JPWO2020157927A1 - Diagnostic system and diagnostic method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide suitable judgment material regarding planning and execution of maintenance work or value of a maintenance target. SOLUTION: In a diagnostic system 100, a storage device 101 holding an observation value 125 of a predetermined event related to a maintenance object and a maintenance history 126 related to the maintenance object, and a predetermined observation value 125 and the maintenance history 126 as inputs. A configuration including an arithmetic device 104 that is applied to the machine learning algorithm of the above and specifies the correlation between the maintenance content indicated by the maintenance history 126 and the fluctuation content of the observed value 125 of the event due to the maintenance implementation in the maintenance content. And.

Description

The present invention relates to a diagnostic system and a diagnostic method.

Plants such as various factories and power generation facilities, or infrastructure such as water networks and power transmission networks are loaded according to their operating contents and surrounding environment, and may deteriorate or break down over time.

Therefore, in order to deal with such deterioration and failure, maintenance at regular intervals, that is, TBM (Time Based Maintenance) is mainly carried out. Since maintenance work naturally requires labor and cost, various conventional techniques for improving its efficiency and accuracy have been proposed.

For example, in a health management system that monitors the health condition of machinery and equipment, sensor data acquired from a plurality of sensors installed in the machinery and equipment, or environment data representing the sensor data and the installation environment of the machinery and equipment can be obtained. The equipment status of the machinery and equipment and the equipment status of the machinery and equipment by a statistical method using the time-series data acquisition unit acquired as time-series data and the normal data which is the time-series data acquired when the machinery and equipment are in a normal state as learning data. A state quantifying unit that quantifies the health condition indicating the performance or quality status of the mechanical equipment, an output unit that displays the quantified equipment condition and the quantified health condition, and / or outputs the quantified health condition to the outside. A health management system (see Patent Document 1), which is characterized by the above-mentioned features, has been proposed.

In addition, it is an abnormality detection / diagnosis method that detects an abnormality in a plant or equipment or a sign thereof and diagnoses the plant or equipment, and detects an abnormality in the plant or equipment based on data acquired from a plurality of sensors. Keywords are extracted from the maintenance history information of the plant or equipment, a diagnostic model of the plant or equipment is generated using the extracted keywords, and the plant or equipment is diagnosed using the generated diagnostic model. Abnormality detection / diagnosis method characterized by. (See Patent Document 2) and the like have also been proposed.

Japanese Unexamined Patent Publication No. 2015-088079 Japanese Unexamined Patent Publication No. 2011-227706

However, no method has been proposed in which the influence of various maintenance work and operation contents on the maintenance target is efficiently estimated in each aspect such as failure probability and required cost, and presented to the user. This means that useful judgment material is not provided regarding the timing and contents of appropriate maintenance work. In addition, the implementation status of the above-mentioned maintenance work and the transition of the failure probability due to it can be used as a material for determining the value of the maintenance target, but such information has not been provided.

Therefore, an object of the present invention is to provide a technique capable of providing suitable judgment material regarding the planning and execution of maintenance work or the value of a maintenance object.

The diagnostic system of the present invention that solves the above problems is a storage device that holds observation values of predetermined events related to the maintenance object and maintenance history of the maintenance object, and a predetermined machine that inputs the observation values and the maintenance history. It is characterized by including an arithmetic device that is applied to a learning algorithm and specifies the correlation between the maintenance content indicated by the maintenance history and the fluctuation content of the observed value of the event due to the maintenance implementation in the maintenance content. do.

In addition, the diagnostic system of the present invention performs maintenance on a predetermined maintenance object at a predetermined time with one or more maintenance contents, and with one or more operation contents, respectively. It is characterized by including an arithmetic unit that graphically displays the estimation information of the failure probability transition at least after the predetermined time in the maintenance target in at least one of the cases when the operation is carried out.

Further, in the diagnostic method of the present invention, the information processing system includes a storage device that holds an observed value of a predetermined event related to the maintenance object and a maintenance history related to the maintenance object, and inputs the observed value and the maintenance history. It is characterized in that it is applied to a predetermined machine learning algorithm to specify the correlation between the maintenance content indicated by the maintenance history and the fluctuation content of the observed value of the event due to the maintenance implementation in the maintenance content.

Further, in the diagnostic method of the present invention, when the information processing system performs maintenance on a predetermined maintenance object at a predetermined time with one or a plurality of maintenance contents, and one or a plurality of operation contents. It is characterized in that, when each operation is carried out, the estimation information of the failure probability transition at least after the predetermined time in the maintenance target object in at least one of them is displayed as a graph.

According to the present invention, it is possible to provide suitable judgment material regarding the planning and execution of maintenance work or the value of a maintenance target.

It is a network configuration diagram including the diagnostic system in this embodiment. It is a figure which shows the hardware configuration example of the diagnostic system in this embodiment. It is a figure which shows the configuration example of the sensor DB in this embodiment. It is a figure which shows the structural example of the measurement data in this embodiment. It is a figure which shows the configuration example of the event DB in this embodiment. It is a figure which shows the structural example of the operation DB in this embodiment. It is a figure which shows the structural example of the estimation result DB in this embodiment. It is a figure which shows the structural example of the incentive fluctuation standard DB in this embodiment. It is a figure which shows the structural example of the credit value DB in this embodiment. It is a figure which shows the flow example 1 of the diagnostic method in this embodiment. It is a figure which shows the flow example 2 of the diagnostic method in this embodiment. It is a figure which shows the flow example 3 of the diagnostic method in this embodiment. It is a figure which shows the output example 1 in this embodiment. It is a figure which shows the flow example 4 of the diagnostic method in this embodiment. It is a figure which shows the output example 2 in this embodiment. It is a figure which shows the flow example 5 of the diagnostic method in this embodiment. It is a figure which shows the flow example 6 of the diagnostic method in this embodiment. It is a figure which shows the example of the calculation formula of the credit value in this embodiment. It is a figure which shows the output example 3 in this embodiment.

--- Network configuration ---
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a diagram showing an example of a network configuration including the diagnostic system 100 of the present embodiment. The diagnostic system 100 shown in FIG. 1 is a computer system for making it possible to provide suitable judgment material regarding the planning and execution of maintenance work or the value of a maintenance target.

The diagnostic system 100 can be assumed to be, for example, an information processing system operated by a so-called O & M (Operation & Management) operator and used when formulating and executing a maintenance plan for maintenance objects. Of course, this operation mode and the like is an example, and may be operated and used by other businesses such as insurance companies and financial institutions that find business value in the maintenance target.

Such a diagnostic system 100 is connected to an appropriate network 10 such as the Internet or a dedicated line, and can perform data communication with a plant 200, a sensor unit 300, an insurance company system 400, and a financial institution system 500, which are maintenance objects. The diagnostic system 100 may include the sensor unit 300.

Of these, the plant 200 is an example of an object to be maintained as described above, and specifically, a power plant, a chemical plant, or the like can be assumed. The sensor unit 300 is arranged at or around a predetermined location of the plant 200, obtains observation values related to a sensing target, that is, an observation event, as measurement data 325 at regular intervals, and provides the observation values to the diagnostic system 100 via the network 10. Will be.

As the observation event of the sensor unit 300, various things such as temperature, vibration amount, rotation speed, current value, voltage value, flow rate, flow velocity, etc. of various mechanisms, workpieces, and conveyed objects in the plant 200 can be assumed. be.

Further, the insurance company system 400 is a system of a business operator who has concluded an insurance contract with the plant owner or the above-mentioned O & M (Operation & Management) business operator with respect to the plant 200 which is the maintenance target. The insurance covered by the contract here is insurance that compensates for damages caused by a failure or accident at the plant 200, and a predetermined insurance amount is paid when the event occurs.

Further, the financial institution system 500 is a system of a financial institution that provides a loan to a plant owner with the above-mentioned plant 200 as collateral. Therefore, this financial institution needs to be aware of the value of the plant 200 with high accuracy.

--- Hardware configuration example ---
The hardware configuration of the diagnostic system 100 is as shown in FIG. That is, the diagnostic system 100 includes a storage device 101, a memory 103, a CPU 104 (arithmetic logic unit), an input device 105, an output device 106, and a communication device 107.

Of these, the storage device 101 is composed of an appropriate non-volatile storage element such as a hard disk drive.

Further, the memory 103 is composed of a volatile memory element such as a RAM.

Further, the CPU 104 is an arithmetic unit that executes the program 102 held in the storage device 101 by reading it into the memory 103, etc., to perform integrated control of the system itself, and to perform various determinations, calculations, and control processes.

Further, the input device 105 assumes a keyboard, a mouse, and a microphone that accept a user's input operation.

Further, the output device 106 assumes a display or a speaker that outputs the processing result of the CPU 104 in an appropriate form such as an image or an audio.

Further, the communication device 107 is a network interface card or the like that is connected to the network 10 and is responsible for communication processing with other devices such as the sensor unit 300, the insurance company system 400, and the financial institution system 500.

--- Data configuration example ---
Subsequently, the information used by the diagnostic system 100 of the present embodiment will be described. FIG. 3 shows an example of the sensor DB 125 in this embodiment.

The sensor DB 125 is a database in which the diagnostic system 100 stores the observed values obtained from the sensor unit 300, that is, the measurement data 325 (FIG. 4).

The data structure is, for example, a collection of records composed of data such as a sensor ID that uniquely identifies the sensor unit 300 of a data resource, an observed event, and an observed value, using the measurement date and time as a key.

As shown in FIG. 4, the above-mentioned measurement data 325 has the same configuration as the record of the sensor DB 125.

Further, FIG. 5 is a diagram showing a configuration example of the event DB 126 in the present embodiment. The event DB 126 in this embodiment is a database that stores information on various events such as failures and accidents that occurred in the plant 200, and maintenance work that was performed in the plant 200.

Such information is, for example, input by the manager of the plant 200 by operating the input device 105, or a predetermined signal (corresponding to a failure or accident) transmitted from the sensor unit 300 as a diagnostic system. It can be assumed that 100 has acquired it.

The data structure is a collection of records consisting of data such as event type, occurrence location, and cost required for dealing with the occurrence date and time as a key.

Further, FIG. 6 is a diagram showing a configuration example of the operation DB 127 according to the present embodiment. The operation DB 127 in the present embodiment is a database that stores information on the operation contents actually set and operated in the plant 200.

The data structure is a collection of records composed of data such as a target and contents, with the target period as a key.

Further, FIG. 7 is a diagram showing a configuration example of the estimation result DB 128 in the present embodiment. The estimation result DB 128 in the present embodiment is a database in which the estimation results of the failure probability transition in the plant 200 are accumulated.

The data structure uses the device ID that uniquely identifies the device to be estimated as a key, and the maintenance content or operation content assumed for the device, and the cost when the maintenance content or operation content is performed (maintenance execution cost). , Failure cost, and accident cost), failure probability transition, and expected cost value, which is a collection of records.

Since the failure probability transition is a mathematical formula showing a line segment representing the time transition of the failure probability, in the example of FIG. 7, a file in which the mathematical formula is described is stored as a value.

In addition, the expected cost value exemplifies the expected cost value for each month, for example, one month later, two months later, and so on. This monthly time lapse may assume the most recent scheduled maintenance date to come next.

Further, in the estimation result DB 128, if the diagnostic method of the present embodiment is not implemented, the record is not stored.

Further, FIG. 8 is a diagram showing a configuration example of the incentive fluctuation reference DB 129 in the present embodiment. The incentive fluctuation standard DB129 in the present embodiment sets a threshold at which the premiums and claims to be applied fluctuate with respect to the failure probability of the insurance target, that is, the maintenance target, for the insurance contract held by the operator of the plant 200 described above. It is a database accumulated as a fluctuation reference value.

The data structure is based on data such as the insurance name, the reference value of the failure probability that is the threshold of insurance premium fluctuation, and the reference value of the failure probability that is the threshold of insurance claim fluctuation, using the insurance contract identification information such as the securities number as a key. It is a collection of records consisting of.

Of these, the reference values for failure probabilities related to insurance premium fluctuations are "maintenance" and "10% increase", and the standard values for failure probabilities related to insurance claims fluctuations are "maintenance" and "10". An example is shown that defines each condition such as "% decrease". In the case of "maintenance", it indicates the conditions under which the premium and insurance money at the time of contracting the insurance can be maintained. Further, in the case of "10% increase", it indicates a condition that the insurance premium at the time of contracting the insurance is increased by 10%. In addition, in the case of "10% reduction", it indicates a condition that the insurance money at the time of contracting the insurance is reduced by 10%.

Further, FIG. 9 is a diagram showing a configuration example of the credit value DB 130 in the present embodiment. The credit value DB 130 in this embodiment is a database that stores information on the credit value related to the plant 200. This credit value indicates the ratio of the failure probability to a predetermined level or less in the total operating time of the plant 200. The credit value is at least one calculated for the entire plant 200 and for each device included in the plant 200.

The data structure is a collection of records composed of data such as a device type, a target period, and a credit value, using a device ID that uniquely identifies the target device as a key.

--- Flow example 1 ---
Hereinafter, the actual procedure of the diagnostic method in the present embodiment will be described with reference to the drawings. Various operations corresponding to the diagnostic methods described below are realized by a program read by the diagnostic system 100 into the memory 103 and executed. Then, these programs are composed of codes for performing various operations described below.

FIG. 10 is a diagram showing a flow example 1 of the diagnostic method in the present embodiment. In this case, first, the sensor unit 300 in the plant 200 transmits the measurement data 325 held by the sensor unit 300 with respect to its own observation target to the diagnostic system 100 (s1).

On the other hand, the diagnostic system 100 receives the measurement data 325 from the sensor unit 300 described above (s2).

Further, the diagnostic system 100 generates a record including the measurement data 325 obtained in s2d (s3), stores the record in the sensor DB 125 (s4), and ends the process. The sensor DB 125 is generated by the processing up to this point.

--- Flow example 2 ---
FIG. 11 is a diagram showing a flow example 2 of the diagnostic method in the present embodiment. Next, the event DB generation flow will be described.

In this case, the sensor unit 300 of the plant 200 transmits the measurement data 325 to the diagnostic system 100 (s5).

On the other hand, the diagnostic system 100 collates the measurement data 325 obtained from the above-mentioned sensor unit 300 with a predetermined determination rule to determine whether or not an event has occurred (s6).

The above-mentioned judgment rule is, for example, a range at the time of normal, a range at the time of failure, and a range at the time of an accident in the observed value, which are predetermined for each observation event, and the measurement data 325 is included in any of these ranges. It is possible to assume a form in which the event occurrence status related to the observation event is determined based on whether or not it belongs.

As a result of the above determination, when the determination as normal is made (s7: N), the diagnostic system 100 ends the process.

On the other hand, when the occurrence of a failure or accident event is identified as a result of the above determination (s7: Y), the diagnostic system 100 determines the occurrence date and time, which is the measurement date and time indicated by the measurement data 325, and the event indicated by the event determination result. A record including each value of the type and the occurrence location based on the observation event indicated by the measurement data 325 is generated, stored in the event DB 126 (s8), and the process is terminated.

Event DB 126 will be generated by the processing up to this point.

--- Flow example 3 ---
FIG. 12 is a diagram showing a flow example 3 of the diagnostic method in the present embodiment. Next, a flow for estimating the failure probability transition in the plant 200 will be described.

In this case, the diagnostic system 100 applies the observation value indicated by the sensor DB 125 and the maintenance history or operation content indicated by the event DB 126 to the machine learning algorithm 110, and observes the operation content and the corresponding event by performing the operation. The correlation with the fluctuation content of the value or the correlation between the maintenance content indicated by the above-mentioned maintenance history and the fluctuation content of the observed value of the corresponding event due to the maintenance implementation in the maintenance content is specified (s10). As the correlation specified here, for example, when "Maintenance A" is performed as the maintenance content, the observation event "axis temperature" of the sensor "S1" is assumed to be "5% down".

Subsequently, the diagnostic system 100 applies the correlation information obtained in s10 to the failure probability evaluation engine 111, and at a predetermined time such as one month or two months from the present time, "maintenance A" and "maintenance B". At least when maintenance is carried out for each of multiple types of maintenance content such as "operation rate: 75%" and when operation is carried out for each of multiple types of operation content such as "operation rate: 75%" and "operation rate: 70%". In any of these, the transition of the failure probability in the plant 200 after the above-mentioned predetermined time is estimated (s11).

The above-mentioned failure probability evaluation engine 111 has a failure probability evaluation algorithm including an explanatory variable formula that predetermines the influence of an observation event in the sensor unit 300 on the failure probability of the plant 200 or its constituent devices with respect to the plant 200. Applicable.

The above-mentioned explanatory variable formula defines the failure probability Y by a polynomial in which a predetermined weighting coefficient (example: a to c) is added to a variable (example: X to Z) for inputting an observation value for each observation event. A form such as = aX + bY + cZ can be assumed.

The above-mentioned explanatory variable expression is an expression of the failure probability when the variation of the life (which can be said to be the probability of failure occurrence) indicated by the predetermined learned life model is minimized. This trained life model is a failure rate function that expresses the distribution of the probability of failure over time (eg, normal distribution).

This failure rate function is, for example, a function obtained by statistical processing based on the failure history and maintenance history of the plant 200, which is the maintenance target (information on failure events and maintenance events in the event DB 126).

In this case, the failure probability evaluation engine 111 of the diagnostic system 100 holds in advance, for example, the time corresponding to the current time (eg, the elapsed time from a predetermined time to the current time) before performing the maintenance work for each maintenance content. The failure probability Y1 is obtained by substituting the observation values of the latest observation events obtained at the present time into the explanatory variable equations.

Further, the failure probability evaluation engine 111 of the diagnostic system 100 specifies, for example, the fluctuation state of the observed value in each observed event when the maintenance work is performed for each maintenance content based on the above-mentioned correlation. That is, the observed value of each observed event after the maintenance work is carried out is estimated. Here, it is assumed that such fluctuations in the observed values occur one month after the maintenance work is carried out.

Subsequently, the failure probability evaluation engine 111 of the diagnostic system 100 performs the maintenance work according to each of the above-mentioned maintenance contents. Substitute (variable due to maintenance work) to obtain the failure probability Y2 one month after the present time.

The diagnostic system 100 can estimate the transition of the failure probability by sequentially executing such estimation of the failure probability for each period.

Subsequently, the diagnostic system 100 carries out maintenance for each of a plurality of types of maintenance contents such as "maintenance A" and "maintenance B" at a predetermined time such as one month or two months after the present time obtained in s11. Transition of failure probability at plant 200 in at least one of the cases where the operation is performed and when the operation is performed for each of multiple types of operation contents such as "operation rate: 75%" and "operation rate: 70%" ( The screen 1000) of FIG. 13 is displayed on, for example, the display device 106 (s12), and the process is terminated.

As illustrated on the screen 1000 of FIG. 13, the user can also specify each condition that defines the operation content when estimating the transition of the failure probability. Therefore, in order to accept this user designation, the diagnostic system 100 displays the interface 1001 on the screen 1000.

This interface 1001 accepts each item such as the operating rate of the plant 200, which is the maintenance object, the amount of materials used, and the genuine / non-genuine product which is an attribute of the component, as a user specification.

In this case, the diagnostic system 100 executes each step of the above-mentioned flow in the same manner when the operation is performed with the operation contents defined under the conditions specified by the user, and after the maintenance work and the like in the plant 200 are performed. The transition of the failure probability in

--- Flow example 4 ---
FIG. 14 is a diagram showing a flow example 4 of the diagnostic method in the present embodiment. Next, the calculation flow of the expected cost value will be described.

In this case, the diagnostic system 100 performs maintenance on the plant 200 for each maintenance content based on the cost at the time of maintenance indicated by the maintenance history in the event DB 126 and each cost of failure and accident. , And generate a table that defines the maintenance cost, the failure cost, and the accident cost at the plant 200 in at least one of the cases where the operation is performed for each operation content (s15).

Specifically, the average or median cost at the time of maintenance, the average or median cost at the time of failure, and the average or median cost at the time of an accident, which are shown in the maintenance history, are set for each device. Calculate for plant 200, and set this calculated value in each cost column of the estimation result DB128.

Subsequently, the diagnostic system 100 multiplies the failure probability at each period indicated by the transition of the failure probability obtained in the flow of FIG. 12 by the failure cost specified in the cost column of the estimation result DB 128 obtained in s15. The maintenance implementation cost in at least one of the cases where the maintenance is performed for each maintenance content and the operation for each operation content is added to the above-mentioned multiplication processing result, and the plant 200 is used. Calculate the expected cost value (s16).

The diagnostic system 100 shall calculate the above-mentioned expected cost value for each device in the plant 200.

Subsequently, the diagnostic system 100 displays the expected cost value obtained in s16 on the display device 106 (screen 1100 in FIG. 17) together with the transition of the failure probability, and ends the process (s17).

The example of the screen 1100 of FIG. 17 shows a mode in which the transition of the failure probability and the expected cost value are displayed together with respect to the “equipment A” and the “equipment B” in the plant 200.

--- Flow example 5 ---
FIG. 16 is a diagram showing a flow example 5 of the diagnostic method in the present embodiment. Next, the calculation flow of the future value of the plant 200, which is the maintenance target, will be described.

Here, the diagnostic system 100 is an insurance company for each of the useful life, the production amount per unit time, the downtime during maintenance, the annual premium when the failure probability of a predetermined standard is maintained, and the maintenance implementation opportunity in the plant 200. It is assumed that each information of the insurance money paid from the storage device 101 is stored in advance in the storage device 101.

In this case, the diagnostic system 100 receives the user designation for each of the maintenance content scheduled to be performed for the plant 200 and the failure probability that triggers the maintenance implementation in the maintenance content by the input device 105 (s20).

Subsequently, the diagnostic system 100 counts the frequency of the failure probability specified by the user in the above-mentioned s20 in the transition of the failure probability estimated in the flow example of FIG. 12 described above, and the counted frequency. Is multiplied by the above-mentioned useful life to calculate the number of maintenance operations (s21). Further, the diagnostic system 100 multiplies the downtime at the time of maintenance with the maintenance content specified by the user in s20 from the remaining operating time from the present time to the above-mentioned useful life by the number of maintenances obtained in s21. The total operating time is calculated by subtracting the value (s22).

Further, the diagnostic system 100 calculates the total production amount by multiplying the total operating time by the above-mentioned production amount per unit time (s23).

Further, the diagnostic system 100 calculates the total maintenance cost by multiplying the above-mentioned maintenance cost at the time of maintenance by the number of maintenances obtained in s21 (s24).

Further, the diagnostic system 100 calculates the total premium paid by multiplying the above-mentioned annual premium by the above-mentioned useful life (s25).

Further, the diagnostic system 100 calculates the total amount of insurance money by multiplying the above-mentioned insurance money for each maintenance implementation opportunity by the number of maintenance operations obtained in s21 (s26).

Subsequently, the diagnostic system 100 subtracts the total maintenance cost and the total premium paid in s24 from the total production amount obtained in s23 described above, and adds the total amount of insurance money obtained in s26 to the subtraction result. By adding, the future value of the plant 200 is calculated (s27).

Further, the diagnostic system 100 displays the future value on the display device 106 or transmits it to the insurance company system 400 and the financial institution system 500 (s28), and ends the process.

--- Flow example 6 ---
FIG. 17 is a diagram showing a flow example 6 of the diagnostic method in the present embodiment. Next, the calculation flow of the credit value will be described.

In this case, the diagnostic system 100 has a failure probability of a predetermined level or less in the total operating time based on the total operating time obtained in the flow of FIG. 16 and the transition of the failure probability obtained in the flow of FIG. (Example: 10% or less) is calculated as the credit value of the maintenance object (s30). Specifically, the credit value shall be calculated using the formula shown in FIG.

Further, the diagnostic system 100 displays the credit value obtained in s30 on the display device 106 or transmits it to the insurance company system 400 and the financial institution system 500 (s31), and ends the process.

The diagnostic system 100 refers to the incentive fluctuation standard DB129 (FIG. 8) for the insurance contracted for the plant 200, and the annual insurance to be paid to the insurance company when the failure probability of the predetermined standard is maintained in the plant 200. Information on the premium and the insurance money paid by the insurance company for each maintenance opportunity for the plant 200 is extracted, and based on this information, the annual premium and the incentive fluctuation line showing the threshold value at which the insurance money fluctuates (FIG. 18). It is preferable to display the line segments 1201 and 1202) on the screen 1200 of the above.

The incentive fluctuation line 1201 in FIG. 18 shows the reference values of the failure probability related to the premium fluctuation, which "maintain" the premium and the insurance money at present.

In addition, in the incentive fluctuation line 1202, the insurance premium is "10% increase" from the current standard value of the failure probability related to the insurance premium fluctuation, and "10" from the current standard value of the failure probability related to the insurance money fluctuation. It shows what is "% decrease".

According to this, for the person in charge of operation and maintenance of the plant 200, which is the object of maintenance, from the viewpoint of insurance premiums and insurance money, maintenance work is carried out at an appropriate timing and frequency, setting of operation contents, etc. Can clearly state the need. It can be expected that this will lead to an increase in maintenance awareness.

According to the diagnostic system and the diagnostic method of the present embodiment, the maintenance work and the operation are carried out by inputting the history of various events such as the maintenance work, the operation setting, and the failure in the plant and the sensor data related to the time. It is possible to estimate the relationship with the transition of sensor data accompanying the process using machine learning and present it to the user.

In addition, by applying such a relationship to a predetermined failure probability evaluation system, how the failure probability will change in the future depending on the selection of maintenance work and operation method, and how much cost will be required for maintenance and failure. It will be possible to simulate whether or not, and present the result to the user.

On the user side, it is possible to easily recognize the transition of the failure probability and the fluctuation of the cost for each of a plurality of maintenance / operation scenarios, and utilize it as a judgment material regarding the timing and contents of the maintenance work. This can lead to the transition from TBM to CBM (Condition Based Maintenance) for users who perform maintenance management of plants and the like, and as a result, maintenance costs can be reduced and optimized.

In addition, since the above-mentioned transition of the failure probability can be a variable factor of the insurance premium and the insurance money in the insurance contract related to the insurance target, the user can be presented with the fluctuation standard of the insurance premium and the insurance money related to the failure probability. It is expected to improve and maintain the motivation for maintenance work in Japan.

In addition, based on the above-mentioned transition of failure probability, the value of the plant to be maintained can be calculated and presented to insurance companies, banks, etc. for use in insurance product design work and loan project credit work. It will be possible.

That is, it is possible to provide suitable judgment material regarding the planning and execution of maintenance work or the value of the maintenance target.

The description herein reveals at least the following: That is, in the diagnostic system of the present embodiment, the storage device further holds the operation history of the maintenance object, and the arithmetic unit inputs the observation value and at least one of the maintenance history or the operation history. Applicable to a predetermined machine learning algorithm as, at least one of the operation content indicated by the operation history or the maintenance content indicated by the maintenance history, and at least one of the operation execution in the operation content or the maintenance execution in the maintenance content. It may be said that the correlation with the fluctuation content of the observed value of the above-mentioned event is specified.

According to this, in addition to the maintenance work on the plant or the like, it is possible to identify the correlation of the fluctuation of the observed value such as the sensor data based on the influence of the operation content of the plant or the like. This in turn makes it possible to provide more suitable judgment material regarding the planning and execution of maintenance work or the value of the maintenance target.

Further, in the diagnostic system of the present embodiment, the arithmetic device is based on a failure probability evaluation algorithm including a failure probability evaluation algorithm including a predetermined explanatory variable expression regarding the influence of the event on the failure probability of the maintenance object, and the correlation. It is also possible to estimate the transition of the failure probability of the maintenance object after the predetermined time when the maintenance is performed for each of the plurality of maintenance contents at the predetermined time.

According to this, it is possible to more specifically identify the influence of the maintenance work and the operation contents on the maintenance object on the fluctuation of the observed value in the form of the transition of the failure probability in the maintenance object. As a result, it becomes possible to provide more suitable judgment material regarding the planning and execution of maintenance work or the value of the maintenance target.

Further, in the diagnostic system of the present embodiment, the arithmetic unit performs maintenance for each of the one or a plurality of the maintenance contents at the predetermined time based on the explanatory variable expression and the correlation, and It may be possible to estimate the transition of the failure probability of the maintenance target object at least after the predetermined time in at least one of the operations when one or a plurality of operation contents are performed.

According to this, in addition to the maintenance work on the maintenance object, it is possible to estimate the transition of the failure probability in a form that can take into account the influence of the operation content of the maintenance object. As a result, it becomes possible to provide more suitable judgment material regarding the planning and execution of maintenance work or the value of the maintenance target.

Further, in the diagnostic system of the present embodiment, when estimating the transition of the failure probability, the arithmetic unit accepts a user designation of a predetermined condition that defines the operation content for each of the one or a plurality of operation contents, and the condition. It is also possible to estimate the transition of the failure probability of the maintenance object at least after the predetermined time when the operation with the operation content defined in the above is performed at the predetermined time.

According to this, it is possible to specify the transition of the failure probability in a form in which the conditions of the operation content are specified in detail. As a result, it becomes possible to provide more suitable judgment material regarding the planning and execution of maintenance work or the value of the maintenance target.

Further, in the diagnostic system of the present embodiment, the arithmetic unit accepts at least one of the operating rate of the maintenance object, the materials used, and the attributes of the component parts as the user designation of the predetermined condition. May be good.

According to this, it is possible to specify the transition of the failure probability in a form in which more specific conditions of the operation content are specified. As a result, it becomes possible to provide more suitable judgment material regarding the planning and execution of maintenance work or the value of the maintenance target.

Further, in the diagnostic system of the present embodiment, the storage device further holds information on each cost required when a past failure or accident occurs in the maintenance object, and the arithmetic device has the maintenance history. Based on the cost at the time of performing maintenance and the costs of the failure and the accident, the maintenance target is to be maintained for each of the above-mentioned one or a plurality of the above-mentioned maintenance contents, and the above-mentioned 1 Or, it is assumed that a table is generated that defines the maintenance cost, the failure cost, and the accident cost for the maintenance target in at least one of the cases where the operation is performed in each of the plurality of operation contents. May be good.

According to this, it is possible to identify various costs that may differ depending on the maintenance work and the operation content. As a result, it becomes possible to provide more suitable judgment material regarding the planning and execution of maintenance work or the value of the maintenance target.

Further, in the diagnostic system of the present embodiment, the arithmetic unit performs a process of multiplying the failure probability at each time indicated by the transition of the failure probability by the failure cost defined by the table, and the above-mentioned 1 at the predetermined time. Alternatively, the processing result of multiplying the maintenance execution cost in at least one of the cases where the maintenance is performed for each of the plurality of maintenance contents and the operation for each of the one or more operation contents is performed. In addition to, the process of calculating the expected cost value of the maintenance object may be further executed.

According to this, it is possible to identify the expected cost value that may differ depending on the maintenance work and the operation content. As a result, it becomes possible to provide more suitable judgment material regarding the planning and execution of maintenance work or the value of the maintenance target.

Further, in the diagnostic system of the present embodiment, the arithmetic unit may calculate the expected cost value for a plurality of maintenance objects.

According to this, it is possible to specify the above-mentioned expected cost value which can be different for each maintenance target and also different depending on the maintenance work and the operation content. As a result, it becomes possible to provide more suitable judgment material regarding the planning and execution of maintenance work or the value of the maintenance target.

Further, in the diagnostic system of the present embodiment, the storage device maintains the useful life, the production amount per unit time, the downtime during maintenance, and the failure probability of a predetermined standard in the maintenance object. , And the insurance money paid by the insurance company for each maintenance implementation opportunity, and the computing device is used as the maintenance content scheduled to be implemented for the maintenance object and the opportunity for maintenance implementation with the maintenance content. In the process of accepting the user designation for each of the failure probabilities to be performed and the transition of the estimated failure probability, the frequency of the failure probability receiving the user designation is counted, and the counted frequency is multiplied by the useful life. Then, from the process of calculating the number of maintenance executions and the remaining operating time up to the useful life, the value obtained by multiplying the downtime at the time of maintenance execution of the maintenance contents specified by the user by the number of maintenance executions is subtracted. Then, the total operating time is calculated, the total operating time is multiplied by the production amount per unit time to calculate the total production amount, and the maintenance execution cost is multiplied by the maintenance execution number to obtain the total maintenance cost. , The process of multiplying the annual insurance premium by the useful life to calculate the total premium paid, and the process of multiplying the insurance money for each maintenance implementation opportunity by the number of maintenance implementations to obtain the total insurance money. And the total production amount, the total maintenance cost and the total premium paid are subtracted, and the total amount of the insurance money is added to the result of the subtraction to calculate the future value of the maintenance object. It may be said that the processing to be performed and the processing to be performed are further executed.

According to this, it becomes possible to identify the future value of the maintenance target under predetermined conditions such as the maintenance content and operation content specified by the user, and by extension, provide more suitable judgment material regarding the planning and execution of the maintenance work or the value of the maintenance target. It will be possible.

Further, in the diagnostic system of the present embodiment, the arithmetic unit maintains the ratio of the failure probability to a predetermined level or less in the total operating time based on the transition of the total operating time and the estimated failure probability. It may be calculated as the credit value of the object.

According to this, the maintenance is aimed at insurance companies that need to value the maintenance target as insured, or financial institutions that consider the maintenance target as collateral, for example, as movable property and perform credit screening, etc. It is possible to identify and present the value as a credit value from the viewpoint that the object can continuously generate goods. As a result, it becomes possible to provide more suitable judgment material regarding the planning and execution of maintenance work or the value of the maintenance target.

Further, in the diagnostic system of the present embodiment, the diagnostic system includes a storage device that holds an observed value of a predetermined event related to the maintenance object, a maintenance history related to the maintenance object, and an operation history of the maintenance object. By inputting the observed value and at least one of the maintenance history or the operation history to a predetermined machine learning algorithm, at least one of the operation contents indicated by the operation history or the maintenance contents indicated by the maintenance history and the operation. Regarding the effect of the event on the failure probability of the maintenance object by identifying the correlation with the fluctuation content of the observed value of the event due to at least one of the operation implementation in the content or the maintenance implementation in the maintenance content. At least the predetermined time on the maintenance target when the maintenance for each of the plurality of maintenance contents is performed at the predetermined time based on the failure probability evaluation algorithm including the explanatory variable expression specified in advance and the correlation. The subsequent transition of the failure probability may be estimated, and the information on the transition of the estimated failure probability may be displayed as a graph as the information on the transition of the failure probability.

According to this, it is possible to visually clarify the transition of the failure probability of the maintenance object to the person in charge of operation / maintenance of the plant, etc., which is the maintenance object, or the person in charge of the financial institution or the insurance company. It will be possible. As a result, it becomes possible to provide more suitable judgment material regarding the planning and execution of maintenance work or the value of the maintenance target.

Further, in the diagnostic system of the present embodiment, when estimating the transition of the failure probability, the arithmetic unit accepts a user designation of a predetermined condition that defines the operation content for each of the one or a plurality of operation contents, and the condition. It is also possible to estimate the transition of the failure probability of the maintenance object at least after the predetermined time when the operation with the operation content defined in the above is performed at the predetermined time.

According to this, it is possible to identify the transition of the failure probability in a form in which the conditions of the operation content are specified in detail, and visually present this to the above-mentioned person in charge or the like. As a result, it becomes possible to provide more suitable judgment material regarding the planning and execution of maintenance work or the value of the maintenance target.

Further, in the diagnostic system of the present embodiment, the arithmetic unit accepts at least one of the operating rate of the maintenance object, the materials used, and the attributes of the component parts as the user designation of the predetermined condition. May be good.

According to this, it is possible to identify the transition of the failure probability in a form in which more specific conditions of the operation content are specified, and visually present this to the above-mentioned person in charge or the like. As a result, it becomes possible to provide more suitable judgment material regarding the planning and execution of maintenance work or the value of the maintenance target.

Further, in the diagnostic system of the present embodiment, regarding the insurance contracted for the maintenance object, the annual insurance premium to be paid to the insurance company and the maintenance target when the failure probability of the predetermined standard is maintained in the maintenance object. The computing device is provided with a storage device that holds at least one of the information of the insurance money paid by the insurance company for each maintenance implementation opportunity for the object, and the computing device displays the annual insurance premium when displaying the graph of the estimation information of the failure probability transition. And, based on at least one of the information of the insurance money, the annual insurance premium and the failure probability standard which is the threshold value at which the insurance money fluctuates are displayed as an incentive fluctuation line of the insurance in the graph display. There may be.

According to this, it is possible to clearly present to the person in charge of operation / maintenance of the plant or the like, which is the object of maintenance, the necessity of carrying out maintenance work at an appropriate timing and frequency, setting the operation content, and the like. It can be expected that this will lead to an increase in maintenance awareness. As a result, it becomes possible to provide more suitable judgment material regarding the planning and execution of maintenance work or the value of the maintenance target.

10 Network 100 Diagnostic system 101 Storage device 102 Program 103 Memory 104 CPU (arithmetic unit)
105 Input device 106 Output device 107 Communication device 110 Machine learning engine 111 Failure probability evaluation engine 125 Sensor DB
126 Event DB
127 Operation DB
128 Estimated result DB
129 Incentive fluctuation standard DB
130 Credit value DB
200 plants (maintenance target)
300 Sensor unit 325 Measurement data 400 Insurance company system 500 Financial institution system

Claims (18)

A storage device that holds observation values of predetermined events related to the maintenance object and maintenance history related to the maintenance object.
By applying the observed value and the maintenance history to a predetermined machine learning algorithm as inputs, the correlation between the maintenance content indicated by the maintenance history and the fluctuation content of the observed value of the event due to the maintenance implementation in the maintenance content. And the arithmetic unit that identifies
A diagnostic system characterized by including. The storage device is
Further retain the operation history of the maintenance object,
The arithmetic unit
By inputting the observed value and at least one of the maintenance history or the operation history to a predetermined machine learning algorithm, the operation content indicated by the operation history or at least one of the maintenance contents indicated by the maintenance history and the maintenance content. It is intended to identify the correlation with the fluctuation content of the observed value of the above-mentioned event due to at least one of the operation implementation according to the operation content and the maintenance implementation according to the maintenance content.
The diagnostic system according to claim 1. The arithmetic unit
Based on the failure probability evaluation algorithm including the explanatory variable expression defined in advance regarding the influence of the event on the failure probability of the maintenance object and the correlation, maintenance is performed for each of the plurality of maintenance contents at a predetermined time. This is to estimate the transition of the failure probability of the maintenance object after the predetermined time when it is carried out.
The diagnostic system according to claim 1 or 2. The arithmetic unit
Based on the explanatory variable formula and the correlation, when maintenance is performed for each of the maintenance contents of one or more at the predetermined time, and when operation is performed for each of the operation contents of one or more. , At least one of the above, estimates the transition of the failure probability of the maintenance object after at least the predetermined time.
The diagnostic system according to claim 3, wherein the diagnostic system is characterized in that. The arithmetic unit
In estimating the transition of the failure probability, for each of the one or more operation contents, the user designation of the predetermined conditions for defining the operation contents is accepted, and the operation with the operation contents defined by the conditions is carried out at the predetermined time. In this case, the transition of the failure probability of the maintenance object at least after the predetermined time is estimated.
The diagnostic system according to claim 4, wherein the diagnostic system is characterized in that. The arithmetic unit
As a user designation of the predetermined condition, at least one of the operating rate of the maintenance object, the materials used, and the attributes of the component parts is accepted.
The diagnostic system according to claim 5, wherein the diagnostic system is characterized in that. The storage device is
Further retains information on each cost required in the event of past failures and accidents on the maintenance object.
The arithmetic unit
Based on the cost at the time of performing maintenance indicated by the maintenance history and the costs of each of the failure and the accident, the maintenance target is maintained for each of the above-mentioned one or a plurality of the above-mentioned maintenance contents. And, when the operation is performed in each of the above-mentioned one or a plurality of operation contents, a table that defines the maintenance execution cost, the failure time cost, and the accident time cost in the maintenance target object in at least one of them is generated. be,
The diagnostic system according to claim 4, wherein the diagnostic system is characterized in that. The arithmetic unit
When the process of multiplying the failure probability at each time indicated by the transition of the failure probability by the failure cost specified in the table and the maintenance for each of the above-mentioned one or a plurality of the above-mentioned maintenance contents are performed at the above-mentioned predetermined time. , And the cost at the time of performing the maintenance in at least one of the operations in each of the one or more operation contents is added to the processing result of the multiplication to calculate the expected cost value in the maintenance object. And further execution of the processing to be performed,
The diagnostic system according to claim 7. The arithmetic unit
The expected cost value is calculated for a plurality of maintenance objects.
8. The diagnostic system according to claim 8. The storage device is
The service life, production volume per unit time, downtime during maintenance, annual premium when maintaining a predetermined failure probability, and insurance money paid by the insurance company for each maintenance opportunity, etc. Further retain each information of
The arithmetic unit
A process for accepting user specifications for each of the maintenance contents scheduled to be performed for the maintenance target and the failure probability that triggers the maintenance performance in the maintenance contents.
In the transition of the estimated failure probability, the frequency of the failure probability specified by the user is counted, and the counted frequency is multiplied by the useful life to calculate the number of maintenance operations.
The total operating time is calculated by subtracting the value obtained by multiplying the downtime at the time of performing maintenance for the maintenance content specified by the user by the number of times the maintenance is performed from the remaining operating time up to the useful life, and the total operating time is calculated. The process of multiplying the operating time by the production amount per unit time to calculate the total production amount,
The process of calculating the total maintenance cost by multiplying the maintenance cost by the number of maintenance operations.
The process of multiplying the annual premium by the useful life to calculate the total premium paid,
The process of calculating the total amount of insurance money by multiplying the insurance money for each maintenance implementation opportunity by the number of maintenance operations.
A process of calculating the future value of the maintenance object by subtracting the total maintenance cost and the total premium paid from the total production amount and adding the total amount of the insurance money to the result of the subtraction.
Is to further execute,
9. The diagnostic system according to claim 9. The arithmetic unit
Based on the total operating time and the transition of the estimated failure probability, the ratio of the total operating time in which the failure probability is equal to or lower than a predetermined level is calculated as the credit value of the maintenance object.
10. The diagnostic system according to claim 10. At least one of the cases where maintenance of one or more maintenance contents is carried out for a predetermined maintenance object at a predetermined time and the operation is carried out for each of one or more operation contents. A diagnostic system comprising a calculation device for displaying a graph of estimation information of a failure probability transition after at least the predetermined time on the maintenance object. It is provided with a storage device that holds observation values of predetermined events related to the maintenance object, maintenance history of the maintenance object, and operation history of the maintenance object.
The arithmetic unit
By inputting the observed value and at least one of the maintenance history or the operation history to a predetermined machine learning algorithm, the operation content indicated by the operation history or at least one of the maintenance contents indicated by the maintenance history and the maintenance content. Identify the correlation with the fluctuation content of the observed value of the above event due to at least either the operation implementation according to the operation content or the maintenance implementation according to the maintenance content.
Based on the failure probability evaluation algorithm including the explanatory variable formulas defined in advance regarding the influence of the event on the failure probability of the maintenance object and the correlation, maintenance of each of the plurality of maintenance contents is performed at the predetermined time. Estimate the transition of the failure probability of the maintenance object at least after the predetermined time when it is carried out.
The information on the transition of the estimated failure probability is displayed as a graph as the information on the transition of the failure probability.
12. The diagnostic system according to claim 12. The arithmetic unit
In estimating the transition of the failure probability, for each of the one or more operation contents, the user designation of the predetermined conditions for defining the operation contents is accepted, and the operation with the operation contents defined by the conditions is carried out at the predetermined time. In this case, the transition of the failure probability of the maintenance object at least after the predetermined time is estimated.
13. The diagnostic system according to claim 13. The arithmetic unit
As a user designation of the predetermined condition, at least one of the operating rate of the maintenance object, the materials used, and the attributes of the component parts is accepted.
14. The diagnostic system according to claim 14. Regarding the insurance contracted for the maintenance object, the insurance company will pay the annual premium to be paid to the insurance company when the failure probability of the maintenance object is maintained, and the maintenance implementation opportunity for the maintenance object. Equipped with a storage device that holds at least one of the insurance claims paid,
The arithmetic unit
When displaying the graph of the estimation information of the failure probability transition, based on at least one of the information of the annual insurance premium and the insurance money, the standard of the failure probability, which is the threshold value at which the annual insurance premium and the insurance money fluctuate, is It is displayed in the graph display as an incentive fluctuation line for the insurance.
12. The diagnostic system according to claim 12. Information processing system
It is equipped with a storage device that holds observation values of predetermined events related to the maintenance object and maintenance history related to the maintenance object.
By applying the observed value and the maintenance history to a predetermined machine learning algorithm as inputs, the correlation between the maintenance content indicated by the maintenance history and the fluctuation content of the observed value of the event due to the maintenance implementation in the maintenance content. To identify,
A diagnostic method characterized by that. Information processing system
At least one of the cases where maintenance of one or more maintenance contents is carried out for a predetermined maintenance object at a predetermined time and the operation is carried out for each of one or more operation contents. , At least the estimation information of the failure probability transition after the predetermined time in the maintenance object is displayed as a graph.
A diagnostic method characterized by that.

Images