of times the model is used and the importance of the model.

[0093] Next, the token processing unit 14 calculates tokens indicating incentives when the model and data related to the model are utilized, using a token calculation function that calculates tokens based on the evaluation in step S12 (step S13).

[0094] In order to prevent token fluctuations, the token calculation function may perform preprocessing or data conversion using a relaxation function such as a moving average or a sigmoid function. The token processing unit 14 also manages the parameter sets for the calculation function and preprocessing using a distributed ledger.

[0095] Next, a method for utilizing a digital twin model using a distributed ledger will be described using the case of a surrogate model as an example. Fig. 13 is a diagram showing an example (in the case of a surrogate model) of a method for utilizing a digital twin model using a distributed ledger according to an embodiment. The example in Fig. 13 shows processing when a request for surrogate model registration / update is made, when surrogate model information is input, when a customer request is input, and when model management data stored in the distributed ledger (blockchain) is searched.

[0096] Next, an example of a user interface (UI) that presents a scenario, a model, and data (data set) will be described. Fig. 14 is a diagram showing an example of a UI that presents presentation information according to an embodiment.

[0097] 14 shows an example in which multiple candidates are extracted for predictive maintenance of an infrastructure system. The presentation control unit 12 displays the scenario candidates on the display device and accepts selection of each candidate from the user: an operation scenario, a load scenario, a hazard scenario, and a maintenance scenario.

[0098] In the example of FIG. 14, the presentation control unit 12 presents, for the selected scenario, model data set candidates for performing damage probability prediction based on life prediction and abnormality sign diagnosis (cooling performance diagnosis).

[0099] In the example of FIG. 14, the presentation control unit 12 also presents a candidate set with a relevance level of 0.9 or more and an importance level of 0.9 or more, and a simulation result output for the assumed scenario.

[0100] Each candidate model dataset for predictive infrastructure maintenance also displays its relevance, importance, value index, risk index, and required tokens. For example, candidate model dataset No. 1 for predictive infrastructure maintenance has a relevance of 0.99, importance of 0.95, value of 1, risk of 0.1, and required tokens of 7.

[0101] For example, the value index for model dataset candidate No. 1 is 1, which is based on the output of the value model. For example, the output of the value model is based on whether or not it is possible to predict the temporal change in failure probability through life prediction and whether or not it is possible to predict the dependence of the temporal change in failure probability on cooling performance (both predictions are feasible: 1, either prediction is not feasible: 0).

[0102] As shown in Figure 14, the presentation control unit 12 may display display information on a display device or the like that accepts the selection of a scenario to be utilized from one or more candidate scenarios, and may display display information including combinations of candidate models and candidate data, and for each combination, tokens, value, and risk, depending on the selection of the scenario to be utilized.

[0103] As described above, in the model utilization system 1 of the embodiment, the distributed ledger processing unit 11 processes the distributed ledger that stores the scenarios used in analyzing the target system, the digital twin model used in the analysis, and the data to be input into the digital twin model. The token processing unit 14 calculates tokens that represent incentives for the institution that provided the digital twin model to be utilized and the institution that provided the data to be utilized. The presentation control unit 12 then searches the distributed ledger for candidate scenarios that satisfy the user's requests, candidate digital twin models that satisfy the user's requests, and candidate data that satisfy the user's requests, and presents the searched candidate scenarios, candidate digital twin models, and candidate data, as well as the tokens required to utilize the combination of the searched candidate digital twin models and candidate data.

[0104] As a result, the model utilization system 1 of the embodiment can utilize digital twin models more efficiently while ensuring credit enhancement. Specifically, the model utilization system 1 of the embodiment can realize, for example, (1) a mechanism that can ensure credit enhancement for model utilization, (2) a mechanism for calculating and distributing tokens that benefits institutions that provide data and models, and (3) a mechanism for creating important scenarios and extracting models and datasets related to the scenarios.

[0105] Finally, an example of the hardware configuration of the model utilization system 1 according to the embodiment will be described.

[0106] [Example of hardware configuration] 15 is a diagram illustrating an example of the hardware configuration of a model utilization system 1 according to an embodiment. The model utilization system 1 is realized by, for example, one or more computers (information processing devices) including a processor 301, a main storage device 302, an auxiliary storage device 303, a display device 304, an input device 305, and a communication IF 306. The processor 301, the main storage device 302, the auxiliary storage device 303, the display device 304, the input device 305, and the communication IF 306 are connected via a bus 310.

[0107] The processor 301 executes a program read from the auxiliary storage device 303 to the main storage device 302. The main storage device 302 is memory such as a read-only memory (ROM) and a random access memory (RAM). The auxiliary storage device 303 is a hard disk drive (HDD), a solid state drive (SSD), a memory card, or the like.

[0108] The display device 304 displays the above-mentioned presentation information, etc. The input device 305 accepts input from the user. Note that the model utilization system 1 does not necessarily have to include the display device 304 and the input device 305.

[0109] The communication IF 306 is an interface for communicating with other devices. If the model utilization system 1 does not have the display device 304 and the input device 305, the display function and input function of an external terminal connected via the communication IF 306 may be used.

[0110] The programs executed by the model utilization system 1 are provided as computer program products stored in installable or executable format files on computer-readable storage media such as CD-ROMs, memory cards, CD-Rs, and DVDs (Digital Versatile Discs).

[0111] In addition, the program executed by the model utilization system 1 may be stored on a computer connected to a network such as the Internet, and may be provided by being downloaded via the network.

[0112] Furthermore, the programs executed by the model utilization system 1 may be configured to be provided via a network such as the Internet without being downloaded.

[0113] Furthermore, the programs executed by the model utilization system 1 may be provided in advance by being stored in a ROM or the like.

[0114] The programs executed by the model utilization system 1 have a modular configuration that includes functions that can be realized by the programs, among the functional configuration of the model utilization system 1 described above. The functions realized by the programs are loaded into the main memory device 302 by the processor 301 reading and executing the programs from a storage medium such as the auxiliary memory device 303. In other words, the functions realized by the programs are generated on the main memory device 302.

[0115] Note that some or all of the functions of the model utilization system 1 may be realized by hardware such as an integrated circuit (IC). The IC is, for example, a processor that executes dedicated processing.

[0116] Furthermore, when each function is realized using a plurality of processors, each processor may realize one of the functions, or may realize two or more of the functions.

[0117] Although several embodiments of the present invention have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These novel embodiments can be embodied in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their modifications are included within the scope and spirit of the invention, and are also included in the scope of the invention and its equivalents as defined in the claims. [Explanation of symbols]

[0118] 1. Model utilization system 2 nodes 11 Distributed Ledger Processing Unit 12 Presentation control unit 13 Calculation section 14 Token Processing Unit 301 processor 302 Main storage 303 Auxiliary storage device 304 Display device 305 Input Device 306 Communication Interface 310 Bus

Claims

1. a distributed ledger processing unit that stores a scenario used in analyzing a target system, a digital twin model used in the analysis, and data to be input into the digital twin model in a distributed ledger; a token processing unit that calculates tokens for the digital twin model to be utilized and the data to be utilized using at least one of the number of times the digital twin model is used and the importance of the digital twin model; a presentation control unit that searches for candidate scenarios that satisfy user requests from the scenarios stored in the distributed ledger, searches for candidate digital twin models that satisfy user requests and candidate data that satisfy user requests from the scenarios stored in the distributed ledger based on relevance information with the candidate scenarios, and presents the searched candidate scenarios, the searched candidate digital twin models, and the searched candidate data, and presents the tokens required for using a combination of the searched candidate digital twin models and the searched candidate data; A model utilization system equipped with:

2. A calculation unit is further provided that calculates a risk calculated from the candidate digital twin model and a value indicating the usefulness of the candidate digital twin model based on the candidate scenario, The presentation control unit searches for candidates for the digital twin model whose risks and values satisfy the user request. The model utilization system according to claim 1.

3. The calculation unit calculates an index expressed as a probability distribution by performing a simulation using the candidate digital twin model and the candidate data based on the candidate scenario, and calculates the risk based on the probability of occurrence of an event in the probability distribution that causes the index to fall outside an acceptable range determined in accordance with the user request, or the value obtained by multiplying the probability of occurrence of an event in the probability distribution that causes the index to fall outside the acceptable range by a loss cost. The model utilization system according to claim 2.

4. The calculation unit calculates an index expressed by a probability distribution by performing a simulation using the digital twin model candidate and the data candidate based on the scenario candidate, calculates model reliability by uncertainty evaluation based on the index, and calculates the risk to be higher the lower the model reliability. The model utilization system according to claim 2.

5. The calculation unit calculates an index expressed by a probability distribution by performing a simulation using the digital twin model candidate and the data candidate based on the scenario candidate, and calculates the value based on at least one of an expected value and an average value of the index. The model utilization system according to any one of claims 2 to 4.

6. The calculation unit calculates an index expressed by a probability distribution by performing a simulation using the digital twin model candidate and the data candidate based on the scenario candidate, calculates model reliability by uncertainty evaluation based on the index, and calculates the value to be higher the higher the model reliability. The model utilization system according to any one of claims 2 to 4.

7. The calculation unit calculates a first importance of the digital twin model candidate to be higher the more the usage history of the digital twin model candidate is, calculates a second importance of the data candidate to be higher the more the usage history of the data candidate is, and calculates a value to be higher the higher the first and second importance are. The model utilization system according to any one of claims 2 to 4.

8. The presentation control unit displays, on a display device, display information for accepting a selection of a scenario to be utilized from one or more candidate scenarios, and, in accordance with the selection of the scenario to be utilized, displays, on the display device, display information presenting combinations of the candidate digital twin models and candidate data, and, for each combination, the token, the value, and the risk. The model utilization system according to any one of claims 2 to 4.

9. The distributed ledger processing unit performs at least one of the following processes: authentication of a user accessing the distributed ledger, access control of the distributed ledger according to the user's authority, consensus formation, transaction execution, fee determination, block registration in the distributed ledger, and mining of the distributed ledger. A model utilization system according to any one of claims 1 to 4.

10. the target system is an infrastructure system, the data includes at least one of a usage history of the infrastructure system, a load history of the infrastructure system, and monitoring data of the infrastructure system; The digital twin model is a surrogate model used for at least one of abnormality sign diagnosis of the infrastructure system, lifespan prediction of the infrastructure system, and failure risk prediction of the infrastructure system. A model utilization system according to any one of claims 1 to 4.

11. The surrogate model is a model generated from a function candidate library based on the user request. The model utilization system according to claim 10.

12. the target system is an infrastructure system, the digital twin model is a surrogate model used to calculate a control input for feedback control of the infrastructure system, the calculation unit performs a simulation to determine the control input using the state of the infrastructure system and the surrogate model, and calculates a risk and a value of the surrogate model based on a result of the simulation. The model utilization system according to claim 2.

13. the target system is an infrastructure system, the digital twin model is a surrogate model used to calculate a control input for feedforward control of the infrastructure system, the calculation unit uses the surrogate model as an observer for the feedforward control to predict the state of the infrastructure system, performs a simulation to determine the control input, and calculates the risk and value of the surrogate model based on the results of the simulation; The model utilization system according to claim 2.

14. a step in which the model utilization system stores in a distributed ledger a scenario used to analyze the target system, a digital twin model used in the analysis, and data to be input into the digital twin model; a step in which the model utilization system calculates tokens for the digital twin model to be utilized and the data to be utilized using at least one of the number of times the digital twin model is used and the importance of the digital twin model; the model utilization system searches for candidate scenarios that satisfy user requests from the scenarios stored in the distributed ledger, searches for candidate digital twin models that satisfy user requests and candidate data that satisfy user requests from the scenarios stored in the distributed ledger based on relevance information with the candidate scenarios, and presents the searched candidate scenarios, the searched candidate digital twin models, and the searched candidate data, and presents the tokens required for using a combination of the searched candidate digital twin models and the searched candidate data; How to utilize the model, including:

15. Computer, a distributed ledger processing unit that stores a scenario used in analyzing a target system, a digital twin model used in the analysis, and data to be input into the digital twin model in a distributed ledger; a token processing unit that calculates tokens for the digital twin model to be utilized and the data to be utilized using at least one of the number of times the digital twin model is used and the importance of the digital twin model; a presentation control unit that searches for candidate scenarios that satisfy user requests from the scenarios stored in the distributed ledger, searches for candidate digital twin models that satisfy user requests and candidate data that satisfy user requests from the scenarios stored in the distributed ledger based on relevance information with the candidate scenarios, and presents the searched candidate scenarios, the searched candidate digital twin models, and the searched candidate data, and presents the tokens required for using a combination of the searched candidate digital twin models and the searched candidate data; A program to function as a

Images