Prediction device, prediction method, and program

Abstract

The prediction device (1) includes an acquisition control unit (11) that performs control to acquire a probability distribution indicating the probability that the health of the target equipment will continue, and a prediction unit (12) that predicts a transition in the health of the target equipment based on the probability distribution acquired by the acquisition control unit (11).

Description

[Technical field]

[0001] The present disclosure relates to a prediction device, a prediction method, and a program. [Background technology]

[0002] Asset management of infrastructure and other objects requires management throughout their entire lifespan. To achieve this, it is important to properly manage the object at each stage of its life cycle, and to minimize maintenance costs, it is necessary to repair or replace the object at the appropriate time.

[0003] For example, Patent Document 1 describes a technology that predicts the remaining lifespan of a part by referring to limit test results that indicate the life characteristics of the part that affect the soundness of the equipment, and selects replacement parts that minimize maintenance costs while ensuring the reliability of the equipment based on a maintenance plan, using the remaining lifespan of the part, the equipment lifespan, and cost information. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2023-47259 A Summary of the Invention [Problem to be solved by the invention]

[0005] In the conventional technology described in Patent Document 1, the remaining life of a part is predicted by referring to the limit test results, so there was a problem that the prediction accuracy decreases when there is a large discrepancy between the limit test results and the actual soundness value of the part. For example, even if the limit test results are referred to, the deterioration trend of the object can be grasped, but the change in the state of the object cannot be grasped, so the prediction accuracy of the soundness of the object is insufficient.

[0006] The present disclosure is intended to solve the above-mentioned problem, and has an object to provide a prediction device that can provide information useful for predicting the remaining lifespan of an object. [Means for solving the problem]

[0007] A prediction device according to the present disclosure is a prediction device that predicts a transition in evaluation of an object whose evaluation changes over time, and includes an acquisition control unit that controls acquisition of a probability distribution indicating a probability that the evaluation of the object will continue, and a prediction unit that predicts the transition in the evaluation of the object based on the acquired probability distribution. The transition of the evaluation is the probability that the evaluation of the object will continue over time. do. Effect of the Invention

[0008] According to the present disclosure, a probability distribution indicating the probability that the evaluation of an object will continue is obtained, and a transition in the evaluation of the object is predicted based on the obtained probability distribution. In this way, the prediction device according to the present disclosure can provide a prediction result of the transition in the evaluation of the object, which is information useful for predicting the remaining life of the object. [Brief description of the drawings]

[0009] [Figure 1] 1 is a block diagram showing an example of the configuration of a prediction device according to a first embodiment. [Diagram 2] 4 is a flowchart showing a prediction method according to the first embodiment. [Diagram 3] 2 is a block diagram showing an example of the configuration of a prediction unit included in the prediction device according to the first embodiment. FIG. [Figure 4] 13 is a flowchart illustrating an example of a prediction process. [Diagram 5] FIG. 13 is a diagram showing a simulation result of a transition in the health degree of an object. [Figure 6] 6A and 6B are diagrams showing prediction results of transition of health condition of an object in the first embodiment. [Figure 7] 1 is a block diagram showing an example of a hardware configuration for implementing the functions of a prediction device according to a first embodiment. [Figure 8] FIG. 11 is a block diagram showing an example of the configuration of a prediction device according to a second embodiment. [Figure 9] FIG. 11 is a block diagram showing an example of the configuration of a generation unit included in a prediction device according to a second embodiment. [Figure 10] 13 is a flowchart showing a probability distribution generating process according to the second embodiment. [Figure 11] FIG. 13 is a diagram showing the relationship between the duration of health of an object and a probability distribution. [Figure 12] FIG. 11 is a block diagram showing an example of the configuration of a prediction device according to embodiment 3. [Figure 13] FIG. 11 is a block diagram showing an example of the configuration of a prediction unit included in a prediction device according to embodiment 3. [Figure 14] 13 is a flowchart showing a prediction method according to the third embodiment. [Figure 15] FIG. 13 is a diagram showing an overview of a process for correcting a simulation result of a transition in the health degree of an object. [Figure 16] 16A and 16B are diagrams showing prediction results of transition of health condition of an object in the third embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] Embodiment 1 FIG. 1 is a block diagram showing an example of the configuration of a prediction device 1 according to the first embodiment. In FIG. 1, the prediction device 1 is a device for predicting the transition of the evaluation of an object that changes over time, and obtains a probability distribution indicating the probability that the evaluation of the object will continue, and predicts the transition of the evaluation of the object based on the obtained probability distribution. For example, the prediction device 1 obtains the probability distribution from an external device via a network. The prediction device 1 may also obtain information required for generating the probability distribution, and generate the probability distribution based on the obtained information. Furthermore, the prediction device 1 predicts the remaining life of the object based on the prediction result of the transition of the evaluation of the object. The prediction device 1 may also output information of the prediction result to the external device, so that the external device may predict the remaining life of the object based on the information of the prediction result.

[0011] The prediction device 1 may be a device connected to the display device 2 via a wired or wireless connection. The display device 2 displays information output from the prediction device 1. For example, the display device 2 acquires and displays information indicating the prediction result of the transition of the evaluation of the object and the remaining life of the object from the prediction device 1. By referring to the prediction result of the transition of the evaluation of the object and the remaining life of the object displayed on the display device 2, an inspection worker can repair or replace the object at a timing that minimizes maintenance costs. The display device 2 may be a display unit included in the prediction device 1, or may be a display unit included in a terminal communicatively connected to the prediction device 1 via a network.

[0012] (Overview of the object) The object may be anything that can be maintained and managed, and may include machines and machine parts, electrical and electronic equipment, transportation equipment, buildings and infrastructure facilities, energy facilities, medical equipment, and environmental facilities. Machines and machine parts include, for example, machine parts or devices, mechanical systems, and other machines used in manufacturing or industrial fields. This also includes parts such as motors, pumps, bearings, gears, and valves. Electrical and electronic equipment includes, for example, electrical or electronic equipment, controls, sensors, circuits, and other devices. This also includes computers, control panels, and communication equipment. Transportation equipment includes transportation equipment such as automobiles, trucks, trains, ships, and aircraft. This also includes engines, tires, brakes, suspensions, and aircraft airframes.

[0013] Buildings and infrastructure facilities include, for example, buildings or facilities, roads, bridges, dams, and other infrastructure structures. These structures are also strongly affected by the deterioration or aging of building materials or structural members. Energy facilities include, for example, energy facilities such as power plants, transmission lines, and substations. These facilities are composed of mechanical parts, electrical equipment, control systems, and the like, and life span management is important.

[0014] Medical devices include, for example, medical equipment and medical tools, as well as equipment. For example, x-ray machines, MRI scanners, artificial joints or surgical tools. Environmental facilities include, for example, water treatment plants, waste treatment facilities or drainage facilities. These facilities also include mechanical parts, electrical equipment, chemicals, etc.

[0015] (Evaluation Summary) The evaluation of an object is an evaluation that changes over time and is an index for evaluating the state of the object. For example, the evaluation may be a health level indicating the degree to which the object is healthy on an N-level scale. The health level is status information indicating the state of the object, which is recorded by periodic inspection work of the object. The health may indicate the health of the object in terms of its mechanical, electrical, chemical, or thermal condition. The mechanical health indicator may include, for example, an indicator of the degree of mechanical damage such as wear, fatigue, distortion, cracks, etc. of a mechanical part or device. Parameters such as vibration, temperature, voltage, or pressure of the object may also be included. The electrical health indicators include, for example, the operational status or electrical parameters of an electrical or electronic device. The health level indicating the chemical state may include, for example, parameters indicating the degree of degradation, corrosion, alteration, etc. of a chemical process or material. The thermal condition may include parameters that indicate the degree of temperature change or thermal stress of the component.

[0016] The evaluation of the object may be a deterioration degree indicating the degree of deterioration of the object, for example, in terms of a mechanical state, an electrical state, a chemical state, or a thermal state. In the following, an example will be described in which the target object is a target device such as an electronic device, and the evaluation is the soundness of the target device.

[0017] The prediction device 1 includes an acquisition control unit 11 and a prediction unit 12. For example, the prediction device 1 is realized by a computer. A memory included in the computer stores a program constituting an information processing application for realizing each function of the acquisition control unit 11 and the prediction unit 12. Each function of the acquisition control unit 11 and the prediction unit 12 is realized by a processor included in the computer executing the information processing application read from the memory.

[0018] (Acquisition control unit) The acquisition control unit 11 performs control to acquire the probability distribution. For example, when the prediction device 1 includes a communication unit not shown in FIG. 1 and is capable of communication connection with an external device not shown in FIG. 1 by the communication unit, the acquisition control unit 11 controls the communication unit to acquire the probability distribution from the external device. Examples of communication methods used for communication with the external device include Bluetooth (registered trademark), Wi-Fi, Zigbee (registered trademark), LoRa (registered trademark), and NFC (Near Field Communication). The communication unit is, for example, a wireless communication module compatible with any of these communication methods. The acquisition control unit 11 controls the communication unit to execute procedures such as pairing, network connection, and address designation to establish communication with the external device. When communication is established, the communication unit requests information indicating a probability distribution related to the target object from the external device, and the external device returns the requested information. The information received by the communication unit is acquired by the acquisition control unit 11.

[0019] (Summary of Probability Distribution) The probability distribution is made up of the probability of the duration of each health level of the target equipment. For example, when the health level of the target equipment is N=1 to 3, the probability distribution includes the probability of the duration of the health level at N=3, the probability of the duration of the health level at a value between N=3 and 2, the probability of the duration of the health level at N=2, the probability of the duration of the health level at a value between N=2 and 1, and the probability of the duration of the health level at N=1. The probability distribution is generated based on the expected value of the duration of each health level at the time of designing the target equipment and the actual value of the duration of each health level of the target equipment. Since the actual value of the duration of each health level of the target equipment is taken into account in the probability distribution, it is possible to make a prediction with a small deviation from the actual value of the health level.

[0020] (Prediction section) The prediction unit 12 predicts the transition of the health of the target device based on the probability distribution acquired by the acquisition control unit 11. For example, the prediction unit 12 acquires a probability distribution indicating the probability that the health of the target device will continue from the acquisition control unit 11. This probability distribution is estimated based on, for example, an inspection work history (past data or experience). For example, the probability that the health of the target device will continue is estimated based on the measurement results of the health of the target device included in the inspection work history.

[0021] Next, the operation of the prediction device 1 will be described. FIG. 2 is a flowchart showing the prediction method according to the first embodiment. The acquisition control unit 11 performs control to acquire a probability distribution indicating the probability that the health of the target device will continue (step ST1). The prediction unit 12 predicts the transition of the health of the target device based on the probability distribution acquired by the acquisition control unit 11 (step ST2). The prediction device 1 can provide prediction information on the transition of the health of the target device, which is useful for predicting the remaining life of the target device, by executing the prediction method shown in FIG.

[0022] As a method for predicting the transition of the health of the target device, there is a prediction by numerical calculation. For example, when the probability distribution is an exponential distribution, the prediction unit 12 calculates the probability of reaching each health level after a certain time by numerical calculation, and predicts the transition of the health of the target device based on the calculated probability.

[0023] A machine learning model may be used to predict the transition of the health of the target device. For example, when information indicating a probability distribution is input, the prediction unit 12 predicts the transition of the health of the target device using a machine learning model that outputs prediction information of the transition of the health of the target device. The prediction information includes at least one of a prediction result of the transition of the health of the target device and a prediction result of the remaining life of the target device. For example, the machine learning model may be a model such as Transformer, BERT, or GPT-3, which is machine-trained to receive information indicating a probability distribution as an input and output prediction information of the transition of the health of the target device. In this case, the prediction unit 12 may not include a simulation unit described later with reference to FIG. 3.

[0024] The machine learning model is stored in a storage unit (not shown in FIG. 1) included in the prediction device 1. The machine learning model may also be stored in an external storage device that is provided separately from the prediction device 1 and can be communicatively connected. In this case, the prediction unit 12 communicatively connects to the external storage device by controlling the communication unit. A probability distribution transmitted from the prediction unit 12 via the communication unit is input to the machine learning model stored in the external storage device, and prediction information output from the machine learning model is received by the communication unit and acquired by the prediction unit 12.

[0025] The prediction unit 12 may also predict the remaining life of the target equipment based on the predicted transition of the health of the target equipment. The remaining life is the remaining period from the current time until the representative value of the transition of the health reaches a certain value. The representative value of the transition of the health is the expected value, the median, or the mode in the transition of the health. The prediction result of the transition of the health of the target equipment calculated by the prediction unit 12 and information indicating the remaining life of the target equipment are output to the display device 2, for example. The display device 2 displays the prediction result of the transition of the health of the target equipment and the information indicating the remaining life of the target equipment output from the prediction device 1. By referring to the prediction result of the transition of the health of the target equipment and the information indicating the remaining life of the target equipment displayed on the display device 2, an inspection worker can repair or replace the target equipment at a timing that minimizes maintenance costs.

[0026] Variation 1. (Simulation of evaluation transition) The transition of the health state of the target device may be predicted by simulation. Fig. 3 is a block diagram showing an example of the configuration of the prediction unit 12 included in the prediction device 1. For example, as shown in Fig. 3, the prediction unit 12 may include a simulation unit 121 and a transition prediction unit 122. The simulation unit 121 simulates the transition of the health of the target device. The transition prediction unit 122 predicts the transition of the health of the target device based on the result of the simulation by the simulation unit 121. Similarly, when simulating the transition in the health of the target equipment, the simulation result of the transition in the health of the target equipment calculated by the prediction unit 12 and information indicating the remaining life of the target equipment are output to, for example, the display device 2. The display device 2 displays the simulation result of the transition in the health of the target equipment and the information indicating the remaining life of the target equipment output from the prediction device 1. By referring to the simulation result of the transition in the health of the target equipment and the information indicating the remaining life of the target equipment displayed on the display device 2, an inspection worker can repair or replace the target equipment at a timing that minimizes maintenance costs.

[0027] FIG. 4 is a flowchart illustrating an example of a prediction process. The simulation unit 121 performs a transition simulation process to simulate the transition of the health level of the target device (step ST1A). As a transition prediction process, the transition prediction unit 122 predicts a transition of the health level of the target device based on the simulation result A obtained by the simulation unit 121 (step ST2A).

[0028] For example, the simulation unit 121 simulates the transition of the health of the target device based on the probability distribution, and predicts the transition of the health of the target device based on the result of the simulation. For example, specifically, the simulation unit 121 simulates the possible transition of the health of the target device X times based on the probability distribution indicating the probability that the health of the target device will continue. Note that the number of simulations X may be any number that can obtain a plausible prediction result as the transition of the health of the target device, and may be, for example, about 10,000 times.

[0029] (Simulation using random numbers) The simulation unit 121 may generate random numbers based on a probability distribution and perform a simulation using the random numbers. For example, the simulation unit 121 generates random numbers, which are random values, from a probability distribution using a Monte Carlo simulation, and models the generated random numbers as changes in the health level. The simulation unit 121 randomly determines the increase or decrease in the health level of the target device at each point in time, and performs X number of simulations of the transition of the health level.

[0030] Next, the transition prediction unit 122 predicts the health of the target device that changes over time by analyzing the simulation results of the transition of the health of the target device, and predicts the future state of the target device based on the prediction results. For example, the transition prediction unit 122 calculates the probability of reaching each health level, the mode, median, or expected value of the health level at a specific time point by analyzing the simulation results. The expected value may include expected value (1), which is the average value of the health levels, or expected value (2), which is a value obtained by rounding expected value (1).

[0031] FIG. 5 is a diagram showing a simulation result of the transition of the health of the target device, and shows a simulation result A of the transition of the health of the target device by the simulation unit 121. In FIG. 5, the horizontal axis is the elapsed period (days), and the vertical axis is the health of the target device. The simulation result A is a plot of possible transitions of the health calculated by the simulation unit 121 performing X number of simulations based on the probability distribution. The simulation result A shown in FIG. 5 shows that, for example, when the elapsed period is up to 600 days, the health of the target device is highly likely to transition from N=5.0 to 4.0. In addition, when the elapsed period approaches 2000 days, the possibility of the health of the target device transitioning from N=2.0 to 1.0 becomes high.

[0032] For example, the simulation unit 121 generates pseudo-random numbers from a probability distribution of the continuation of the health level of the target device, and randomly simulates the change in the health level based on the generated pseudo-random numbers. This simulation is repeated a predetermined number of times X. In each simulation, the simulation unit 121 probabilistically determines whether the current health level will continue. For example, if the generated pseudo-random number is smaller than a threshold, the simulation unit 121 determines that the health level will not continue and randomly decreases the health level. On the other hand, if the generated pseudo-random number is equal to or greater than the threshold value, the simulation unit 121 determines that the soundness level will continue, and maintains the soundness level without any change. The prediction unit 12 may display the simulation result A shown in Fig. 5 on the display device 2. This provides the simulation result A, which shows how the health degree of the target device changes over time, as shown in Fig. 5.

[0033] Variation 2. (Evaluation prediction results are aggregated by time) The prediction unit 12 may tally up the prediction results of the transition of the health of the target device for each hour, and use the tally up results for each hour to calculate the probability of reaching each health level for each hour. For example, the prediction unit 12 may tally up the prediction results of the transition of the health of the target device for each hour, and use the tally up results for each hour to calculate the probability of reaching each health level for each hour. In addition, the prediction unit 12 may calculate an expected value, a median, a mode, or the like.

[0034] The prediction unit 12 may predict the future transition of the health of the target device based on a representative value that is a result of tallying up the health at each time. For example, if the average health tends to decrease over time, the prediction unit 12 predicts the future transition of the health based on this. The prediction unit 12 may also predict the remaining life of the target device based on the predicted transition of the health of the target device.

[0035] 5 for each hour, and calculates the probability of reaching the health level for each hour using the results of the hourly aggregation. The aggregation results obtained for each hour for the simulation result A include the probability of reaching each health level, the expected value, the median, or the most frequent value.

[0036] 6A and 6B are diagrams showing the results of prediction of the transition of the health of the target device. FIG. 6A is a graph showing the probability that the health of the target device will continue with respect to elapsed time. In FIG. 6A, the horizontal axis is the elapsed period (days), and the vertical axis is the probability that the health of the target device will continue. Area B1 is an area of ​​the probability of the period during which the health of the target device will continue at N=5 with respect to the elapsed period. Area B2 is an area of ​​the probability of the period during which the health of the target device will continue at N=4 with respect to the elapsed period. Area B3 is an area of ​​the probability of the period during which the health of the target device will continue at N=3 with respect to the elapsed period. Area B4 is an area of ​​the probability of the period during which the health of the target device will continue at N=2 with respect to the elapsed period. Area B5 is an area of ​​the probability of the period during which the health of the target device will continue at N=1 with respect to the elapsed period.

[0037] Fig. 6B is a graph showing the health of the target device versus elapsed time. In Fig. 6B, the horizontal axis is the elapsed period (days) and the vertical axis is the health of the target device. The solid line A1 shows the change in the average (expected) health at each time during the elapsed period. The dashed line A2 shows the change in the median health at each time during the elapsed period. Furthermore, the dashed-dotted line A3 shows the change in the mode of the health at each time during the elapsed period.

[0038] For example, the prediction unit 12 plots the probabilities included in the probability distribution acquired by the acquisition control unit 11 against the elapsed time period as shown in Fig. 6A. The prediction unit 12 calculates the average, median, and mode of the health degree of the target device at each time, and plots the calculated values ​​against the elapsed time period as shown in Fig. 6B. Display information for displaying the graph shown in Fig. 6A and the graph shown in Fig. 6B is generated by the prediction unit 12. The display device 2 displays the graphs shown in Fig. 6A and 6B based on the display information generated by the prediction unit 12. By referring to these, the transition of the health degree of the target device can be grasped.

[0039] Furthermore, the prediction unit 12 may calculate the remaining time until the representative value of the transition of health of the target device reaches a certain value as the "remaining life". For example, when the representative value of the transition of health is an average value, the prediction unit 12 calculates the remaining time from the current time T until the representative value reaches N=1 as the remaining life. For example, the prediction unit 12 may generate display information in which the remaining life is displayed on a graph shown in FIG. 6B. Based on this display information, the display device 2 displays the graph shown in FIG. 6B in which the remaining life is set. By referring to this graph, the remaining life of the target device can be grasped.

[0040] Variation 3. (Calculation of risk value) The prediction unit 12 may perform risk assessment regarding the risk of the target device for each time based on the probability of reaching each health level for each time. For example, the prediction unit 12 calculates the probability that the target device will reach each health level, thereby calculating the probability that the target device will fail, and calculates a risk assessment value of the target device failing. The risk assessment value can be calculated by multiplying the loss at the time of failure by the probability that the target equipment will fail, where the probability that the target equipment will fail is the probability that the health of the target equipment will reach the minimum value (e.g., N=1). The loss in the event of a breakdown is a loss that is expected to occur when the target device breaks down. For example, the loss in the event of a breakdown may be the amount of damage that occurs when the target device breaks down. The prediction unit 12 may generate display information on which the risk assessment value is displayed. The display device 2 displays the risk assessment value based on this display information. This enables the prediction device 1 to provide an assessment value that is useful for risk assessment of the target device.

[0041] Next, a hardware configuration for realizing the functions of the prediction device 1 will be described. FIG. 7 is a block diagram showing an example of a hardware configuration for implementing the functions of the prediction device 1. As shown in FIG. Each function of the acquisition control unit 11 and the prediction unit 12 included in the prediction device 1 is realized by a processing circuit. That is, the prediction device 1 includes a processing circuit for executing the processes of steps ST1 to ST2 shown in Fig. 2. The processing circuit may be a CPU (Central Processing Unit) that executes a program stored in a memory.

[0042] The acquisition control unit 11 acquires information indicating a probability distribution received from an external device by a communication unit included in the prediction device 1, for example, via the input interface 100. In addition, when information indicating a probability distribution is stored in a storage unit included in the prediction device 1, the acquisition control unit 11 reads and acquires the information indicating the probability distribution from the storage unit via the input interface 100. In this case, the prediction device 1 does not need to include a communication unit.

[0043] The prediction unit 12 outputs the prediction result of the transition of the evaluation of the object to the display device 2, for example, via the output interface 101. In addition, the prediction unit 12 may transmit the prediction result to an external device by controlling the communication unit via the output interface 101.

[0044] Each function of the acquisition control unit 11 and the prediction unit 12 included in the prediction device 1 is realized by software, firmware, or a combination of software and firmware. The software or firmware is written as a program and stored in the memory 103 .

[0045] The processor 102 reads out and executes a program stored in the memory 103 to realize the functions of the acquisition control unit 11 and the prediction unit 12 included in the prediction device 1. For example, the prediction device 1 includes a memory 103 for storing a program that, when executed by the processor 102, results in the processing of steps ST1 to ST2 shown in FIG. 2 being executed. These programs cause a computer to execute the procedure or method of the processing performed by the acquisition control unit 11 and the prediction unit 12. The memory 103 may be a computer-readable storage medium in which a program for causing a computer to function as the acquisition control unit 11 and the prediction unit 12 is stored.

[0046] Memory 103 may be, for example, a non-volatile or volatile semiconductor memory such as a Random Access Memory (RAM), a Read Only Memory (ROM), a flash memory, an Erasable Programmable Read Only Memory (EPROM), an Electrically-EPROM (EEPROM) (registered trademark), a magnetic disk, a flexible disk, an optical disk, a compact disk, a mini disk, a DVD, etc.

[0047] A part of the functions of the acquisition control unit 11 and the prediction unit 12 included in the prediction device 1 may be realized by dedicated hardware, and the other part may be realized by software or firmware. For example, the function of the acquisition control unit 11 may be realized by a processing circuit which is dedicated hardware, and the function of the prediction unit 12 may be realized by the processor 102 reading and executing a program stored in the memory 103. In this way, the processing circuit can realize the above functions by hardware, software, firmware, or a combination of these.

[0048] As described above, the prediction device 1 according to the first embodiment includes the acquisition control unit 11 that performs control to acquire a probability distribution indicating the probability that the health of the target device will continue, and the prediction unit 12 that predicts a transition in the health of the target device based on the probability distribution acquired by the acquisition control unit 11. In this way, the prediction device 1 can provide prediction information on the transition in the health of the target device, which is useful for predicting the remaining life of the target device.

[0049] In the prediction device 1 according to the first embodiment, the prediction unit 12 simulates the transition of the health of the target device based on the probability distribution and predicts the transition of the health of the target device based on the result of the simulation. This allows the prediction device 1 to provide the simulation result of the transition of the health of the target device.

[0050] In the prediction device 1 according to the first embodiment, the prediction unit 12 generates random numbers based on a probability distribution and performs a simulation using the random numbers. This allows the prediction device 1 to provide information useful for predicting the remaining life of a target device, that is, a simulation result of the transition of the health of the target device.

[0051] In the prediction device 1 according to the first embodiment, the prediction unit 12 tally up the prediction results of the transition of the health level for each hour, and calculates the probability of reaching the health level for each hour using the result of the tally for each hour. This allows the prediction device 1 to provide information useful for predicting the remaining life of the target device, which is the probability of reaching the health level for each hour.

[0052] In the prediction device 1 according to the first embodiment, the prediction unit 12 performs risk assessment on the risk of the target device for each time based on the probability of reaching the health level for each time. This allows the prediction device 1 to provide an assessment value that is useful for risk assessment of the target device.

[0053] The prediction method according to the first embodiment includes a step (ST1) in which the acquisition control unit 11 performs control to acquire a probability distribution indicating the probability that the health of the target device will continue, and a step (ST2) in which the prediction unit 12 predicts a transition in the health of the target device based on the probability distribution acquired by the acquisition control unit 11. Thereby, the prediction method according to the first embodiment can provide prediction information of the transition in the health of the target device, which is information useful for predicting the remaining life of the target device.

[0054] The computer that executes the program according to the first embodiment functions as an acquisition control unit 11 that performs control to acquire a probability distribution indicating the probability that the health of the target device will continue, and a prediction unit 12 that predicts a transition in the health of the target device based on the probability distribution acquired by the acquisition control unit 11. In this way, the computer that executes the program according to the first embodiment can provide prediction information on the transition in the health of the target device, which is information that is useful for predicting the remaining life of the target device.

[0055] Embodiment 2 The prediction device of embodiment 1 predicts the transition of an evaluation of an object based on a probability distribution indicating the probability that the evaluation of the object will continue, while the prediction device of embodiment 2 generates a probability distribution to be used for prediction.

[0056] FIG. 8 is a block diagram showing a configuration example of a prediction device 1A according to the second embodiment. In FIG. 8, the prediction device 1A is a device for predicting the transition of the evaluation of an object that changes over time, generating a probability distribution indicating the probability that the evaluation of the object will continue, and predicting the transition of the evaluation of the object based on the generated probability distribution. For example, the prediction device 1A acquires design information 3 or an inspection work history 4, which are information necessary for generating the probability distribution, and generates a probability distribution based on the acquired design information 3 or inspection work history 4. Note that, although FIG. 8 shows the generation unit 13 that acquires both the design information 3 and the inspection work history 4, the generation unit 13 may acquire either the design information 3 or the inspection work history 4, and generate a probability distribution using either the acquired design information 3 or the inspection work history 4.

[0057] (Design information overview) The design information 3 is a limit test result indicating the life characteristics of the object based on a limit test corresponding to the environment required for the object. For example, the design information 3 includes an expected value of the period during which the evaluation at the time of design will continue. The expected value of the period during which the evaluation at the time of design will continue is calculated by a limit test on the object. When the object is a target device such as an electronic device and the evaluation is the soundness of the target device, the design information 3 includes an expected value of the period during which each soundness of the target device at the time of design will continue. The design information 3 is stored, for example, in a storage unit included in the prediction device 1A, or in an external device with which the prediction device 1A can communicate.

[0058] (Overview of inspection work history) The inspection work history 4 is obtained by past inspection work on the target object, and includes performance values ​​for a period during which the evaluation of the target object continues. For example, when the target object is a target device such as an electronic device, and the evaluation is the health of the target device, the inspection work history 4 includes performance values ​​for a period during which each health of the target device continues. Note that, similar to the design information 3, the inspection work history 4 is stored, for example, in a storage unit included in the prediction device 1A, or in an external device with which the prediction device 1A can communicate. Below, an example will be described in which the target object is a target device such as an electronic device, and the evaluation is the health of the target device.

[0059] The prediction device 1A includes an acquisition control unit 11, a prediction unit 12, and a generation unit 13. For example, the prediction device 1A is realized by a computer. A memory included in the computer stores a program constituting an information processing application for realizing each function of the acquisition control unit 11, the prediction unit 12, and the generation unit 13. A processor included in the computer executes the information processing application read from the memory, thereby realizing each function of the acquisition control unit 11, the prediction unit 12, and the generation unit 13.

[0060] The acquisition control unit 11 performs control to acquire the probability distribution. For example, the acquisition control unit 11 controls the generation unit 13 to acquire the probability distribution generated by the generation unit 13. The prediction unit 12 predicts a transition of the health level of the target device based on the probability distribution acquired by the acquisition control unit 11. For example, the prediction unit 12 acquires the probability distribution generated by the generation unit 13 from the acquisition control unit 11.

[0061] (Generation part) The generation unit 13 generates a probability distribution indicating the probability that the health of the target device will continue. As shown in Fig. 8, the generation unit 13 acquires the design information 3 or the inspection work history 4, and generates a probability distribution based on the acquired design information 3 or the inspection work history 4. For example, the generation unit 13 generates a probability distribution of the duration of each health state based on the expected value of the duration of each health state at the time of design, which is included in the design information 3, and the actual value of the duration of each health state of the target equipment, which is included in the inspection work history 4. The probability distribution is estimated by Bayesian estimation, for example, on the assumption that the probability distribution of the duration of each soundness level follows a Weibull distribution, and the above information included in the design information 3 is used as prior information.

[0062] FIG. 9 is a block diagram showing a configuration example of the generating unit 13. As shown in FIG. 9, the generating unit 13 includes a design information acquiring unit 131, an inspection work history acquiring unit 132, and a probability distribution estimating unit 133. The design information acquiring unit 131 acquires the design information 3. For example, when the prediction device 1A includes a communication unit not shown in FIG. 9 and is capable of communicating with an external device not shown in FIG. 9 through the communication unit, the design information acquiring unit 131 acquires the design information 3 from the external device by controlling the communication unit. Also, when the design information 3 is stored in a storage unit included in the prediction device 1A, the design information acquiring unit 131 reads out and acquires the design information 3 from the storage unit. In this case, the prediction device 1A does not need to include a communication unit. Although FIG. 9 shows the generation unit 13 having both the design information acquisition unit 131 and the inspection work history acquisition unit 132, the generation unit 13 may also have either the design information acquisition unit 131 or the inspection work history acquisition unit 132.

[0063] The inspection work history acquisition unit 132 acquires the inspection work history 4. For example, similar to the design information acquisition unit 131, the inspection work history acquisition unit 132 acquires the inspection work history 4 from an external device by controlling a communication unit. Furthermore, when the inspection work history 4 is stored in a memory unit included in the prediction device 1A, the inspection work history acquisition unit 132 reads and acquires the inspection work history 4 from the memory unit. In this case, the prediction device 1A does not need to be equipped with a communication unit.

[0064] (Probability distribution is generated based on design information and inspection work history) The probability distribution estimation unit 133 estimates a probability distribution indicating the probability that each soundness level of the target device will be maintained based on at least one of the design information 3 and the inspection work history 4. For example, the probability distribution estimation unit 133 selects an appropriate probability distribution based on the expected value of the duration of each health state at the time of design, which is included in the design information 3, and the actual value of the duration of each health state of the target equipment, which is included in the inspection work history 4. The probability distribution may be an exponential distribution or a Weibull distribution. These distributions are suitable for modeling the duration of the health state. For example, the probability distribution is assumed to follow the Weibull distribution. The probability distribution estimation unit 133 calculates parameters of the selected probability distribution, and generates a probability distribution of the duration of the health of the target device using the calculated parameters. This probability distribution is a model of the probabilistic distribution of the duration of each health level of the target device.

[0065] (Probability distribution generated based on inspection work history) Furthermore, the probability distribution estimation unit 133 may generate the probability distribution based on the inspection work history 4 of the target equipment. For example, the probability distribution estimation unit 133 generates the probability distribution based on the performance values ​​of the periods during which each health level of the target equipment continues, which are included in the inspection work history 4. Generally, probability distributions include exponential distribution, Weibull distribution, normal distribution, etc. The probability distribution estimation unit 133 selects from these the distribution that best fits the actual value. Next, the probability distribution estimation unit 133 uses a statistical method such as maximum likelihood estimation or least squares method to estimate the parameters of the selected probability distribution. For example, when an exponential distribution is assumed, the average duration of the health state is estimated from past records. The probability distribution estimation unit 133 uses the estimated parameters to generate a probability distribution of the duration of the health of the target device. This makes it possible to model the probabilistic distribution of the duration of the health. The generated probability distribution is used to predict the future transition of the health of the device.

[0066] (Generate probability distribution based on design information) Furthermore, the probability distribution estimation unit 133 may generate the probability distribution based on the design information 3 for the target device. For example, the probability distribution estimation unit 133 generates the probability distribution based on the expected value of the duration of each health level at the time of design, which is included in the design information 3. The expected value of the duration for which each state of health at the time of design will continue is calculated based on prediction or evaluation of the target equipment at the design stage. The probability distribution estimation unit 133 selects an appropriate probability distribution based on the expected value. Generally, a log-normal distribution or an exponential distribution is used. Next, the probability distribution estimation unit 133 calculates parameters of the selected probability distribution, and generates a probability distribution for the duration of the target device's health using the calculated parameters. This probability distribution is a model of the probabilistic distribution of the duration of each health level of the target device. This makes it possible to predict the duration of the future health level. The probability distribution generated by the probability distribution estimation unit 133 is acquired by the acquisition control unit 11.

[0067] Furthermore, the generation unit 13 may generate a probability distribution of the target equipment using a machine learning model that outputs a probability distribution indicating the probability that the soundness of the target equipment will continue when at least one of the design information 3 and the inspection work history 4 is input. For example, a model such as Transformer, BERT, or GPT-3 is adopted as the machine learning model, and the machine learning is performed to receive at least one of the design information 3 and the inspection work history 4 as an input and output a probability distribution.

[0068] The machine learning model is stored in a storage unit (not shown in FIG. 8) included in the prediction device 1A. The machine learning model may also be stored in an external storage device that is provided separately from the prediction device 1A and can be connected for communication. In this case, the generation unit 13 controls the communication unit to connect for communication to the external storage device. At least one of the design information 3 and the inspection work history 4 transmitted from the generation unit 13 by the communication unit is input to the machine learning model stored in the external storage device, and the probability distribution output from the machine learning model is received by the communication unit and acquired by the acquisition control unit 11.

[0069] Next, the operation of the prediction device 1A will be described. FIG. 10 is a flowchart showing the probability distribution generation process according to the second embodiment. The design information acquisition unit 131 acquires the design information 3, and the inspection work history acquisition unit 132 acquires the inspection work history 4 (step ST1B). The probability distribution estimation unit 133 estimates a probability distribution based on the design information 3 and the inspection work history 4 (step ST2B). The prediction device 1A executes the probability distribution generating method shown in FIG. 10, thereby generating a probability distribution of the continuation of the health of the target device, which is used for predicting the transition of the health of the target device.

[0070] Fig. 11 is a diagram showing the relationship between the duration of health of a target device and the probability distribution. In Fig. 11, the horizontal axis is the duration for which each health of the target device continues, and the vertical axis is the probability that the health of the target device will continue. The dashed line C1 is the probability distribution estimated from the design information 3. The solid line C2 is the probability distribution estimated by the probability distribution estimation unit 133 from the design information 3 and the inspection work history 4. The bar graph D is the duration of the probability that the health will continue.

[0071] The probability distribution estimation unit 133 estimates the probability distribution shown by the solid line C2 based on the design information 3 and the inspection work history 4. For example, the probability distribution estimation unit 133 estimates the probability distribution shown by the solid line C2 by Bayesian estimation. First, the probability distribution estimation unit 133 acquires the expected value of the duration of each soundness at the time of design, which is included in the design information 3, and the actual value of the duration of each soundness of the target equipment, which is included in the inspection work history 4. In Bayesian estimation, it is necessary to select a prior distribution. The probability distribution estimation unit 133 sets parameters of the selected prior distribution based on the expected value of the duration of each soundness at the time of design, which is included in the design information 3. The probability distribution estimation unit 133 defines a likelihood function based on the actual value of the period during which the soundness of the target equipment continues, which is included in the inspection work history 4. The likelihood function is a function that indicates how likely the parameter values ​​of the model are when observed data are given. Next, the probability distribution estimation unit 133 calculates the posterior distribution by Bayesian estimation. The posterior distribution is a probability distribution of the parameters after taking into account the observed data. The prior distribution and the likelihood function are combined to calculate the probability distribution shown by the solid line C2 as the posterior distribution.

[0072] As described above, the prediction device 1A according to the second embodiment includes the generation unit 13 that generates a probability distribution. This allows the prediction device 1A to generate a probability distribution of the duration for which each health level of the target device will continue.

[0073] In the prediction device 1A according to the second embodiment, the generation unit 13 generates a probability distribution based on the inspection work history of the target equipment. This enables the prediction device 1A to generate a probability distribution of the duration for which each health level of the target equipment will continue.

[0074] In the prediction device 1A according to the second embodiment, the generation unit 13 generates a probability distribution based on the limit test result for the target device. This enables the prediction device 1A to generate a probability distribution for the duration of each health level of the target device.

[0075] Embodiment 3 The prediction device according to the first embodiment simulates the transition of the evaluation of the object based on a probability distribution indicating the probability that the evaluation of the object will continue, and predicts the transition of the evaluation of the object based on the result of the simulation. The prediction device according to the third embodiment corrects the result of the simulation based on the probability distribution using the obtained evaluation of the object.

[0076] Fig. 12 is a block diagram showing a configuration example of a prediction device 1B according to the third embodiment. In Fig. 12, the prediction device 1B is a device that predicts the transition of the evaluation of an object that changes over time. The prediction device 1B generates a probability distribution indicating the probability that the evaluation of the object will continue, simulates the transition of the evaluation of the object based on the generated probability distribution, corrects the result of the simulation, and predicts the transition of the evaluation of the object based on the corrected simulation result. Below, an example will be described in which the object is a target device such as an electronic device, and the evaluation is the soundness of the target device.

[0077] The prediction device 1B includes an acquisition control unit 11 and a prediction unit 12A. For example, the prediction device 1B is realized by a computer. A memory included in the computer stores a program constituting an information processing application for realizing each function of the acquisition control unit 11 and the prediction unit 12A. The processor included in the computer executes the information processing application read from the memory, thereby realizing each function of the acquisition control unit 11 and the prediction unit 12A.

[0078] The acquisition control unit 11 controls the acquisition of the probability distribution. For example, the acquisition control unit 11 acquires the probability distribution from an external device by controlling the communication unit. In addition, when information indicating the probability distribution is stored in a storage unit included in the prediction device 1B, the acquisition control unit 11 may read and acquire the information indicating the probability distribution from the storage unit. Furthermore, when the prediction device 1B includes a generation unit 13, the acquisition control unit 11 may acquire the probability distribution generated by the generation unit 13 by controlling the generation unit 13.

[0079] The prediction unit 12A performs a simulation of the transition of the health of the target device based on the probability distribution acquired by the acquisition control unit 11, corrects the simulation result so as to satisfy the past state transition of the target device included in the inspection work history 4A, and predicts the transition of the health based on the corrected simulation result. The inspection work history 4A includes information indicating that the target device has a certain health at a certain time. For example, the prediction unit 12A corrects the result of the simulation to a state in which the health value is a certain value at a certain time included in the inspection work history 4A so as to satisfy the past inspection work history 4A corresponding to the state transition of the target device up to now. Note that the inspection work history 4A is stored, for example, in a storage unit provided in the prediction device 1B, or stored in an external device with which the prediction device 1B can communicate.

[0080] Fig. 13 is a block diagram showing a configuration example of a prediction unit 12A included in the prediction device 1B. As shown in Fig. 13, the prediction unit 12A includes a simulation unit 121, a transition prediction unit 122A, and a simulation result correction unit 123. The simulation unit 121 performs a simulation of the transition of the health of the target device. For example, the simulation unit 121 generates random numbers from a probability distribution by a Monte Carlo simulation, and performs X times a simulation of possible transitions of the health using the generated random numbers.

[0081] The transition prediction unit 122A predicts the transition of the health of the target device based on the results of the simulation corrected by the simulation result correction unit 123. For example, the transition prediction unit 122 predicts the transition of the health from representative values ​​such as the probability of reaching each health level, the expected value of the health level, the median, or the most frequent value by aggregating the corrected simulation results at each time point. Furthermore, the transition prediction unit 122 may calculate the remaining life of the target device based on the predicted transition of the health level.

[0082] The simulation result correction unit 123 corrects the result of the simulation performed based on the probability distribution using the obtained health degree of the target device. For example, the simulation result correction unit 123 acquires the simulation result generated by the simulation unit 121, and acquires the health of the target equipment at a certain time from the inspection work history 4A. The simulation result correction unit 123 corrects the simulation result so as to satisfy the past state transition of the target equipment included in the inspection work history 4A.

[0083] Furthermore, the simulation result correction unit 123 may correct the simulation result using a machine learning model that outputs a corrected simulation result when the simulation result generated by the simulation unit 121 and the inspection work history 4A are input. For example, the machine learning model may be a model such as Transformer, BERT, or GPT-3, and is machine-trained to receive the simulation result and the inspection work history 4A as input and output a corrected simulation result.

[0084] The machine learning model is stored in a storage unit (not shown in FIG. 12) included in the prediction device 1B. The machine learning model may also be stored in an external storage device that is provided separately from the prediction device 1B and can be connected for communication. In this case, the simulation result correction unit 123 controls the communication unit to connect for communication to the external storage device. The simulation result and the inspection work history 4A transmitted by the communication unit are input from the simulation result correction unit 123 to the machine learning model stored in the external storage device, and the simulation result output from the machine learning model is received by the communication unit and acquired by the simulation result correction unit 123.

[0085] Next, the operation of the prediction device 1B according to the third embodiment will be described. FIG. 14 is a flowchart showing a prediction method according to the third embodiment. First, the simulation unit 121 performs a simulation of the transition of the health of the target device (step ST1C). Next, the simulation result correction unit 123 corrects the result of the simulation by the simulation unit 121 using the health of the target device acquired from the inspection work history 4A (step ST2C).

[0086] 15 is a diagram showing an overview of the simulation result correction process for the transition of the health degree of the target device, in which the horizontal axis represents the elapsed time (days) and the vertical axis represents the health degree of the target device. For example, the simulation result correction unit 123 acquires, from the inspection work history 4A, a health level E1 of the target equipment in the inspection work when a period of 300 days has elapsed, and a health level E2 of the target equipment in the inspection work when a period of 400 days has elapsed. Then, the simulation result correction unit 123 corrects the simulation result A shown in FIG. 5 so as to satisfy the acquired health levels E1 and E2. The simulation result corrected in this manner is the simulation result F shown in FIG. 15. The prediction unit 12A may cause the display device 2 to display the simulation result F shown in FIG. 15.

[0087] Next, the transition prediction unit 122A predicts the transition of the health level of the target device based on the simulation result F corrected by the simulation result correction unit 123 (step ST3C). For example, the transition prediction unit 122A aggregates the corrected simulation result F for each time period, and calculates the probability of reaching the health level for each time period using the results of the aggregation for each time period. The aggregation results for each time period of the corrected simulation result F include the probability of reaching each health level, an expected value, a median, or a mode.

[0088] FIG. 16A and FIG. 16B are diagrams showing the results of prediction of the transition of the health of the target device. FIG. 16A is a graph showing the probability that the health of the target device will continue with respect to elapsed time. In FIG. 16A, the horizontal axis is the elapsed period (days), and the vertical axis is the probability that the health of the target device will continue. Area G1 is an area of ​​the probability of the period during which the health of the target device will continue at N=5 with respect to the elapsed period. Area G2 is an area of ​​the probability of the period during which the health of the target device will continue at N=4 with respect to the elapsed period. Area G3 is an area of ​​the probability of the period during which the health of the target device will continue at N=3 with respect to the elapsed period. Area G4 is an area of ​​the probability of the period during which the health of the target device will continue at N=2 with respect to the elapsed period. Area G5 is an area of ​​the probability of the period during which the health of the target device will continue at N=1 with respect to the elapsed period.

[0089] The simulation result correction unit 123 corrects the simulation result shown in Fig. 5 based on the inspection work history P1 at time T1 and the inspection work history P2 at time T2. As a result, the area B1 shown in Fig. 6A becomes the area G1 shown in Fig. 16A, and the area B2 shown in Fig. 6A becomes the area G2 shown in Fig. 16A. The inspection work history P1 includes the health degree E1 of the target equipment at time T1, and the inspection work history P2 includes the health degree E2 of the target equipment at time T2.

[0090] Fig. 16B is a graph showing the health of the target device versus elapsed time. In Fig. 16B, the horizontal axis is the elapsed period (days) and the vertical axis is the health of the target device. A solid line A4 shows the change in the average (expected) health at each time during the elapsed period. A dashed line A5 shows the change in the median health at each time during the elapsed period. Furthermore, a dashed-dotted line A6 shows the change in the mode of the health at each time during the elapsed period.

[0091] For example, the simulation result correction unit 123 corrects the simulation result shown in FIG. 5 based on the inspection work history P1 at time T1 and the inspection work history P2 at time T2. As a result, the change in the average value (expected value) of the soundness at each time of the elapsed period shown by the solid line A1 in FIG. 6B becomes the change in the average value of the soundness at each time of the elapsed period shown by the solid line A4 in FIG. 16B. The change in the median value of the soundness at each time of the elapsed period shown by the dashed line A2 in FIG. 6B becomes the change in the median value of the soundness at each time of the elapsed period shown by the dashed line A5 in FIG. 16B. Furthermore, the change in the mode value of the soundness at each time of the elapsed period shown by the dashed line A3 in FIG. 6B becomes the change in the mode value of the soundness at each time of the elapsed period shown by the dashed line A6 in FIG. 16B.

[0092] The display information for displaying the graph shown in Fig. 16A and the graph shown in Fig. 16B is generated by the transition prediction unit 122A. The display device 2 displays the graphs shown in Fig. 16A and 16B based on the display information generated by the transition prediction unit 122A. By referring to these, it is possible to limit the health transition to those that the target device can take, and predicting the health transition based on the limited simulation results improves prediction accuracy.

[0093] Variation 1. (Predicting changes in evaluation using evaluation of object) Although the process of correcting the simulation result of the transition of the health of the target device has been described above, the prediction unit 12A may predict the transition of the health of the target device based on a probability distribution indicating the probability that the health of the target device will continue and the acquired health of the target device. For example, when predicting the transition of the health of the target device based on the probability distribution, the prediction unit 12A performs the prediction so as to satisfy the transition of the health of the target device so far acquired from the inspection work history 4A. This makes it possible to obtain a prediction result with little deviation from the actual value of the health. In this case, since the transition prediction unit 122A acquires the probability distribution and the inspection work history 4A to make a prediction, the prediction unit 12A only needs to be equipped with the transition prediction unit 122A, and does not need to be equipped with the simulation unit 121 and the simulation result correction unit 123.

[0094] Also, for example, the prediction unit 12A may calculate the probability of reaching each health level after a certain time by numerical calculation, and predict the transition of the health level of the target equipment based on the calculated probability. In this case, the prediction unit 12A performs the prediction so as to satisfy the transition of the health level of the target equipment so far acquired from the inspection work history 4A. For example, if the health level of the target equipment is N=5 from the first inspection until 200 days, the prediction of the transition of the health level of the target equipment is performed so that the health level until 200 days satisfies N=5.

[0095] Furthermore, the prediction unit 12A may predict the health transition of the target equipment using a machine learning model that outputs a prediction result of the health transition of the target equipment that satisfies the past health transition of the target equipment when a probability distribution indicating the probability that the health of the target equipment will continue and the past health transition of the target equipment included in the inspection work history 4A are input. For example, the machine learning model may be a model such as Transformer, BERT, or GPT-3, and is machine-trained to receive the probability distribution and the inspection work history 4A as input and output a prediction result of the health transition of the target equipment.

[0096] (Simulates the transition of evaluation using the evaluation of the object) Furthermore, the prediction unit 12A may simulate the transition of the health of the target device based on the probability distribution indicating the probability that the health of the target device will continue and the acquired health of the target device, and predict the transition of the health of the target device based on the result of the simulation. This makes it possible to limit the simulation result to one that satisfies the state of the target device, that is, a certain health at a certain time obtained from the inspection work history, and obtain a prediction result with little deviation from the actual value of the health.

[0097] For example, the prediction unit 12A acquires the past trends in the health of the target equipment from the inspection work history 4A, and simulates the past trends in the health of the target equipment based on the probability distribution so as to satisfy the acquired past trends in the health of the target equipment. Specifically, the simulation unit 121 performs X number of simulations of possible transitions in the health of the target device that satisfy the transitions in the health of the target device so far, based on a probability distribution indicating the probability that the health of the target device will continue. The number of simulations X may be any number that can obtain a plausible prediction result as the transition of the health of the target device, and may be, for example, about 10,000 times. The simulation may utilize the Monte Carlo simulation described in the first modification of the first embodiment. In this case, since the simulation unit 121 acquires the probability distribution and the inspection work history 4A and performs a simulation, the prediction unit 12A only needs to include the simulation unit 121 and the transition prediction unit 122A, and does not need to include the simulation result correction unit 123.

[0098] As described above, in the prediction device 1B according to the third embodiment, the prediction unit 12A corrects the result of the simulation performed based on the probability distribution using the acquired health of the target device. This allows the prediction device 1B to correct the simulation result A performed based on the probability distribution to the simulation result F limited to the health transitions that the target device could have taken in the past.

[0099] In the prediction device 1 according to the first embodiment, the prediction unit 12 predicts the transition of the health of the target device based on the probability distribution and the acquired health of the target device. This allows the prediction device 1 to provide information useful for predicting the remaining life of the target device, that is, prediction information of the health transition changed using the acquired health of the target device.

[0100] In the prediction device 1 according to the first embodiment, the prediction unit 12 simulates the transition of the health of the target device based on the probability distribution and the acquired health of the target device, and predicts the transition of the health of the target device based on the result of the simulation. This allows the prediction device 1 to provide information useful for predicting the remaining life of the target device, which is the simulation result of the transition of the health changed using the acquired health of the target device.

[0101] It is possible to combine the embodiments, modify any of the components of the embodiments, or omit any of the components of the embodiments. [Industrial Applicability]

[0102] A prediction device according to the present disclosure can be used, for example, for maintenance management of objects in various fields. [Explanation of symbols]

[0103] 1, 1A, 1B prediction device, 2 display device, 3 design information, 4, 4A inspection work history, 11 acquisition control unit, 12, 12A prediction unit, 13 generation unit, 100 input interface, 101 output interface, 102 processor, 103 memory, 121 simulation unit, 122, 122A transition prediction unit, 123 simulation result correction unit, 131 design information acquisition unit, 132 inspection work history acquisition unit, 133 probability distribution estimation unit.

Claims

1. A prediction device for predicting a transition of an evaluation of an object whose evaluation changes over time, comprising: an acquisition control unit that performs control to acquire a probability distribution indicating a probability that the evaluation of the object will continue; a prediction unit that predicts a transition of the evaluation of the object based on the acquired probability distribution; Equipped with The change in the evaluation is a probability that the evaluation of the object will continue over time. A prediction device comprising:

2. A prediction device for predicting a transition of an evaluation of an object whose evaluation changes over time, comprising: an acquisition control unit that performs control to acquire a probability distribution indicating a probability that the evaluation of the object will continue; a prediction unit that predicts a transition of the evaluation of the object based on the acquired probability distribution; Equipped with The evaluation has three or more different values. A prediction device comprising:

3. The prediction unit simulates a transition of the evaluation of the object based on the probability distribution, and predicts the transition of the evaluation of the object based on a result of the simulation.

3. The prediction device according to claim 1 or 2.

4. The prediction unit generates random numbers based on the probability distribution and performs the simulation using the random numbers. The prediction device according to claim 3 .

5. The evaluation is a health level indicating the degree to which the object is healthy.

3. The prediction device according to claim 1 or 2.

6. a generation unit for generating the probability distribution 3. The prediction device according to claim 1 or 2.

7. The generation unit generates the probability distribution based on an inspection work history for the object. The prediction device according to claim 6 .

8. The generating unit generates the probability distribution based on a limit test result for the object. The prediction device according to claim 6 .

9. The prediction unit predicts a transition of the evaluation of the object based on the probability distribution and the acquired evaluation of the object.

3. The prediction device according to claim 1 or 2.

10. The prediction unit simulates a transition of the evaluation of the object based on the probability distribution and the acquired evaluation of the object, and predicts the transition of the evaluation of the object based on a result of the simulation. The prediction device according to claim 9 .

11. The prediction unit corrects a result of the simulation performed based on the probability distribution using the obtained evaluation of the object. The prediction device according to claim 10 .

12. The prediction unit compiles a prediction result of the transition of the evaluation for each time period, and calculates a probability of reaching the evaluation for each time period using the compilation result for each time period.

3. The prediction device according to claim 1 or 2.

13. The prediction unit performs a risk assessment regarding a risk of the object for each time based on a probability of reaching the assessment for each time. The prediction device according to claim 12 .

14. A prediction method executed by a prediction device that predicts a transition of an evaluation of an object whose evaluation changes over time, comprising: An acquisition control unit performs control to acquire a probability distribution indicating a probability that the evaluation of the object will continue; A step of predicting a transition of the evaluation of the object based on the acquired probability distribution by a prediction unit; Equipped with The change in the evaluation is a probability that the evaluation of the object will continue over time. A prediction method comprising:

15. A prediction method executed by a prediction device that predicts a transition of an evaluation of an object whose evaluation changes over time, comprising: An acquisition control unit performs control to acquire a probability distribution indicating a probability that the evaluation of the object will continue; A step of predicting a transition of the evaluation of the object based on the acquired probability distribution by a prediction unit; Equipped with The evaluation has three or more different values. A prediction method comprising:

16. A program that causes a computer to function as a prediction device that predicts a transition of an evaluation of an object whose evaluation changes over time, comprising: The computer, an acquisition control unit that controls acquisition of a probability distribution indicating a probability that the evaluation of the object will continue; and a prediction unit that predicts a transition of the evaluation of the object based on the acquired probability distribution; The change in the evaluation is a probability that the evaluation of the object will continue over time. A program characterized by:

17. A program for causing a computer to function as a prediction device for predicting a transition of an evaluation of an object whose evaluation changes over time, comprising: The computer, an acquisition control unit that controls acquisition of a probability distribution indicating a probability that the evaluation of the object will continue; and a prediction unit that predicts a transition of the evaluation of the object based on the acquired probability distribution; The evaluation has three or more different values. A program characterized by:

Images