Equipment maintenance system and equipment maintenance method

Abstract

Proposed is a maintenance method for achieving high cost performance by comparing the cost of an on-site maintenance service with the cost of a remote maintenance service.　This equipment maintenance system determines a response method for responding to a failure of equipment, the equipment maintenance system comprising: an asset knowledge database in which equipment failures and causes thereof are registered; an input unit that receives failure information output from the equipment; a failure cause estimation unit that estimates the cause of a failure, with reference to the asset knowledge database, on the basis of the failure information received by the input unit; a response method determination unit that determines a response method for responding to the cause of the failure estimated by the failure cause estimation unit; a cost estimation unit that estimates the cost required for the response method determined by the response method determination unit by separating the cost into the cost of an on-site response, in which an engineer visits the site where the equipment is installed, and the cost of a remote response, in which the engineer responds remotely; and an output unit that outputs the response method with the lowest cost among the costs estimated by the cost estimation unit.

Description

Equipment maintenance system and equipment maintenance method

[0001] The present invention relates to an equipment maintenance system and an equipment maintenance method.

[0002] When a malfunction occurs in equipment such as an industrial multifunction printer, the maintenance department must listen to the user's symptoms and infer possible causes of the malfunction. Based on this inference, it is necessary to decide whether to visit the user's site where the equipment is installed or to handle the malfunction remotely.

[0003] There are two methods for responding to failures: remote maintenance and maintenance performed by visiting the user site. Patent Document 1 discloses a maintenance support system that includes a maintenance support tree information management unit that manages maintenance support tree information including candidate remedial actions required for recovery from malfunctions and information on diagnostic actions for identifying them, an optimal action calculation unit that calculates a starting point for actions that minimizes the expected value of the action cost or action time within the maintenance support tree using the number of past cases, the experience of maintenance workers, diagnostic actions, and cost information related to each action, and a diagnostic action execution unit that updates the maintenance support tree information by reflecting the results of the executed actions in order to repeat the calculation of the optimal action until the recovery of the maintenance target device is completed.

[0004] JP 2013-29881 A

[0005] However, the maintenance support system of Patent Document 1 does not take into consideration the travel time and transportation costs required for a business trip when performing maintenance services.

[0006] The object of this invention is to use the results of failure cause estimation to quantify the costs associated with travel time and expenses required for on-site visits, compare the costs of on-site maintenance services and remote maintenance services, and propose a maintenance method with good cost performance.

[0007] The object of the present invention is achieved by an equipment maintenance system that determines a response method for responding to equipment failures, the system comprising: an asset knowledge database in which equipment failures and their causes are registered; an input unit that receives failure information output from the equipment; a failure cause estimation unit that references the asset knowledge database based on the failure information received by the input unit and estimates the cause of the failure; a response method determination unit that determines a response method for responding to the cause of the failure estimated by the failure cause estimation unit; a cost estimating unit that estimates the costs required for the response method determined by the response method determination unit by dividing it into the cost of on-site response in which an engineer visits the site where the equipment is installed and the cost of remote response in which an engineer responds remotely; and an output unit that outputs the response method with the lowest cost of the costs estimated by the cost estimating unit.

[0008] According to the present invention, it is possible to quantify the costs related to travel time and transportation expenses required for business trips and to propose a cost-effective maintenance method.

[0009] Example of a system configuration diagram in an embodiment of the present invention Example of maintenance knowledge data in an embodiment of the present invention Example of failure probability in an embodiment of the present invention Example of child node abnormality probability when parent node is abnormal in an embodiment of the present invention Example of child node abnormality probability when parent node is normal in an embodiment of the present invention Example of a maintenance knowledge Bayesian network in an embodiment of the present invention Example of a flowchart for generation processing of a maintenance knowledge Bayesian network in an embodiment of the present invention Example of a flowchart for failure cause estimation processing in an embodiment of the present invention Example of failure cause estimation result in an embodiment of the present invention Example of a failure cause response method table in an embodiment of the present invention Example of a business trip expense table in an embodiment of the present invention Example of response method data in an embodiment of the present invention Example of business trip response data in an embodiment of the present invention Example of remote response data in an embodiment of the present invention Example of a flowchart for business trip / remote response method decision support processing in an embodiment of the present invention Example of process S34 in the flowchart for business trip / remote response method decision support processing in an embodiment of the present invention Example of process S35 in the flowchart for business trip / remote response method decision support processing in an embodiment of the present invention Example of process S36 in the flowchart for business trip / remote response method decision support processing in an embodiment of the present invention Example of process S37 in the flowchart for business trip / remote response method decision support processing in an embodiment of the present invention Example of an input screen in an embodiment of the present invention Example of a result screen in an embodiment of the present invention

[0010] Hereinafter, embodiments of the present invention will be described with reference to the drawings. In each drawing for explaining the embodiments, the same components are given the same names and reference numerals as much as possible, and repeated explanations thereof will be omitted.

[0011] The present invention is not limited to the following examples, and includes various modifications and equivalent configurations within the spirit and scope of the appended claims. For example, the above examples have been described in detail to clearly explain the present invention, and the present invention is not necessarily limited to those including all of the described configurations.

[0012] Furthermore, the processing units described in the embodiments may be realized in hardware, for example by designing some or all of them as integrated circuits, or may be realized in software by having a processor interpret and execute a program that realizes each function.

[0013] The tables, areas, etc. described in the embodiments may be a database (DB) or data stored in a main memory. Hereinafter, the embodiments for carrying out the present invention will be described in detail with reference to the accompanying drawings.

[0014] 1 is an example of a system configuration diagram according to an embodiment of the present invention. The equipment maintenance system is configured with a computer including a CPU (Central Processing Unit) 2, memory 3, and external storage device 4. In this embodiment, a configuration using a standalone computer will be described, but it may also be realized by a server on a cloud, or each function may be divided and realized by multiple computers.

[0015] The memory 3 includes a weighting coefficient determination unit 7 that determines whether to prioritize monetary cost or time cost in the calculation, a response method determination unit 8 that determines whether maintenance should be performed remotely or on-site by visiting the site, a failure cause estimation unit 9 that estimates the cause of a failure based on input equipment status and alarms, a cost estimation unit 19 that estimates the cost of responding to a failure, and an input / output unit 10 that performs input / output processing.

[0016] The external storage device 4 has a failure mode response method database 11 and an asset knowledge database 12. The failure mode response method database 11 stores a failure cause response TBL (table) 13 that determines whether to respond to a failure remotely or by on-site, and a trip cost TBL (table) that lists the time and cost required for a trip.

[0017] The asset knowledge database 12 stores maintenance knowledge data 15 that associates functional failures and alarms with the failure modes that are the causes of those failures, failure probabilities 16 that store the probability of the occurrence of the failure causes, child node abnormality probabilities 17 when the parent node is abnormal, and child node abnormality probabilities 18 when the parent node is normal.

[0018] The asset knowledge database is a database that stores maintenance knowledge. Maintenance knowledge is information extracted from, for example, maintenance manuals and FMEA, and includes, for example, the maintenance knowledge data shown in FIG. 2 and the probability setting data shown in FIGS. 3 to 5. FIG. 2 is an example of maintenance knowledge data in an embodiment of the present invention. Functional failure indicates a failure phenomenon such as cooling water not circulating or part 2 not being able to be stored. Failure mode indicates the cause of the failure. Check items are columns that store items to be checked, such as sensor data of the equipment, environment, equipment, components, etc.

[0019] The check items are used to identify the phenomena that can occur when a failure mode occurs and to check whether or not the phenomenon has occurred. There is not necessarily one check item corresponding to one failure mode. Alarm information that often occurs when that functional failure occurs is also included.

[0020] 3 shows an example of failure probabilities in an embodiment of the present invention. This is data for setting the probability when creating a Bayesian network. It is composed of a failure mode 41, which stores the name of the failure mode, a state 42, and a probability 43. Based on this table, the occurrence probability of each failure mode can be known.

[0021] The state 42 stores Y or N, which respectively indicate the occurrence and non-occurrence of a failure mode. These states are set so that the sum of the probabilities of Y and N related to the same failure mode is 1.

[0022] The probability 43 stores the probability of the failure mode state. Here, the failure mode state Y and N may each be set to a fixed value such as 50%. However, this is not limiting and calculations can also be made from past history, etc. The probability of the failure mode occurrence state is, for example, the value obtained by dividing the number of occurrences of the failure mode in the failure history by the total number of occurrences.

[0023] 4 shows an example of the child node anomaly probability when a parent node is abnormal in an embodiment of the present invention. It is composed of a parent node 51, a state 52, a child node 53, a child node state 54, and a probability 55. When a certain failure mode occurs, the check items indicate the probability of abnormality and normality.

[0024] The name of the failure mode is stored in the parent node 51. In this table, Y is stored in the state 52, which means the state in which the failure mode occurs.

[0025] The child node 53 stores a check item, in this case the sensor name. The child node status 54 stores the state of the check item, in this case the state of the sensor. The probability 55 stores the probability of the state of the check item, in this case the probability of the sensor state. The sum of the abnormal and normal probabilities for the same check item state is set to 1.0 (100%).

[0026] A fixed value such as 100% when abnormal and 0% when normal can be set, or it may be calculated from the past failure history.

[0027] 5 shows an example of the probability of a child node being abnormal when the parent node is normal in an embodiment of the present invention. It includes a child node 61, a child node state 62, and a probability 63.

[0028] A check item is stored in a child node 61, and in this case, the check item is stored. A child node status 62 stores check details, and in this case, the status of the check item is stored. A probability 63 stores the probability of the check item status, and in this case, the probability of the sensor status is stored. The sum of the abnormal and normal probabilities related to the status of the same check item is set to 1.0 (100%).

[0029] It may be set to a fixed value such as 0% when abnormal and 100% when normal, or may be calculated from the history of past failures.

[0030] When an abnormality occurs in equipment, the failure cause estimation unit in Figure 1 generates a maintenance knowledge Bayesian network for cause estimation, inputs check items into the maintenance knowledge Bayesian network according to information on the symptoms of the failure input by a maintenance technician or user, and calculates the probability of occurrence of each failure mode. <Maintenance Knowledge Bayesian Network Generation Process> Figure 7 is an example of a flowchart of the maintenance knowledge Bayesian network generation process in an embodiment of the present invention. This process is executed by the failure cause estimation unit 9 using the maintenance knowledge data 15 stored in the asset knowledge database 12.

[0031] 6 shows an example of a maintenance knowledge Bayesian network in an embodiment of the present invention. This maintenance knowledge Bayesian network is a Bayesian network with a four-layer structure of alarms, functional failures, failure modes, and check items, and arrows indicate causal relationships.

[0032] The start of the arrow is the parent node and indicates the cause. The tip of the arrow is the child node and indicates the result. The maintenance knowledge Bayesian network generation processing flow is as follows:

[0033] First, the maintenance knowledge data 15 is acquired from the asset knowledge database (S11). Next, a Bayesian network node is created from each cell of the maintenance knowledge data 15 (S12). At this time, cells with the same content are generated as one node.

[0034] Then, a parent-child link is established between each node in accordance with the causal relationships described in the maintenance knowledge data 15 (S13).

[0035] Links are established between the nodes in parent-child relationships according to the causal relationships described in the maintenance knowledge data 15. Specifically, arrows are set to form causal relationships (parent-child relationships) such as "failure mode" → "functional failure," "failure mode" → "check item," and "functional failure" → "alarm."

[0036] The failure probability 16 is set as a priori probability for the set arrow (S14). Next, a posterior probability is set for the arrow indicating the parent-child relationship using the child node abnormality probability 17 when the parent node is abnormal and the child node abnormality probability 18 when the parent node is normal (S15).

[0037] 8 is an example of a flowchart of a fault cause estimation process in an embodiment of the present invention. Information on the fault symptoms is received from a maintenance worker or a user via the input unit (S21). Using an input screen such as that shown in FIG. 16, "abnormal" or "normal" is entered in the check items according to the equipment fault information provided by the maintenance worker or the equipment user.

[0038] "Abnormal" means that something is occurring, and "Normal" means that something is not occurring. If it is unknown whether something is occurring or not, do not enter anything. For example, if there is information that "oil is leaking," enter "Abnormal" in the check item "Oil is leaking." Do not enter anything in other check items.

[0039] A maintenance knowledge Bayesian network is generated based on the input information (S22). The occurrence probability of each failure mode is calculated using the maintenance knowledge Bayesian network, and the estimation results are output to the response method determination unit 8. The failure cause estimation results are as shown in Figure 10. An occurrence probability 93 corresponding to each failure mode 92 is calculated.

[0040] Next, the decision support process for the on-site / remote response method will be described.

[0041] 10 shows an example of a fault cause handling method table in the embodiment of the present invention, which stores fault causes 101 and fault handling methods 102.

[0042] 11 is an example of a travel expense table in an embodiment of the present invention. It stores, in association with each other, the customer 103 who owns the equipment, the round-trip travel time 104 to the location where the equipment is installed, and the transportation cost 105 for the travel.

[0043] 12 shows an example of response method data in an embodiment of the present invention. For each failure mode 107, a response method 108 indicating whether the failure requires on-site response or can be handled remotely, and a probability 109 indicating the possibility of the failure mode are stored in association with each other.

[0044] 13 shows an example of business trip response data in an embodiment of the present invention. This is a table that collects only those cases where the response method is business trip from the data in FIG.

[0045] 14 shows an example of remote correspondence data in an embodiment of the present invention. This is a table that collects only those correspondence methods that are remote from the data in FIG.

[0046] FIG. 15 illustrates an example of a flowchart of a decision support process for a business trip / remote response method in an embodiment of the present invention that uses these tables to propose a response method with high cost performance.

[0047] The failure cause response method and user information input from the failure mode response method database from the maintenance worker terminal 5 are received via the input / output unit 10. At this time, weighting information indicating whether monetary cost or time cost is given more importance may also be received. Based on the received user information, a failure cause estimation result is obtained using a Bayesian network (S31).

[0048] A countermeasure is determined based on each failure cause by referring to the failure cause countermeasure TBL 13, and countermeasure determination information 91 is created in which the countermeasure is associated with the probability of the estimated failure cause (S32).

[0049] For each failure mode 107 in the response method determination information 91, the response methods 108 are divided into remote failures and on-site failures (S33).

[0050] The cost estimator 19 calculates the monetary cost M remote for the failure whose response method is remote (S34), and calculates the time cost T remote for each of the failure whose response method is remote (S35).

[0051] The calculation process for M remoteness performed in S34 is shown in Figure 15A. First, M remoteness and Mkj are initialized to 0 (S001). The number of rows k of remote support data is received, and this process is repeated k times. The probability of the failure mode for each row shown in Figure 14 is set to "P remoteness k" (S002). Each time process S002 is performed, the number of rows j of on-site support data is received, and the process is repeated. The probability of the failure mode for each row shown in Figure 13 is set to "P on-site j", and the monetary cost is calculated according to the following formula (1) (S003).

[0052] T traffic is the travel time shown in Figure 11, M traffic is the transportation cost, and M labor cost is the hourly labor cost of the maintenance worker who performs the maintenance work.

[0053] Mkj=P business trip j×(1−P remote k)×(T transportation × M labor cost·H+M transportation cost) (1) Every time step S003 is performed, Mkj is added to M remote (S004). Step S003 is performed k×j times to obtain M remote (S005).

[0054] The calculation process for T remoteness performed in S35 is shown in Figure 15B. First, T remoteness and Tkj are initialized to 0 (S011). The number of rows k of remote correspondence data is received, and this process is repeated k times. The probability of the failure mode for each row shown in Figure 14 is set to "P remoteness k" (S012). Each time process S002 is performed, the number of rows j of on-site correspondence data is received, and the process is repeated. The probability of the failure mode for each row shown in Figure 13 is set to "P on-site j", and the time cost is calculated according to the following formula (1) (S013).

[0055] Tkj=P business trip j×(1−P distance k)×T transportation (2) Every time step S003 is performed, Tkj is added to T distance (S014). Step S003 is performed k×j times to obtain M distance (S015).

[0056] Based on the weighting information, the monetary cost and the time cost are multiplied by their respective weights and summed to obtain the cost remote, which is the cost of remote support (S36). The weighting information may be set to a predetermined weight, or the monetary cost and the time cost may be weighted equally.

[0057] Next, the cost estimating unit 19 calculates M trips, which is the total monetary cost, even if the response method is a trip to a breakdown (S37), and calculates T trips, which is the total time cost (S38).Similarly, it calculates the cost trip based on the weighting information (S39).

[0058] The calculation process for M business trips performed in S36 is shown in Figure 15C. First, M business trips and Mkj are initialized to 0 (S021). The number of rows of remote correspondence data, k, is received, and this process is repeated k times. The probability of the failure mode for each row shown in Figure 14 is set to "P remote k" (S022). Each time process S022 is performed, the number of rows of business trip correspondence data, j, is received, and the process is repeated. The probability of the failure mode for each row shown in Figure 13 is set to "P business trip j," and the monetary cost is calculated according to the following formula (1) (S023). Mkj = P remote k x (1 - P business trip j) x [(T remote + T transportation) x M labor costs x H + M transportation costs] (3) Each time process S023 is performed, Mkj is added to M business trips (S024). Process S023 is performed k x j times to obtain M business trips (S025).

[0059] The calculation process for T trips performed in S37 is shown in Figure 15D. First, T trips and Tkj are set and initialized (S031). The number of rows of remote correspondence data, k, is received, and this process is repeated k times. The probability of the failure mode for each row shown in Figure 14 is set to "P remote k" (S032). Each time process S032 is performed, the number of rows of remote correspondence data, j, is received, and the process is repeated. The probability of the failure mode for each row shown in Figure 13 is set to "P trip j," and the time cost is calculated according to the following formula (1) (S033). Tkj = P trip j x (1 - P remote k) x T transportation (4) Each time process S033 is performed, Tkj is added to T trips (S034). Process S033 is performed k x j times to obtain T trips (S035).

[0060] The cost estimation unit 19 compares the cost of remote response with the cost of on-site response (S40), and if the cost of remote response is higher, the output unit outputs information such as the result display message 170 shown in FIG. 17 "On-site response method is less costly, so on-site response is recommended" to the maintenance staff (S41), and if the on-site response cost is higher, the output unit outputs the result display message 170 such as the information "Remote response method is less costly, so remote response is recommended" to the maintenance staff (S42).

[0061] Finally, "M remote", "T remote", "M business trip", and "T business trip" are output to the cost calculation result 172 (S43).

[0062] 17 is an example of a result display screen in an embodiment of the present invention. Cost calculation results 172 show costs for four cases. The four cases are a combination of the response methods (on-site / remote) for the failure modes that actually occurred and the response methods (on-site / remote) adopted at the discretion of the maintenance personnel.

[0063] Among the four cases, there are two ideal cases: when a failure mode occurs that requires on-site response and maintenance personnel respond by on-site response, and when a failure mode occurs that requires remote response and maintenance personnel respond by remote response. The ideal cases mean that no unnecessary costs or time are incurred.

[0064] For example, if a failure mode F1 occurs that requires a site visit and a maintenance technician visits the facility to respond based on the symptoms, the only costs incurred are the cost and time required for the site visit, which is an ideal case.

[0065] If a failure mode F1 occurs that requires an on-site response, and the maintenance staff determines from the symptoms that they will respond remotely, the equipment will not recover quickly after the remote response, and an on-site response will be required, which will incur not only the cost and time of the on-site response but also the cost of the remote response, so this is not an ideal case.

[0066] The cost calculation result 172 shows the cost of four cases, and the expenses and time required compared to the ideal case, as (1) to (4). (1) and (4) are ideal cases, and the deviations from the ideal are displayed as cost: 0 yen and work time: 0 hours. In (2), the deviations from the ideal are displayed as cost: "M remote" yen and work time: "T remote" hours. In (3), the deviations from the ideal are displayed as cost: "M on-site" yen and work time: "T on-site" hours. "M remote", "T remote", "M on-site", and "T on-site" are values ​​calculated using the flowchart shown in FIG. 15.

[0067] 1 Equipment maintenance system, 2 CPU, 3 Memory, 4 External storage device, 5 Maintenance worker terminal, 6 Maintenance worker, 7 Weight coefficient determination unit, 8 Response method determination unit, 9 Failure cause estimation unit, 10 Input / output unit, 11 Failure mode response method database, 12 Asset knowledge database, 13 Failure cause response TBL, 14 Travel cost TBL, 15 Maintenance knowledge data, 16 Failure probability, 17 Child node abnormality probability when parent node is abnormal, 18 Child node abnormality probability when parent node is normal, 19 Cost estimation unit

Claims

1. An equipment maintenance system that determines a response method for responding to an equipment failure, comprising: an asset knowledge database in which equipment failures and their causes are registered; an input unit that accepts failure information output from the equipment; a failure cause estimation unit that refers to the asset knowledge database based on the failure information accepted by the input unit and infers the cause of the failure; a response method determination unit that determines a response method for responding to the cause of the failure inferred by the failure cause estimation unit; a cost estimating unit that estimates the costs required for the response method determined by the response method determination unit, dividing the costs into on-site response costs, in which an engineer visits the site where the equipment is installed, and remote response costs, and an output unit that proposes the response method that is cheaper than the on-site response costs and remote response costs estimated by the cost estimating unit.

2. The facility maintenance system according to claim 1, wherein said cost estimating section estimates the cost of on-site response by including the transportation costs and time required for an engineer to travel to the site as costs.

3. The equipment maintenance system according to claim 2, wherein the asset knowledge database stores failure information and a probability of occurrence of the failure cause associated therewith, and the failure cause estimation unit estimates the failure cause based on the probability of occurrence of the failure cause.

4. The equipment maintenance system of claim 2, wherein the input unit receives cost information indicating whether cost or time for responding to a failure is to be prioritized, and the cost estimation unit is further provided with a weighting coefficient calculation unit that determines a weighting coefficient indicating whether cost or time is to be prioritized when estimating costs based on the cost information.

5. The equipment maintenance system of claim 2, wherein the failure cause estimation unit estimates a plurality of failure causes, the cost estimation unit estimates the cost for each failure cause, calculates the total of on-site response costs and the total of remote response costs, and the output unit outputs the smaller of the total of on-site response costs and the total of remote response costs.

6. An equipment maintenance method for determining a response method for responding to an equipment failure, comprising: an input unit receiving failure information output from the equipment; a failure cause estimation unit estimating the cause of the failure based on the failure information received by the input unit by referring to an asset knowledge database in which equipment failures and their causes are registered; a response method determination unit determining a response method for responding to the cause of the failure estimated by the failure cause estimation unit; a cost estimation unit dividing the costs required for the response method determined by the response method determination unit into an on-site response cost in which an engineer visits the site where the equipment is installed and a remote response cost in which the engineer responds remotely; and an output unit proposing the response method which is the lesser of the on-site response cost and the remote response cost estimated by the cost estimation unit.

Images