Information processing device, information processing method, and program

Abstract

To more accurately predict at least one of the system's performance metrics and reliability metrics. [Solution] In the information processing apparatus of this embodiment, the prediction unit predicts at least one of the performance indicator and reliability indicator of a device using a prediction model that predicts at least one of the performance indicator and reliability indicator by sampling probability distribution data of global variables and probability distribution data of local variables. The control unit controls the controllable variable value, the range of the controllable variable value, or the probability distribution of the controllable variable value among the local variables based on at least one of the performance indicator and reliability indicator. The simulator unit generates a glocal adapter management model that performs a simulation in which the prediction model predicts at least one of the performance indicator and reliability indicator based on the controlled local variables.

Description

Technical Field

[0004] , , , ,

[0005] , , , , ,

[0003] , , , ,

[0001] Embodiments of the present invention relate to an information processing apparatus, an information processing method, and a program. [[ID=​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​The information processing device of the embodiment comprises a variable acquisition unit, a prediction unit, a control unit, and a simulator unit. The variable acquisition unit acquires probability distribution data of global variables, which include at least one variable in a global system that includes the device to be adjusted as a component, and probability distribution data of local variables, which include at least one variable in the device. The prediction unit predicts at least one of the performance indicator and reliability indicator of the device using a prediction model that predicts at least one of the performance indicator and reliability indicator by sampling the probability distribution data of the global variables and the probability distribution data of the local variables. The control unit controls the controllable variable value, the range of the controllable variable value, or the probability distribution of the controllable variable value among the local variables based on at least one of the performance indicator and reliability indicator. The simulator unit generates a glocal adapter management model that performs a simulation in which the prediction model predicts at least one of the performance indicator and reliability indicator based on the controlled local variables. [Brief explanation of the drawing]

[0006] [Figure 1] A diagram illustrating the relationship between the probability of failure and performance indicators, lifespan indicators, performance criteria, and lifespan criteria. [Figure 2] A diagram showing an example of the functional configuration of the information processing device according to the first embodiment. [Figure 3] A diagram showing an example of the functional configuration of the generation unit of the first embodiment. [Figure 4] A diagram showing an example of a surrogate AI according to the first embodiment. [Figure 5] A flowchart showing an example of the overall flow of the information processing method of the first embodiment. [Figure 6] A diagram illustrating the control of the target being adjusted by the surrogate AI in the first embodiment. [Figure 7] A diagram illustrating the safety margin of the first embodiment. [Figure 8] A schematic diagram illustrating an example of controlling a target object using the surrogate AI of the first embodiment. [Figure 9] A flowchart illustrating an example of a method for generating a surrogate AI according to the first embodiment. [Figure 10] A diagram illustrating a modified example 1 of the first embodiment. [Figure 11] A diagram illustrating a second modified example of the first embodiment. [Figure 12] A diagram showing an example of the functional configuration of the generation unit in the second embodiment. [Figure 13] A diagram showing an example of the functional configuration of the generation unit in the third embodiment. [Figure 14] This figure shows an example of the configuration related to the automatic update function of the generation unit in the fourth embodiment. [Figure 15] A diagram showing an example of the hardware configuration of the information processing device of the first to fourth embodiments. [Modes for carrying out the invention]

[0007] Because the characteristics, installation environment, and degradation rate of device components vary from one product to another, devices and components are designed with a safety margin to account for these variations. However, the performance and degradation rate assumed during the design phase may differ significantly from the actual performance and degradation rate of individual units in the operating environment.

[0008] In such cases, the designed parameters may be inappropriate, potentially leading to a deterioration in performance and lifespan.

[0009] By combining information from both the global side (customer usage environment) and the local side (devices and components), and adjusting the design parameters of the devices and components to appropriate parameters, it becomes possible to provide a function that appropriately manages performance and lifespan to meet customer requirements.

[0010] However, the local-side information also includes the manufacturer's confidential information of the device components (e.g., tolerance of constituent materials and structural dimensions, etc.) and know-how (e.g., detailed relationships of the phenomenon responses to design variables such as load, etc.), and there is a lot of information that is not provided as the device component specifications.

[0011] Also, the customer system-side information also includes confidential information and know-how, etc. in the information regarding the installation environment and usage method, etc., and there are many cases where information cannot be provided to the manufacturer side. Therefore, it is often difficult to appropriately manage the performance and lifespan so as to meet the customer requirements by combining the information of both the global side and the local side.

[0012] Referring to the accompanying drawings below, embodiments of an information processing apparatus, an information processing method, and a program that can more accurately predict at least one of the performance index and the reliability index of the system will be described in detail.

[0013] (First Embodiment) First, the relationship between the damage probability, failure rate, or failure probability (Failure rate) regarding performance and lifespan, etc., and the performance index, lifespan index, performance criteria, and lifespan criteria will be described.

[0014] FIG. 1 is a diagram for explaining the relationship between the damage probability and the performance index, lifespan index, performance criteria, and lifespan criteria.

[0015] As shown in FIG. 1, the probability distribution of the damage probability is determined by the respective probability distributions of the index for predicting performance (performance index), the index for predicting lifespan (lifespan index), performance criteria, and lifespan criteria, as well as by their respective margin designs respectively.

[0016] Also, there are aging changes in the performance criteria, lifespan criteria, performance index, and lifespan index.

[0017] Furthermore, the probability distributions of performance metrics and lifespan metrics depend on both local device component design specifications and individual variation information, as well as global information. For example, global information includes how the customer system is used globally and information about the operating environment, such as the installation environment.

[0018] Furthermore, local manufacturers provide global customers with test results for devices and components under industry-standard accelerated reliability testing conditions. However, detailed information regarding the relationship between design variables, load conditions, and boundary conditions, as well as models related to probability distributions, often cannot be provided due to the inclusion of confidential information and know-how.

[0019] Furthermore, since the configuration, load conditions, installation conditions, environmental conditions, and operating conditions of the global customer system also contain confidential information and know-how, customers often cannot provide this data and probability distribution models to local manufacturers.

[0020] In the first embodiment, an embodiment of a performance and life management simulator (with model identification) using a glocal adapter management model will be described. Specifically, in the first embodiment, a battery system that controls a battery using a glocal adapter management model realized by surrogate AI (Artificial Intelligence) will be used as an example.

[0021] In this embodiment, information relating to the device itself, including the device to be adjusted by the information processing device 1, may be described as "local-side information," a system that includes the device as a component may be described as a "global system," and information relating to the global system may be described as "global system-side information."

[0022] Furthermore, a device component is each of the constituent elements of a device. In the first embodiment, the device is a battery storage system including a plurality of battery modules and a power electronics module. The device components are the battery modules and the power electronics module, which are constituent elements of the battery storage system.

[0023] [Example of functional configuration] Figure 2 shows an example of the functional configuration of the information processing device 1 of the first embodiment. The information processing device 1 of the first embodiment comprises a generation unit 11, a storage unit 12, and a surrogate AI 13.

[0024] The generation unit 11 generates a surrogate AI 13 (glocal adapter management model).

[0025] The memory unit 12 stores a surrogate model base (equipped with surrogate models for various phenomenon analyses such as electrical circuit analysis, thermal analysis, and lifetime analysis) and a phenomenon analysis database.

[0026] The surrogate AI 13 outputs at least one of a performance indicator and a reliability indicator under control conditions that satisfy the device conditions defined by the local variables described below, and the global system requirements defined by the global variables described below. The surrogate AI 13 also controls the parameters of the device under adjustment 2 based on the global and local variables.

[0027] Furthermore, if Surrogate AI13 fails to meet the requirements of the global system defined by the global variables, it modifies at least one of the device conditions defined by the local variables or the global system conditions defined by the global variables, and then re-predicts at least one of the performance indicators and reliability indicators.

[0028] The second set of components to be adjusted are the battery module and the power electronics module, which are components of the battery storage system.

[0029] Figure 3 shows an example of the functional configuration of the generation unit 11 of the first embodiment. The generation unit 11 of the first embodiment includes a global variable acquisition unit 111, a local variable acquisition unit 112, a surrogate model unit 113, a model identification unit 114, a prediction unit 115, a control condition setting unit 116, a control unit 117, and a simulator unit 118.

[0030] The global variable acquisition unit 111 acquires global variables that indicate system conditions, including performance and lifespan indicators, for the global adjustment target 2. The global variables are acquired from the customer system.

[0031] Global variables represent the performance and reliability indicator requirements, system configuration specifications, load conditions, operating environment conditions, and disturbances of the tuned-upon 2 in the global system in which the device component is a constituent element. For example, global variables include at least one of the following: a variable representing the device's requirements, a variable representing the device's system configuration specifications, a variable representing the device's load conditions, a variable representing the device's operating environment conditions, and a variable representing the device's disturbances.

[0032] The local variable acquisition unit 112 acquires local variables that indicate the local device component conditions. These local variables are obtained from the device component manufacturer.

[0033] Local variables represent the local device component conditions that constitute the system. Device component conditions are probability distributions that express structure / material properties / boundary conditions / initial conditions and individual differences. For example, local variables include at least one of the following: a variable indicating the structure of the device, a variable indicating the material properties of the device, a variable indicating the boundary conditions of the device, a variable indicating the initial conditions of the device, and a variable indicating individual differences of the device.

[0034] The global variable acquisition unit 111 and the local variable acquisition unit 112 may be implemented as a single variable acquisition unit.

[0035] The surrogate model unit 113 generates a surrogate AI 13 from the surrogate model base and the phenomenon analysis database described above. The surrogate AI 13 is a model that samples probability distribution data of global variables and probability distribution data of local variables and predicts performance and reliability indicators.

[0036] The surrogate model unit 113 generates a predictive model to be included in the surrogate AI 13 from a group of surrogate models, each containing one or more surrogate models. The surrogate model predicts at least one of the device's performance indicator and reliability indicator by sampling probability distribution data of global variables and probability distribution data of local variables.

[0037] The surrogate model unit 113 is implemented by Operator Learning of differential equations related to load conditions / structural conditions / material conditions / initial conditions / boundary conditions, and by a generative AI model. Specifically, the surrogate model unit 113 is an AI such as Neural Operator with Transformer or Normalizing Flow with VAE.

[0038] The surrogate AI 13 generated by the surrogate model unit 113 is adjusted by the model identification unit 114, prediction unit 115, control condition setting unit 116, control unit 117, and simulator unit 118, and then output from the generation unit 11.

[0039] The model identification unit 114 identifies global variables and model data of the surrogate AI 13 related to performance and reliability indicators from monitoring / measurement data for load patterns during trial operation or operation.

[0040] The prediction unit 115 predicts performance and reliability indicators by sampling probability distribution data of global variables and probability distribution data of local variables.

[0041] Performance and reliability metrics are used to evaluate, for example, the probability or risk of performance or reliability failing to meet required specifications. Specifically, performance and reliability metrics are expressed as failure rate, failure probability distribution, or risk value. For example, the risk value is expressed as the product of the loss cost (which indicates the magnitude of the loss) and the probability of the loss occurring.

[0042] The control condition setting unit 116 sets control conditions. The control conditions include global variables that can be controlled (for example, by model predictive control, consensus control, or control by a multi-agent model), correctable timing, and correctable range (or probability distribution).

[0043] The control unit 117 controls the controllable local variables of the surrogate AI 13 based on predicted performance and reliability indicators. For example, the control unit 117 performs model predictive control, multi-agent model-based control, or consensus control on the controllable local variables of the surrogate AI 13.

[0044] The simulator unit 118 executes the simulation using a surrogate AI 13 whose local variables are controlled by the control unit 117. Specifically, the simulator unit 118 takes into account the local device and component conditions and executes a simulation that outputs a combination of correction control variable values ​​(including the timing of correction control) that satisfies the performance and reliability index requirements in the global system, as well as the performance and reliability index itself.

[0045] The simulator unit 118 generates candidate combinations of control variables for appropriate operation (operation control scenarios) or candidate combinations of management variables for appropriate regeneration (regeneration scenarios), thereby aligning global customer requirements and customer usage environment conditions with local device component specifications and individual differences.

[0046] Furthermore, if the simulator unit 118 does not meet the requirements, it will modify the global or local conditions and rerun the simulation.

[0047] The generation unit 11 outputs a surrogate AI 13 adjusted by the simulator unit 118. The surrogate AI 13 is used as a glocal adapter to control the parameters of the adjustment target 2.

[0048] The following describes how to generate the glocal adapter management model (surrogate AI13).

[0049] Figure 4 shows an example of the surrogate AI 13 of the first embodiment. As a preliminary step, the surrogate model unit 113 described above prepares a surrogate model base (a group of surrogate models for each phenomenon) which includes one or more surrogate models.

[0050] Specifically, the surrogate model unit 113 first conducts a simulation parameter survey while generating sampling points for numerical experiments on local and global variables, and prepares a phenomenon simulation dataset. The phenomenon simulation dataset is stored in the phenomenon analysis database.

[0051] In the first embodiment, the local variables include variables of a battery storage system (e.g., a Traction Energy Storage System (TESS)) that includes multiple battery modules and power electronics modules. Specifically, the local variables include variables relating to the structure / material properties / boundary conditions / initial conditions of the battery modules and power electronics modules.

[0052] Furthermore, global variables include system condition variables of the battery storage system (e.g., system configuration specifications, load conditions, operating environment conditions, and disturbances).

[0053] Hereafter, when referring to both global and local variables collectively, the term "glocal variable" will be used.

[0054] Next, the surrogate model unit 113 generates a surrogate model base that includes one or more surrogate models that take partial glocal variables related to performance and reliability as input and output partial phenomenon indicators, based on a dataset of phenomenon simulation data (combinations of glocal variables and phenomenon result data).

[0055] Next, the surrogate model unit 113 takes all global and local variables related to the system (glocal variables) as input and utilizes a group of surrogate models that output performance and reliability metrics to generate a training dataset for the surrogate AI 13.

[0056] The inputs for Surrogate AI13 are the phenomenon metrics for Surrogate AI training, and all global and local variables related to performance and reliability metrics, which provide sampling data for global and local variables.

[0057] The output of Surrogate AI13 is the control variable for all phenomena indicators.

[0058] Here, performance and reliability indicators are metrics that quantitatively represent the performance or reliability of a device. For example, in the case of a battery, performance includes the battery's charge / discharge characteristics (C-rate, Depth of Discharge, etc.), cooling performance, power electronics performance, battery life, power electronics life, and the probability distribution (probability model and model parameters) that represents the uncertainty of these.

[0059] Next, the surrogate model unit 113 generates a surrogate AI 13 by training it with the spatial and temporal responses of phenomenal state variables related to performance and reliability, phenomenal criteria, and operators of differential equations using Neural Operators or Transformers. The differential equations to be trained include load conditions, boundary conditions, configuration / structural conditions, material property conditions, and initial conditions.

[0060] A key feature of the first embodiment is that, during the generation process of this surrogate AI13, constraints are generated by combinations of function candidates obtained from the function candidate library, thereby imposing regularity conditions to facilitate the convergence of the learning error of the surrogate AI13. The function candidate library includes physical and engineering models based on the phenomenon mechanism, as well as functions representing energy functionals and information entropy, etc.

[0061] Furthermore, during the configuration stage of Surrogate AI13, latent variables (features) may be extracted from glocal variables using generative AI methods such as Normalizing-flow VAE.

[0062] For example, latent variables (features) are features of performance and reliability indicators. Alternatively, latent variables (features) are features of phenomenon criteria. Furthermore, latent variables (features) are features of multivariate probability distributions relating to load conditions, boundary conditions, configuration / structural conditions, material property conditions, and initial conditions.

[0063] In this case, the glocal variables are converted into latent variables (features), and the input / output configuration of the surrogate AI13 is set to the latent variables (features) before the surrogate AI13 is generated.

[0064] During the learning process of the surrogate AI 13, if the learning error does not meet the requirements, the surrogate model unit 113 expands the aforementioned phenomenon simulation dataset and repeats the above process until the learning error meets the requirements.

[0065] The surrogate model unit 113 automatically generates a surrogate AI 13 that implements a glocal adapter through the above processing.

[0066] Next, we will explain the process used when simulating performance and reliability (lifetime) prediction and management using the surrogate AI 13 generated by the surrogate model unit 113.

[0067] Performance and reliability (lifetime) predictions and management are simulated using the model identified by the model identification unit 114.

[0068] The global variable acquisition unit 111 acquires global variables, including performance and reliability indicator requirements, system configuration specifications, load conditions, operating environment conditions, and disturbances, for the adjustment target 2 in the global system.

[0069] Requirements specifications refer to the specifications required by the customer to meet performance, reliability, and other requirements. For example, in the case of TESS, an example of requirements specifications is as follows: • Failure probability of 0.1% or less during the operational period (e.g., 20 years). • Failure probability of 0.1% or less during the period leading up to cascaded reuse (e.g., 10 years from operation). • Failure probability of system components within the period until replacement (e.g., 5 years from operation) is 0.01% • Performance degradation of 5% or less during the operating period (probability of occurrence 0.1% or less)

[0070] System configuration specifications refer to the specifications that make up a global system. For example, in the case of TESS, an example of system configuration specifications is as follows: • Series-parallel configuration of batteries • Configuration of battery cells / battery modules / battery packs / battery systems • Power electronics systems such as inverters, converters, and capacitors • Configuration of each battery pack • Air cooling system configuration

[0071] Load conditions refer to the load conditions on the devices and components of a system and its constituent elements. These load conditions include power waveforms during charging and discharging, State of Charge (SOC) waveforms, temporal and spatial temperature waveforms, and probability distributions representing the degree of uncertainty associated with them.

[0072] Operating environment conditions refer to the conditions of the installation environment in which the system is installed and used. Examples of operating environment conditions include the ambient temperature and humidity of the TESS installation environment.

[0073] Disturbances refer to the effects on a system caused by physical or chemical noise (with temporal and spatial distribution) from outside the system. In the case of TESS, examples of disturbances include electromagnetic noise and environmental vibrations.

[0074] The local variables are as follows: Variables related to the structure / material properties / boundary conditions / initial conditions of the battery cell. Variables related to the structure / material properties / boundary conditions / initial conditions of the battery module. • Parameters of the equivalent electrical network • Parameters of the thermal network • Degradation model parameters • Criteria variables (such as life definition variables)

[0075] In terms of performance, factors include the battery's charge / discharge characteristics (C-rate, Depth of Discharge, etc.), cooling performance, power electronics performance, battery life, power electronics life, and the probability distributions (probability models and model parameters) that represent the uncertainties associated with these factors.

[0076] The prediction unit 115 uses the surrogate AI 13 automatically generated by the surrogate model unit 113 described above to predict performance and reliability from the input information of glocal variables.

[0077] Specifically, first, Surrogate AI13 identifies the glocal variables and model parameters of the surrogate model related to performance and reliability indicators from monitoring / measurement data of load patterns during system trial operation or operation. Then, Surrogate AI13 samples probability distribution data of global variables and probability distribution data of local variables, and uses the identified glocal variables and model parameters of the surrogate model to predict performance and reliability indicators.

[0078] For example, in the case of the equivalent electrical network of each battery pack or each battery module, Surrogate AI13 identifies the surrogate model of the equivalent electrical network from the temperature-time transformation law and dynamic vibration response tests (including frequency changes) of the load pattern, by analogy to the method for identifying Maxwell models, which consist of multiple series and parallel resistors and capacitances.

[0079] Here, a hierarchical surrogate AI13 may be constructed by hierarchically classifying the battery modules and battery packs and identifying the surrogate model.

[0080] Next, we will explain the adjustment process for managing performance and lifespan using Surrogate AI13.

[0081] First, the control condition setting unit 116 sets the control conditions for the glocal variables that can be controlled (model predictive control, consensus control, or control by a multi-agent model), the timing for which correction control is possible, and the range (or probability distribution) for which correction is possible.

[0082] Next, the control unit 117 controls the controllable local variables among the local variables based on the prediction of performance and reliability indicators obtained by sampling.

[0083] Finally, the simulator unit 118 uses the surrogate AI 13, whose local variables are controlled by the control unit 117, to perform simulations for managing performance and lifespan (reliability).

[0084] Figure 5 is a flowchart showing an example of the overall flow of the information processing method of the first embodiment. First, the global variable acquisition unit 111 acquires global variables including customer system requirements (such as a threshold for the probability of damage), system configuration conditions, and load, usage environment, boundary conditions (including monitoring data) (step S1).

[0085] Next, the surrogate model unit 113 sets up a surrogate model to be used to predict the system's performance and lifespan indicators (e.g., probability of damage) identified by the global variables, based on a group of surrogate models that includes one or more surrogate models (step S2).

[0086] Next, the local variable acquisition unit 112 acquires the local variables used in the configured surrogate model (step S3). The local variables include the structural and material conditions of the device / component, detailed criteria information (including probability distributions), and tolerance information (including probability distributions), etc.

[0087] Next, the surrogate model unit 113 sets a control method for the surrogate AI 13 that predicts performance and lifespan indicators (step S4). For example, one control method is a multi-agent model.

[0088] When adopting a control method using a multi-agent model, each individual device may be modeled as an agent, or each individual device may be modeled as a separate agent for each performance and lifespan indicator, or the performance / lifespan calculation unit, control unit, and monitoring unit may be modeled as agents, or a combination of these may be modeled. In a multi-agent model, each agent receives environment variables and state variables of each agent, and performs actions such as predicting performance, lifespan, and failure rate, adjusting control variables, reusing devices, or performing device maintenance (fin cleaning or parts replacement, etc.) to optimize the value function of maximizing lifespan or minimizing failure rate under constraints (e.g., within a predetermined energy consumption limit, within a predetermined total cost limit, etc.).

[0089] The control method for the surrogate AI13 is not limited to those described above. For example, it may be controlled by consensus control or model predictive control.

[0090] When employing consensus control, each agent in a multi-agent model operates in such a way that some of its variables are aligned. The variables to be aligned for each agent may be performance, lifetime, and failure rate, or they may be local variables. As an example of how to determine the operation of each agent, one can calculate the difference between the agent of interest and the agent that is not of interest for the variable to be aligned, and multiply this difference by a coefficient to obtain the amount of operation for the agent of interest. This calculation is performed for each operating agent. When calculating the difference between the agent of interest and the agent that is not of interest, it is not necessary to use all agents other than the agent of interest as the agent that is not of interest; it is sufficient to use only one agent as the agent that is not of interest.

[0091] When employing model predictive control, the system's future waveform is estimated at each sampling stage, and the operating waveform is optimized accordingly. The current time step of the optimized operating waveform is used as the operating waveform for that sampling stage. A surrogate model may be used to estimate the system's future waveform. The value function used during optimization may include performance, lifetime, and failure rate waveforms, local variable waveforms, and the operating waveform. Constraints may be set on the system waveform and operating waveform during optimization.

[0092] Next, the simulator unit 118 manages the system's performance and lifespan using the surrogate model AI generated by the surrogate model unit 113 and adjusted by the model identification unit 114, prediction unit 115, control condition setting unit 116, and control unit 117 (step S5).

[0093] Figure 6 illustrates the control of the adjustment target 2 by the surrogate AI 13 in the first embodiment. The example in Figure 6 shows a case where a multi-agent model is set as the control method to manage performance and lifespan.

[0094] In the example in Figure 6, the surrogate AI13 comprises agents G and L.

[0095] Agent G is equipped with an API (Application Programming Interface) model 131. Agent G receives system requests (customer system requests) from the global system G as input from the monitoring data r.

[0096] Agent L comprises an adjustment unit 132, a feedback controller 133, and a prediction model 134. Agent L receives input from both the global and local variables (glocal variables) mentioned above. Specifically, Agent L receives input from the monitoring data r, including load variables, environment variables, and boundary conditions.

[0097] The parameters of API model 131 and prediction model 134 are identified by the model identification unit 114 described above.

[0098] API model 131 is a model that implements API functionality for modeling the customer interface. API model 131 models customer system usage conditions (load waveform conditions, environmental conditions, boundary conditions, and system configuration conditions) in a way that is consistent with the input format of prediction model 134.

[0099] For example, API model 131 creates a load model with amplitude, period, and load velocity as variables by spectrally analyzing the load waveform over a certain period in the customer's usage environment.

[0100] Predictive model 134 is a surrogate model that implements simulation and prediction functions. When glocal variables are input, predictive model 134 simulates and predicts the probability of damage related to the performance and lifespan indicators of the device component, taking into account the uncertainty of the glocal variables. Then, predictive model 134 outputs a combination of control variables and safety factors (margins) that meet customer requirements.

[0101] Furthermore, before predicting performance and lifetime, the prediction model 134 may utilize the input monitoring data r to identify some variables (or model parameters) of the surrogate model or lifetime model using data assimilation techniques, and then perform performance and lifetime predictions.

[0102] Figure 7 is a diagram illustrating the safety factor (margin) of the first embodiment. The surrogate AI 13 outputs a combination of control variables and a safety factor (margin) that satisfy the customer requirements. By setting the safety factor (margin) of adjustment target 2 to the safety factor (margin) output from the surrogate AI 13, the safety factor (margin) of adjustment target 2 can be adjusted to a more appropriate value.

[0103] Figure 8 is a schematic diagram illustrating an example of controlling the adjustment target 2 using the surrogate AI 13 of the first embodiment.

[0104] The surrogate AI 13's prediction model 134 includes a damage probability calculation unit 141 and a drive threshold calculation unit 142.

[0105] The adjustment target 2 comprises a battery module 21 and a power electronics module 22. The power electronics module 22 includes, for example, power electronics equipment and a cooler for cooling the power electronics equipment.

[0106] The damage probability calculation unit 141 receives the operating environment data (e.g., ambient temperature) as a global variable and the current waveform of the battery module 21 as a local variable.

[0107] The damage probability calculation unit 141 calculates the damage probability of the adjustment target 2 (e.g., container, battery pack, battery string, or battery module 21) based on the input global and local variables.

[0108] The drive threshold calculation unit 142 adjusts the drive threshold of the local variable according to the calculated damage probability.

[0109] For example, local variables include variables that affect the battery charge / discharge waveform, such as battery charge / discharge thresholds (upper and lower limits). Also, for example, local variables include variables that affect the drive waveform and thus the lifespan of the power electronics module 22 (e.g., drive current frequency). Furthermore, for example, local variables include variables related to cooling performance, such as cooling fan speed.

[0110] By adjusting the drive threshold shown in Figure 8 above, the following effects can be obtained, for example: • Containers that have not deteriorated significantly allow for rapid charging and discharging of batteries. • Reduces the energy required to drive the fan, thus saving energy. • In degraded containers and power electronics modules 22, the charge and discharge waveforms can be improved to extend their lifespan. • By increasing the fan's output and enabling more efficient cooling, the lifespan of adjustment target 2 can be extended.

[0111] Battery cells have individual differences, and the battery module 21 and battery pack also have individual differences in their charge and discharge characteristics. By identifying and understanding these individual differences using monitoring data during trial operation, and then predicting performance and lifespan using the surrogate AI 13 (glocal adapter), it becomes possible to present a combination of control variables that can handle diverse load patterns required by customers, while taking individual differences into account.

[0112] Figure 9 is a flowchart showing an example of a method for generating a surrogate AI13 according to the first embodiment.

[0113] First, the global variable acquisition unit 111 acquires probability distribution data of global variables that include at least one variable in the global system which includes the device of the device to be adjusted 2 as a component (step S11).

[0114] Next, the local variable acquisition unit 112 acquires probability distribution data of local variables, which include at least one variable in the device that is a component of the global system (step S12).

[0115] Next, the prediction unit 115 predicts at least one of the performance indicator and reliability indicator of the device using a prediction model 134 that predicts at least one of the performance indicator and reliability indicator by sampling the probability distribution data of global variables and the probability distribution data of local variables (step S13).

[0116] Next, the control unit 117 controls the controllable variable values, the range of controllable variable values, or the probability distribution of controllable variable values ​​among the local variables based on at least one of the performance indicator and the reliability indicator (step S14).

[0117] Next, the simulator unit 118 generates a glocal adapter management model (surrogate AI 13) that runs a simulation in which the prediction model 134 predicts at least one of the performance indicator and the reliability indicator based on controlled local variables (step S15).

[0118] As described above, the information processing device 1 of the first embodiment can more accurately predict at least one of the system's performance indicators and reliability indicators. Specifically, the glocal adapter management model (surrogate AI 13) can predict, for example, the probability of damage (probability distribution of control indicators) from information on global system conditions (global variables) and local device / component conditions (local variables).

[0119] Furthermore, the glocal adapter management model, during prediction, identifies the probability distributions of the performance criteria and lifetime criteria, as well as information regarding the margin design mentioned above, from the global and local variables.

[0120] For example, a glocal adapter management model allows you to input specific global variables and control the system so that the performance and lifespan failure probability values ​​meet the customer requirements specified by those global variables.

[0121] For example, a glocal adapter management model can output performance or lifetime damage probabilities by changing the values ​​of controllable local variables that affect the control indicator. In other words, a glocal adapter management model is a mathematical model that implicitly incorporates models (the surrogate models mentioned above) concerning the relationship and probability distribution of how local variables affect the control indicator.

[0122] The information processing device 1 of the first embodiment can be used, for example, in the "operation and maintenance" department for automatic device correction and predictive maintenance that are consistent with the customer's usage environment, thereby contributing to improved reliability.

[0123] (Modification 1 of the first embodiment) Next, Modification 1 of the First Embodiment will be described. In the description of Modification 1, explanations similar to those of the First Embodiment will be omitted, and the differences from the First Embodiment will be described. Modification 1 describes the case where the object to be adjusted 2 is, for example, an elevator, robot, automobile, motor, electronic equipment, and power electronics equipment such as an air conditioner.

[0124] Figure 10 is a diagram illustrating a modified example of the first embodiment. The surrogate AI 13 prediction model 134 includes a damage probability calculation unit 141 and a switching frequency calculation unit 143.

[0125] The damage probability calculation unit 141 receives the operating environment data (e.g., ambient temperature) as a global variable and the load waveform of the drive unit 23 as a local variable.

[0126] The damage probability calculation unit 141 calculates the damage probability of the adjustment target 2 based on the input global and local variables.

[0127] The switching frequency calculation unit 143 selects the switching frequency of the drive unit 23 circuit according to the calculated damage probability.

[0128] (Modification 2 of the first embodiment) Next, a modification 2 of the first embodiment will be described. In the description of modification 2, explanations similar to those of the first embodiment will be omitted, and the differences from the first embodiment will be described. Modification 2 describes the case where the adjustment target 2 is an HDD (Hard Disk Drive).

[0129] Figure 11 is a diagram illustrating a modified example of the first embodiment. The surrogate AI 13 prediction model 134 comprises a damage probability calculation unit 141 and a light permission threshold calculation unit 144.

[0130] The damage probability calculation unit 141 receives the usage environment data as a global variable and the local variables as described above.

[0131] Specifically, if the adjustment target 2 is an HDD, the system configuration specifications of the global variables would include the configuration of the device in which the HDD is installed (CPU, rack, etc.), the method of fixing the HDD, and the air cooling system configuration.

[0132] Load conditions for global variables include read / write commands, temporal and spatial temperature waveforms, and probability distributions representing the degree of uncertainty of these.

[0133] The usage environment data includes usage environment conditions and disturbances. Usage environment conditions include ambient temperature and humidity where the HDD is installed. Disturbances include acoustic disturbances from the cooling fan (fan vibration), as well as vibrations from other devices and the HDD.

[0134] Local variables include not only local variables related to the HDD's structure, material properties, boundary conditions, and initial conditions, but also control parameters in seek control (head 24 position waveform).

[0135] In terms of performance, factors include head position accuracy during reading and writing, IOPS, power consumption, and the probability distribution (probability model and model parameters) that represents the uncertainty of these factors.

[0136] Model parameters include the frequency transfer characteristics of vibrations and the parameters of the state-space model.

[0137] (Second Embodiment) Next, a second embodiment will be described. In the description of the second embodiment, explanations similar to those of the first embodiment will be omitted, and the differences from the first embodiment will be described. In the second embodiment, a configuration without the model identification unit 114 described above will be described.

[0138] Figure 12 shows an example of the functional configuration of the generation unit 11 of the second embodiment. The generation unit 11 of the second embodiment includes a global variable acquisition unit 111, a local variable acquisition unit 112, a surrogate model unit 113, a prediction unit 115, a control condition setting unit 116, a control unit 117, and a simulator unit 118.

[0139] The surrogate model unit 113 generates a surrogate AI 13 from the surrogate model base (group of surrogate models) and the phenomenon analysis database described above.

[0140] In the second embodiment, as shown in Figure 12, the surrogate AI 13 is adjusted by the prediction unit 115, the control condition setting unit 116, the control unit 117, and the simulator unit 118, without the model identification unit 114.

[0141] According to the second embodiment, it is possible to simulate the performance and lifespan management of device components in customer systems. This is useful, for example, in the "planning and sales" department to visualize the TCO (Total Cost of Ownership) advantages of device components. Also, for example, in the "design and development" department, the information processing device 1 of the second embodiment can be used to improve the reliability of device components under customer system conditions.

[0142] (Third embodiment) Next, a third embodiment will be described. In the description of the third embodiment, explanations similar to those of the first embodiment will be omitted, and the differences from the first embodiment will be described. In the third embodiment, a case will be described in which the performance required of the system includes environmental load, and environmental load prediction is also performed.

[0143] [Example of functional configuration] Figure 13 shows an example of the functional configuration of the generation unit 11 of the third embodiment. The generation unit 11 of the third embodiment includes a global variable acquisition unit 111, a local variable acquisition unit 112, a surrogate model unit 113, a model identification unit 114, a prediction unit 115, a control condition setting unit 116, a control unit 117, a simulator unit 118, a regeneration scenario candidate setting unit 119, and an environmental load prediction unit 120.

[0144] In the fourth embodiment, in addition to the configuration of the first embodiment, a regeneration scenario candidate setting unit 119 and an environmental load prediction unit 120 are added.

[0145] The regeneration scenario candidate setting unit 119 sets candidate regeneration scenarios that indicate the selection of cascaded reuse in the device component lifecycle, the timing of that selection, and the scope of cascaded reuse.

[0146] The global variable acquisition unit 111 acquires global variables that further include variables related to environmental load performance as global variables indicating the system configuration specifications.

[0147] The local variable acquisition unit 112 acquires local variables that further include variables related to environmental load performance as local variables indicating device component conditions.

[0148] The environmental load prediction unit 120 uses a life cycle simulator to predict the environmental load of the regeneration scenario candidates set by the regeneration scenario candidate setting unit 119, based on global and local variables.

[0149] The simulator unit 118 executes the simulation using a surrogate AI 13 whose local variables are controlled by the control unit 117. The simulator unit 118 of the third embodiment outputs a combination of correction control variable values ​​(including the timing of correction control) that further satisfies the environmental load requirements of the global system, taking into account the local device and component conditions, as well as performance and reliability indicators.

[0150] According to the third embodiment, it is useful for considering the regeneration (reuse / recycle / refurbish / dispose of) of a system. For example, the information processing device 1 of the third embodiment can be used to determine the timing of reuse and to generate optimal regeneration scenarios for life extension / reuse / refurbish / recycle.

[0151] Specifically, regarding regeneration scenarios such as the timing of cascaded reuse of battery modules 21, the surrogate AI 13 (glocal adapter management model) can be used to simulate performance, including environmental impact, and lifespan predictions, thereby enabling the presentation of regeneration scenarios.

[0152] (Fourth Embodiment) Next, the fourth embodiment will be described. In the description of the fourth embodiment, explanations similar to those of the first embodiment will be omitted, and the differences from the first embodiment will be explained. In the fourth embodiment, the case in which the glocal adapter management model (surrogate AI13) is automatically updated (automatically generated) will be described.

[0153] Furthermore, the fourth embodiment can be implemented in combination with the first to third embodiments described above.

[0154] [Example of functional configuration] Figure 14 shows an example of the configuration related to the automatic update function of the generation unit 11 of the fourth embodiment. The generation unit 11 of the fourth embodiment includes a data expansion unit 151, a surrogate model generation unit 152, a training dataset generation unit 153, a regularity condition generation unit 154, an extraction unit 155, a learning unit 156, an error determination unit 157, and a surrogate AI update unit 158.

[0155] The data enhancement unit 151 enhances the phenomenon simulation dataset by generating sampling points for numerical experiments on global and local variables while conducting a simulation parameter survey.

[0156] The surrogate model generation unit 152 generates a group of surrogate models, each of which takes partial glocal variables related to performance and reliability as input and outputs partial phenomenon indicators, based on the phenomenon simulation dataset.

[0157] The training dataset generation unit 153 uses the local and global variables input to the surrogate model group, as well as the performance and reliability metrics output from the surrogate model group, to prepare sampling data for the surrogate AI 13's training metrics, global variables, and local variables.

[0158] Local variables represent the structure, material properties, boundary conditions, and initial conditions of the device components that make up the system. Global variables represent the system configuration specifications, load conditions, operating environment conditions, and disturbances of the system in which the device components are made up.

[0159] The regularity condition generation unit 154 generates constraint conditions by combining a physical model based on the phenomenon mechanism, an engineering model, and function candidates obtained from a function candidate library. The function candidate library includes energy functionals and information entropy, etc. The constraint conditions are used to facilitate the convergence of the learning error of the surrogate AI 13.

[0160] The extraction unit 155 extracts phenomenal indicators and criteria related to performance and reliability, as well as features (latent variables) in multivariate probability distributions related to load conditions, boundary conditions, configuration / structural conditions, material property conditions, and initial conditions, using generative AI methods such as Normalizing-flow VAE.

[0161] The learning unit 156 learns the spatial and temporal responses of phenomenal state variables related to performance and reliability, phenomenal criteria, and operators of differential equations using Neural Operators, Transformers, etc. The differential equations represent loading conditions, boundary conditions, constitutive / structural conditions, material property conditions, and initial conditions.

[0162] If the error of the learned surrogate AI 13 does not meet the required specifications, the error determination unit 157 requests the data expansion unit 151 to expand the phenomenon simulation dataset.

[0163] If the error determination unit 157 determines that the error of the learned surrogate AI 13 satisfies the required specifications, it requests the surrogate AI update unit 158 ​​to update (generate) the surrogate AI 13.

[0164] The surrogate AI update unit 158 ​​updates the surrogate AI 13 if the error determination unit 157 determines that the error meets the required specifications.

[0165] According to Embodiment 4, the surrogate AI13 (glocal adapter management model), which operates as a performance and lifespan management simulator, can be continuously updated in accordance with the operating environment, customer requirements, or device / component updates.

[0166] Finally, we will describe an example of the hardware configuration of the information processing device 1 according to the first to fourth embodiments.

[0167] [Example hardware configuration] Figure 15 shows examples of the hardware configuration of the information processing device 1 according to the first to fourth embodiments. The information processing device 1 according to the first to fourth embodiments comprises a processor 201, a main memory 202, an auxiliary storage device 203, a display device 204, an input device 205, and a communication device 206. The processor 201, the main memory 202, the auxiliary storage device 203, the display device 204, the input device 205, and the communication device 206 are connected via a bus 210.

[0168] Note that the information processing device 1 may not be equipped with some of the above-described components. For example, if the information processing device 1 can utilize the input and display functions of an external device, the information processing device 1 may not be equipped with a display device 204 and an input device 205.

[0169] The processor 201 executes the program read from the auxiliary storage device 203 into the main memory device 202. The main memory device 202 is memory such as ROM and RAM. The auxiliary storage device 203 is such as an HDD (Hard Disk Drive) and a memory card.

[0170] The display device 204 is, for example, a liquid crystal display. The input device 205 is an interface for operating the information processing device 1. The display device 204 and the input device 205 may be implemented by a touch panel or the like that has both display and input functions. The communication device 206 is an interface for communicating with other devices.

[0171] For example, a program executed by the information processing device 1 is provided as a computer program product, recorded in an installable or executable file format on a computer-readable storage medium such as a memory card, hard disk, CD-RW, CD-ROM, CD-R, DVD-RAM, or DVD-R.

[0172] Alternatively, for example, the program executed by the information processing device 1 may be stored on a computer connected to a network 200 such as the Internet, and provided by downloading it via the network 200.

[0173] Alternatively, for example, the information processing device 1 may be configured to provide the program via a network 200 such as the Internet without requiring a download. Specifically, the information processing may be performed by a so-called ASP (Application Service Provider) type service, where the server computer does not transfer the program, but only issues execution instructions and retrieves the results to realize the processing function.

[0174] Alternatively, for example, the program for the information processing device 1 may be pre-installed and provided in ROM or the like.

[0175] The program executed by the information processing device 1 has a modular configuration that includes functions that can also be implemented by the program, as described above. In actual hardware terms, each of these functions is loaded onto the main memory 202 by the processor 201 reading and executing the program from the storage medium. In other words, each of the above function blocks is generated on the main memory 202.

[0176] Furthermore, some or all of the above-mentioned functions may be implemented using hardware such as an IC (Integrated Circuit) instead of software.

[0177] Alternatively, multiple processors 201 may be used to implement each function, in which case each processor 201 may implement one of the functions, or two or more of the functions.

[0178] While several embodiments of the present invention have been described, these embodiments are presented as examples only and are not intended to limit the scope of the invention. These novel embodiments can be carried out in a variety of other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their variations are included in the scope and spirit of the invention, as well as in the claims of the invention and its equivalents. [Explanation of Symbols]

[0179] 1. Information Processing Device 2. Items to be adjusted 11 Generation part 12 Storage section 13 Surrogate AI 21 Battery Modules 22 Power Electronics Modules 23 Drive unit 24 heads 111 Global Variable Acquisition Section 112 Local Variable Acquisition Section 113 Surrogate Model Section 114 Model Identification Unit 115 Prediction Section 116 Control Condition Setting Unit 117 Control Unit 118 Simulator Section 119 Regeneration Scenario Candidate Setting Section 120 Environmental Load Prediction Section 131 API Models 132 Adjustment section 133 Feedback Controller 134 Predictive Models 141 Damage Probability Calculation Unit 142 Drive threshold calculation unit 143 Switching frequency calculation unit 144 Light permission threshold calculation unit 151 Data Expansion Department 152 Surrogate Model Generation Unit 153 Training Dataset Generation Unit 154 Regular condition generator 155 Extraction part 156 Learning Department 157 Error judgment section 158 Surrogate AI Update Department 201 Processor 202 Main storage 203 Auxiliary storage device 204 Display device 205 Input device 206 Communication equipment 210 Bus

Claims

1. A variable acquisition unit that acquires probability distribution data of global variables, which include at least one variable in a global system that includes the device to be adjusted as a component, and probability distribution data of local variables, which include at least one variable in the device. A prediction unit predicts at least one of the performance indicators and reliability indicators of the device using a prediction model that predicts at least one of the performance indicators and reliability indicators of the device by sampling the probability distribution data of the global variables and the probability distribution data of the local variables. A control unit controls, based on at least one of the performance indicator and reliability indicator, the control unit controls the controllable variable value, the range of the controllable variable value, or the probability distribution of the controllable variable value among the local variables. A simulator unit that generates a glocal adapter management model that performs a simulation to predict at least one of the performance indicators and reliability indicators using the prediction model based on the controlled local variables, An information processing device equipped with the following features.

2. The global variable includes at least one of the following: a variable indicating the device's requirements, a variable indicating the device's system configuration, a variable indicating the device's load conditions, a variable indicating the device's operating environment conditions, and a variable indicating the device's disturbances. The information processing apparatus according to claim 1.

3. The local variable includes at least one of the following: a variable indicating the structure of the device, a variable indicating the material properties of the device, a variable indicating the boundary conditions of the device, a variable indicating the initial conditions of the device, and a variable indicating individual differences in the device. The information processing apparatus according to claim 1.

4. A surrogate model unit that generates the prediction model from a group of surrogate models, which includes one or more surrogate models that predict at least one of the performance indicators and reliability indicators of the device by sampling the probability distribution data of the global variables and the probability distribution data of the local variables, The information processing apparatus according to claim 1, further comprising:

5. The glocal adapter management model outputs at least one of the performance indicator and the reliability indicator under control conditions that satisfy the device conditions defined by the local variables and the requirements specifications of the global system defined by the global variables. The information processing apparatus according to any one of claims 1 to 4.

6. If the glocal adapter management model fails to meet the requirements of the global system defined by the global variables, it modifies at least one of the device conditions defined by the local variables or the global system conditions defined by the global variables, and then re-predicts at least one of the performance indicator and reliability indicator. The information processing apparatus according to claim 5.

7. The glocal adapter management model outputs at least one of the performance indicators and reliability indicators based on consensus control, model predictive control, or control by a multi-agent model. The information processing apparatus according to claim 1.

8. The glocal adapter management model outputs, as information indicating at least one of the performance indicator and the reliability indicator, the probability of device failure, the probability distribution of the probability of failure, or a risk value indicating the risk of the device. The information processing apparatus according to claim 1.

9. The aforementioned risk value is expressed as the product of the loss cost, which indicates the magnitude of the loss, and the probability of the loss occurring. The information processing apparatus according to claim 8.

10. The information processing device acquires probability distribution data of global variables, which include at least one variable in a global system that includes the device to be adjusted as a component, and probability distribution data of local variables, which include at least one variable in the device. The information processing device predicts at least one of the performance indicator and reliability indicator of the device using a predictive model that predicts at least one of the performance indicator and reliability indicator by sampling the probability distribution data of the global variable and the probability distribution data of the local variable. The information processing device controls, based on at least one of the performance indicator and the reliability indicator, a controllable variable value, a range of controllable variable values, or a probability distribution of controllable variable values ​​among the local variables. The information processing device generates a glocal adapter management model that performs a simulation to predict at least one of the performance indicators and reliability indicators using the prediction model, based on the controlled local variables. Information processing methods including

11. Computers, A variable acquisition unit that acquires probability distribution data of global variables, which include at least one variable in a global system that includes the device to be adjusted as a component, and probability distribution data of local variables, which include at least one variable in the device. A prediction unit predicts at least one of the performance indicators and reliability indicators of the device using a prediction model that predicts at least one of the performance indicators and reliability indicators of the device by sampling the probability distribution data of the global variables and the probability distribution data of the local variables. A control unit controls, based on at least one of the performance indicator and reliability indicator, the control unit controls the controllable variable value, the range of the controllable variable value, or the probability distribution of the controllable variable value among the local variables. A simulator unit that generates a glocal adapter management model that performs a simulation to predict at least one of the performance indicators and reliability indicators using the prediction model based on the controlled local variables. A program designed to function as such.

Images