Quality index calculation device and quality index calculation method

Abstract

To provide a quality index calculation device and method that improve calculation accuracy of a quality index.SOLUTION: A quality index calculation device 2 comprises: a parameter calculation unit 11 that acquires longevity actual data d1,k indicative of respective longevity in a plurality of components including the component different in a specification from an assembly component, in which a factor impacting longevity of the component is in common with the assembly component serving as the component to be assembled to a finished product, and factor information indicative of the factor impacting the longevity of the plurality of components, and calculates a parameter about a longevity probability distribution of the plurality of components on the basis of respective longevity actual data and factor information; a longevity probability distribution estimation unit 12 that estimates a longevity probability distribution, using factor information d2,i and a calculation result P1 of the parameter the parameter calculation unit 11 calculates; and a quality index calculation unit 13 that calculates a quality index indicative of a quality of the assembly component from an estimation result P2 of the longevity probability distribution the longevity probability distribution estimation unit 12 estimates.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] This disclosure relates to a quality index calculation device and a quality index calculation method. [Background technology]

[0002] There is a quality index calculation device that calculates quality indicators that show the quality of the components included in the finished product. For example, Patent Document 1 discloses an apparatus comprising a Weibull analysis step for estimating the distribution of failure rates of components (hereinafter referred to as the "failure probability distribution") from actual data indicating the lifespan of components, and a lifespan determination step for calculating a statistical service life from the failure probability distribution. The service life is the period from the end of the initial failure period of a component to the start of the wear-out failure period. [Prior art documents] [Patent Documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 10-34122 [Overview of the project] [Problems that the invention aims to solve]

[0004] In the apparatus disclosed in Patent Document 1, the service life determination step calculates a statistical service life, and the service life corresponds to the quality index calculated by the quality index calculation device. If a sufficient number of real-world data points are not obtained to represent the lifespan of a component, the accuracy of the failure probability distribution estimation performed by the Weibull analysis step deteriorates. As a result, the accuracy of the lifespan calculation performed by the lifespan determination step may deteriorate. In other words, there was a problem in that the accuracy of quality indicator calculations may deteriorate.

[0005] This disclosure was made to solve the above-mentioned problems, and aims to provide a quality index calculation device that can improve the accuracy of quality index calculation compared to the device disclosed in Patent Document 1. [Means for solving the problem]

[0006] The quality index calculation device according to this disclosure includes a parameter calculation unit that calculates parameters relating to the life probability distribution of multiple parts based on the actual life data and factor information relating to the factors influencing the life of multiple parts, which include multiple parts that have different specifications from the embedded parts but share factors that affect the life of the multiple parts, and a life probability distribution estimation unit that estimates the life probability distribution using the factor information and the parameters calculated by the parameter calculation unit. The quality index calculation device also includes a quality index calculation unit that calculates a quality index indicating the quality of the embedded parts from the life probability distribution estimated by the life probability distribution estimation unit. [Effects of the Invention]

[0007] According to this disclosure, the accuracy of calculating quality indicators can be improved compared to the apparatus disclosed in Patent Document 1. [Brief explanation of the drawing]

[0008] [Figure 1] This is a diagram showing the configuration of the quality index calculation device 2 according to Embodiment 1. [Figure 2] This is a hardware configuration diagram showing the hardware of the quality index calculation device 2 according to Embodiment 1. [Figure 3] This is a hardware configuration diagram of a computer when the quality index calculation device 2 is implemented by software or firmware, etc. [Figure 4] This flowchart shows the quality index calculation method, which is the processing procedure for the quality index calculation device 2. [Figure 5] This is an explanatory diagram showing an example of a lifetime probability distribution f(t). [Figure 6] This is a configuration diagram showing the quality index calculation device 2 according to Embodiment 2. [Figure 7] This is a hardware configuration diagram showing the hardware of the quality index calculation device 2 according to Embodiment 2. [Figure 8] This is a configuration diagram showing the quality index calculation device 2 according to Embodiment 3. [Figure 9] This is a hardware configuration diagram showing the hardware of the quality index calculation device 2 according to Embodiment 3. [Figure 10] This is an explanatory diagram showing the probability of the ranked quality index P4 occurring. [Modes for carrying out the invention]

[0009] To provide a more detailed explanation of this disclosure, the forms for implementing this disclosure will be described below with reference to the attached drawings.

[0010] Embodiment 1. Figure 1 is a configuration diagram showing a quality index calculation device 2 according to Embodiment 1. Figure 2 is a hardware configuration diagram showing the hardware of the quality index calculation device 2 according to Embodiment 1. In Figure 1, the storage device 1 stores actual lifespan data d1, which indicates the lifespan of each of the multiple components. Furthermore, the storage device 1 stores factor information d2, which indicates factors that affect the lifespan of the components.

[0011] The quality index calculation device 2 calculates the component lifespan β from the memory device 1. i Factor x that gives (i=1,···,I) i However, the actual lifespan data d represents the lifespan of each of the N parts that are common to the embedded components, and which include N parts that have different specifications from the embedded components. 1,k Obtain (k=1,···,N). N is an integer greater than or equal to 2, and I is an integer greater than or equal to 1. The built-in component is a component to be incorporated into a finished product and is a component for which the quality index calculation device 2 calculates a quality index. That the built-in component and the specification are the same means, for example, that each of the shape, function, and performance is the same as that of the built-in component. Also, the quality index calculation device 2 has a factor x that affects the lifetimes of N components by β i and obtains factor information d i indicating that factor. 2,i Based on the actual lifetime data d 1,k and the factor information d 2,i the quality index calculation device 2 calculates a quality index indicating the quality of the component. The quality index calculation device 2 includes a parameter calculation unit 11, a lifetime probability distribution estimation unit 12, and a quality index calculation unit 13.

[0012] The parameter calculation unit 11 is realized, for example, by a parameter calculation circuit 21 shown in FIG. 2. The parameter calculation unit 11 obtains from the storage device 1 the actual lifetime data d i (i = 1, ···, I) indicating the lifetimes of N components for which the factor x that gives β i is common to the built-in component and includes N components with different specifications from the built-in component. 1,k (k = 1, ···, N). Also, the parameter calculation unit 11 obtains from the storage device 1 the factor information d i indicating the factor x that affects the lifetimes of the N components by β i . 2,i Based on the actual lifetime data d 1,k and the factor information d 2,i the parameter calculation unit 11 calculates parameters related to the lifetime probability distribution f(t) of the N components. The lifetime probability distribution f(t) indicates the distribution of the probability that actual lifetime data with a failure lifetime of t is observed. The parameter calculation unit 11 outputs the calculation result P1 of the parameters to the lifetime probability distribution estimation unit 12.

[0013] The lifetime probability distribution estimation unit 12 is implemented, for example, by the lifetime probability distribution estimation circuit 22 shown in Figure 2. The lifetime probability distribution estimation unit 12 receives factor information d from the storage device 1. 2,i The parameters (i=1,···,I) are obtained, and the parameter calculation result P1 is obtained from the parameter calculation unit 11. The lifetime probability distribution estimation unit 12 uses factor information d 2,i Using the parameter calculation result P1, we estimate the lifetime probability distribution f(t). The lifetime probability distribution estimation unit 12 outputs the estimated result P2 of the lifetime probability distribution f(t) to the quality index calculation unit 13.

[0014] The quality index calculation unit 13 is implemented, for example, by the quality index calculation circuit 23 shown in Figure 2. The quality index calculation unit 13 obtains the estimated result P2 of the lifetime probability distribution f(t) from the lifetime probability distribution estimation unit 12. The quality index calculation unit 13 calculates a quality index P3 indicating the quality of the embedded component from the lifetime probability distribution f(t) estimated by the lifetime probability distribution estimation unit 12. The quality index calculation unit 13 displays the quality index P3, for example, on a display.

[0015] In Figure 1, the parameter calculation unit 11, lifetime probability distribution estimation unit 12, and quality index calculation unit 13, which are components of the quality index calculation device 2, are assumed to be implemented by dedicated hardware as shown in Figure 2. That is, the quality index calculation device 2 is assumed to be implemented by a parameter calculation circuit 21, a lifetime probability distribution estimation circuit 22, and a quality index calculation circuit 23. Each of the parameter calculation circuit 21, lifetime probability distribution estimation circuit 22, and quality index calculation circuit 23 can be, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a combination thereof.

[0016] The components of the quality index calculation device 2 are not limited to those implemented by dedicated hardware; the quality index calculation device 2 may also be implemented by software, firmware, or a combination of software and firmware. Software or firmware is stored as a program in the computer's memory. A computer refers to the hardware that executes programs, and includes components such as a CPU (Central Processing Unit), processing unit, arithmetic unit, microprocessor, microcomputer, processor, or DSP (Digital Signal Processor).

[0017] Figure 3 is a hardware configuration diagram of a computer when the quality index calculation device 2 is implemented by software or firmware, etc. When the quality index calculation device 2 is implemented by software or firmware, a program is stored in memory 31 that causes a computer to execute the respective processing procedures in the parameter calculation unit 11, the lifetime probability distribution estimation unit 12, and the quality index calculation unit 13. The computer's processor 32 then executes the program stored in memory 31.

[0018] Furthermore, Figure 2 shows an example in which each component of the quality index calculation device 2 is implemented by dedicated hardware, and Figure 3 shows an example in which the quality index calculation device 2 is implemented by software or firmware, etc. However, this is only one example, and some components of the quality index calculation device 2 may be implemented by dedicated hardware, while the remaining components may be implemented by software or firmware, etc.

[0019] Next, we will explain the operation of the quality index calculation device 2 shown in Figure 1. Figure 4 is a flowchart showing the quality index calculation method, which is the processing procedure of the quality index calculation device 2. The parameter calculation unit 11 calculates from the storage device 1 the component's lifespan βi Factor x that gives (i=1,···,I) i However, the actual lifespan data d represents the lifespan of each of the N components that are common to the embedded components. 1,k Obtain (k=1,···,N). N lifetime data points d 1,1 ~d 1,N This includes actual lifespan data for components with different specifications from the built-in components. N actual lifespan data d 1,1 ~d 1,N This may include actual lifespan data for components with the same specifications as the embedded components. Furthermore, the parameter calculation unit 11 calculates the lifespan of N components from the storage device 1, which is influenced by β. i Factor x that gives rise to i Factor information d that indicates 2,i Obtain it.

[0020] The actual lifespan data d1 is data indicating the lifespan of a component. For example, the actual lifespan data d1 includes the manufacturing date of the failed component, the failure date of the failed component, and the cause of failure of the failed component. The component's lifespan can be measured in hours or days. Factor information d2 is factor x that affects the lifespan of the component. i This is information indicating factor x. i These factors include, for example, factors related to the type of part, the manufacturer, the manufacturing plant, the materials used, or the structure. Factor information d2, which indicates the type of part, manufacturer, manufacturing plant, material, and structure, is recorded, for example, in a parts ledger.

[0021] It can be assumed that parts of the same type will have the same factors affecting their lifespan, and that parts from the same manufacturer will have the same factors affecting their lifespan. Furthermore, it can be assumed that parts from the same manufacturing plant will have the same factors affecting their lifespan, and that parts with the same structure will have the same factors affecting their lifespan. Examples of multiple parts of the same type include multiple parts with identical specifications. It is assumed that such multiple parts share the same factors that affect their lifespan. On the other hand, examples of multiple parts of different types include multiple parts that have the same function and performance as specified in the specifications, but whose shapes as specified in the specifications differ from each other. Such multiple parts may have different factors that affect their lifespan. Examples of multiple parts made from the same material include part D and part D', both of which use material C as their primary material, even though their functions and performance specifications are similar. It is assumed that the factors affecting the lifespan of such multiple parts are the same. Examples of multiple components with the same structure include component R, which has similar functions and performance specifications, and component R', which also has Q as its main internal structure. It is assumed that the factors affecting the lifespan of such multiple components are the same.

[0022] The parameter calculation unit 11 calculates the actual life data d 1,k (k=1,···,N) and factor information d 2,i Based on (i=1,···,I), the parameters for the lifetime probability distribution f(t) of N components are calculated (Step ST1 in Figure 4). The parameter calculation unit 11 outputs the parameter calculation result P1 to the lifetime probability distribution estimation unit 12.

[0023] The lifetime probability distribution f(t), as shown in Figure 5, represents the distribution of the probability of observing actual lifetime data d1, for example, where the failure lifetime is t. Failure lifetime refers to the lifespan of a non-repairable system until it fails. Non-repairable systems include disposable items that cannot be repaired even if they fail. Figure 5 is an explanatory diagram showing an example of a lifetime probability distribution f(t). The horizontal axis represents the failure lifetime t, and the vertical axis represents the probability that a lifetime data d1 with a failure lifetime of t is observed.

[0024] The parameter calculation process performed by the parameter calculation unit 11 will be described in detail below. The lifetime probability distribution f(t) is, for example, the probability density function of the Weibull distribution. The lifetime probability distribution f(t) is expressed by two parameters m and λ, as shown in equation (1) below. m is a parameter common to all parts. λ is a shape parameter that indicates the shape of the Weibull distribution. The parameter λ is expressed by the scale parameter η of the Weibull distribution, as shown in equation (2) below.

[0025] TIFF0007876351000001.tif34166

[0026] Specifically, when the two parameters m and λ are, for example, m=1 and λ=1 / 3000, the probability f(t=3000) of observing the actual lifetime data d1, where the failure lifetime is t=3000, is approximately 0.00012, as shown in equation (3) below. Furthermore, when the two parameters m and λ are, for example, m=1 and λ=1 / 6000, the probability f(t=3000) of observing the actual lifetime data d1, where the failure lifetime is t=3000, is approximately 0.000046, as shown in equation (4) below. Therefore, when m=1 and λ=1 / 3000, the probability f(t=3000) of observing the actual lifetime data d1 with a failure lifetime of t=3000 is higher than the probability f(t=3000) of observing the actual lifetime data d1 with a failure lifetime of t=3000 when m=1 and λ=1 / 6000. Therefore, for the two parameters m and λ, m=1 and λ=1 / 3000 are more appropriate than m=1 and λ=1 / 6000.

[0027] TIFF0007876351000002.tif33166

[0028] The parameter calculation unit 11 sets parameter m to a certain value, and then calculates parameter λ k By changing (k=1,···,N), we obtain N lifetime real data d as shown in equations (3) and (4). 1,k The probability f(t) of observing each of (k=1,···,N)k Calculate ). The parameter calculation unit 11 calculates N lifetime actual data d 1,k The probability f(t) of each being observed k The probabilities f(t) are added together. k The cumulative result of ) is expressed by the likelihood function L shown in equation (5) below.

[0029] TIFF0007876351000003.tif38166

[0030] parameter λ k The influence β on the component's lifetime probability distribution f(t) is as shown in equation (6) below. i And, influence β i Factor x that gives rise to i It is represented by and . The likelihood function L shown in equation (7) below is equal to the λ of the likelihood function L shown in equation (5). k For this, the λ shown in equation (6) k This is the value that was substituted.

[0031] TIFF0007876351000004.tif39166 In equation (6), β0 is the influence β i It is the reference value for β i This represents the difference from β0. In equation (7), λ0 is the parameter λ k It is the reference value for λ k This represents the difference from λ0.

[0032] The parameter calculation unit 11 calculates the parameters (m, λ0, β) when the likelihood function L shown in equation (7) is maximized. i We perform maximum likelihood estimation to infer the parameters (m, λ0, β) when the likelihood function L is maximized. i ) is the parameter at which the probability of observing all N lifetime real data d1 is highest. The parameters (m, λ0, β) when the likelihood function L is maximized. i For example, gradient descent can be used as a maximum likelihood estimation method to infer ). The parameter calculation unit 11 calculates the parameters (m, λ0, β) that maximize the likelihood function L. i The output is sent to the lifetime probability distribution estimation unit 12.

[0033] The lifetime probability distribution estimation unit 12 receives factor information d from the storage device 1. 2,i (i=1,···,I) The lifetime probability distribution estimation unit 12 uses factor information d 2,i Using the parameter calculation result P1, the lifetime probability distribution f(t) is estimated (step ST2 in Figure 4).

[0034] For example, factor x i When one of the factors included is a factor due to the type of part, then a concept of parts of the same type includes one or more parts. Therefore, the factor x due to the type of part i The number of actual life data d1 related to this is greater than or equal to the number of actual life data d1 related to a single component. For example, factor x i When one of the factors included is a factor related to the manufacturer of the part, then a concept of parts manufactured by the same manufacturer includes one or more parts. Therefore, the factor x related to the manufacturer of the part i The number of actual life data d1 related to this is greater than or equal to the number of actual life data d1 related to a single component. For example, factor x i When one of the factors included is a factor due to the material of the part, then a concept of parts made of the same material includes one or more parts. Therefore, the factor due to the material of the part x i The number of actual life data d1 related to this is greater than or equal to the number of actual life data d1 related to a single component. Therefore, the estimation accuracy of the lifetime probability distribution f(t) by the lifetime probability distribution estimation unit 12 is greater than or equal to the estimation accuracy of the lifetime probability distribution f(t) estimated from the actual lifetime data d1 related to a certain component. The lifetime probability distribution estimation unit 12 outputs the estimated result P2 of the lifetime probability distribution f(t) to the quality index calculation unit 13.

[0035] The following describes in detail the estimation process of the lifetime probability distribution f(t) by the lifetime probability distribution estimation unit 12. The lifetime probability distribution estimation unit 12 uses factor information d 2,i Factor x indicated by (i=1,···,I) i And the parameters (m, λ0, β) when the likelihood function L is maximized are... i The parameters included in ) (λ0,β i By substituting ) and into equation (6), the parameter λ k Calculate. Factor x indicated by factor information d2 i If the factor x is not expressed numerically, the lifetime probability distribution estimation unit 12 uses a numerical method such as one-hot encoding to determine the factor x i Quantify factor x i Substitute this into equation (6). Then, the lifetime probability distribution estimation unit 12 uses the parameter λ k Substitute λ into equation (1) and obtain the parameter (m, λ0, β i By substituting the parameter m included in (1) into equation (1), the lifetime probability distribution f(t) is calculated.

[0036] The quality index calculation unit 13 obtains the estimated result P2 of the lifetime probability distribution f(t) from the lifetime probability distribution estimation unit 12. The quality index calculation unit 13 calculates a quality index P3 indicating the quality of the embedded component from the lifetime probability distribution f(t) estimated by the lifetime probability distribution estimation unit 12 (step ST3 in Figure 4). The quality index calculation unit 13 displays the quality index P3, for example, on a display.

[0037] The following describes in detail the process by which the quality index calculation unit 13 calculates the quality index P3. Quality indicator P3 can indicate, for example, B10 life or expected lifespan. B10 life indicates the time it takes for 10% of the total parts to fail, while expected lifespan indicates the time it takes for 50% of the total parts to fail. When calculating the quality index P3, for example, B10 life, the quality index calculation unit 13 calculates the value of t at which the cumulative distribution function F(t) obtained from the life probability distribution f(t) becomes 0.1, as shown in equation (8) below. Furthermore, when calculating the expected lifetime as the quality index P3, the quality index calculation unit 13 calculates the expected lifetime E(t) obtained from the lifetime probability distribution f(t) as shown in equation (9) below.

[0038] TIFF0007876351000005.tif37166

[0039] In the above embodiment 1, the quality index calculation device 2 is configured to include a parameter calculation unit 11 that calculates parameters relating to the life probability distribution of multiple parts based on the actual life data and factor information relating to the factors influencing the life of multiple parts, for multiple parts including an embedded part which is a part incorporated into a finished product and parts which have different specifications from the embedded part, and a life probability distribution estimation unit 12 that estimates the life probability distribution using the factor information and the parameters calculated by the parameter calculation unit 11. Furthermore, the quality index calculation device 2 includes a quality index calculation unit 13 that calculates a quality index indicating the quality of the embedded part from the life probability distribution estimated by the life probability distribution estimation unit 12.Therefore, the quality index calculation device 2 can improve the accuracy of quality index calculation compared to the device disclosed in Patent Document 1.

[0040] In the quality index calculation device 2 shown in Figure 1, factor x i However, factors such as the type of part, the manufacturer, the manufacturing plant, the materials, or the structure are assumed to be factors without hierarchical relationships. However, this is merely an example, and there may be hierarchical relationships between the factors. For example, if the factor due to the type of part is the first level and the factor due to the manufacturer is the second level, then for each type of part, a reference value λ0 for the parameter λ is determined, and the variability λ1 relative to λ0 is determined by the factor due to the type of part. Furthermore, the variability λ2 relative to λ1 is determined by the difference in manufacturers. The parameter λ1 from the first level is expressed as shown in equation (10) below. The parameter λ2 from the second level is expressed as shown in equation (11) below. The parameter λ from the h-th level h This can be expressed as shown in equation (12) below, where h is an integer greater than or equal to 3.

[0041] TIFF0007876351000006.tif51166

[0042] In equation (10), x 1,i This is the first-level factor, β 1,i This is the first-level factor x 1,i It is the effect that is given by. In equation (11), x 2,i This is the second-level factor, β 2,i This is the second-level factor x 2,i It is the effect that is given by. In equation (12), x h,i This is the h-th level factor, β h,i This is the h-th level factor x h,i It is the effect that is given by. When there is a hierarchical relationship between the factors, the likelihood function L is given by λ of the likelihood function L shown in equation (5). k For this, λ1~λ shown in equation (6) h These are the values ​​that have been substituted into each of them.

[0043] When there is no hierarchical relationship between the factors, for example, even if the types of parts are different, the difference in part types does not affect the influence of the manufacturer. Therefore, regardless of the type of part, for example, manufacturer G's quality may be uniformly evaluated as poor. On the other hand, there is a hierarchical relationship between each factor. For example, if the first-level factor is the type of part and the second-level factor is the manufacturer, then the difference in part type will result in differences in the influence of the manufacturer. Therefore, for example, the quality of manufacturer G may be poor when the part type is A, but the quality of manufacturer G may be good when the part type is B.

[0044] Embodiment 2. Embodiment 2 describes a quality index calculation device 2 that includes a quality index correction unit 14 for correcting the quality index P3 calculated by the quality index calculation unit 13.

[0045] Figure 6 is a configuration diagram showing the quality index calculation device 2 according to Embodiment 2. In Figure 6, the same reference numerals as in Figure 1 indicate the same or corresponding parts, so their explanation is omitted. Figure 7 is a hardware configuration diagram showing the hardware of the quality index calculation device 2 according to Embodiment 2. In Figure 7, the same reference numerals as in Figure 2 indicate the same or corresponding parts, so their explanation is omitted. In Figure 6, the storage device 1 stores quality-related information d3 in addition to actual lifespan data d1 and factor information d2. Quality-related information d3 is not data indicating the lifespan of a component, but rather information related to the quality of a component that is considered useful when determining its quality. Information related to component quality includes usage data, such as the number of units shipped for each periodic period, as well as information showing the results of quality evaluation tests for the components. The quality evaluation test results are the results of quality evaluation tests for newly adopted components, and include, for example, the pass / fail status of the quality evaluation test, or the tester's opinion.

[0046] The quality index calculation device 2 shown in Figure 6 comprises a parameter calculation unit 11, a lifetime probability distribution estimation unit 12, a quality index calculation unit 13, and a quality index correction unit 14. The quality index correction unit 14 is implemented, for example, by the quality index correction circuit 24 shown in Figure 7. The quality index correction unit 14 receives quality-related information for each of the N components from the storage device 1. 3,k The quality index P3 calculated by the quality index calculation unit 13 is corrected using this method. The quality index correction unit 14 displays the corrected quality index P4 on a display, for example.

[0047] In Figure 6, it is assumed that the parameter calculation unit 11, lifetime probability distribution estimation unit 12, quality index calculation unit 13, and quality index correction unit 14, which are components of the quality index calculation device 2, are each implemented by dedicated hardware as shown in Figure 7. That is, it is assumed that the quality index calculation device 2 is implemented by a parameter calculation circuit 21, a lifetime probability distribution estimation circuit 22, a quality index calculation circuit 23, and a quality index correction circuit 24. Each of the parameter calculation circuit 21, lifetime probability distribution estimation circuit 22, quality index calculation circuit 23, and quality index correction circuit 24 can be, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC, an FPGA, or a combination thereof.

[0048] The components of the quality index calculation device 2 are not limited to those implemented by dedicated hardware; the quality index calculation device 2 may also be implemented by software, firmware, or a combination of software and firmware. When the quality index calculation device 2 is implemented by software or firmware, a program is stored in the memory 31 shown in Figure 3 that causes a computer to execute the respective processing procedures in the parameter calculation unit 11, lifetime probability distribution estimation unit 12, quality index calculation unit 13, and quality index correction unit 14. Then, the processor 32 shown in Figure 3 executes the program stored in the memory 31.

[0049] In addition, FIG. 7 shows an example in which each component of the quality index calculation device 2 is realized by dedicated hardware, and FIG. 3 shows an example in which the quality index calculation device 2 is realized by software, firmware, or the like. However, this is merely an example, and some components in the quality index calculation device 2 may be realized by dedicated hardware, and the remaining components may be realized by software, firmware, or the like.

[0050] Next, the operation of the quality index calculation device 2 shown in FIG. 6 will be described. Except for the quality index correction unit 14, it is the same as the quality index calculation device 2 shown in FIG. 1. Therefore, here, only the operation of the quality index correction unit 14 will be described.

[0051] The quality index correction unit 14 acquires respective quality-related information d 3,k (k = 1, ···, N) of N parts from the storage device 1, and acquires the quality index P3 from the quality index calculation unit 13. The quality index correction unit 14 calculates a weight coefficient w 3,k that satisfies the following conditional expression (13) from the quality-related information d i . The weight coefficient w i is calculated so as to have a larger value as the quality indicated by the quality-related information d 3,k is better.

[0052] TIFF0007876351000007.tif16166

[0053] Hereinafter, the calculation process of the weight coefficient w i by the quality index correction unit 14 will be specifically described. When the quality-related information d 3,k includes information indicating the results of the component quality evaluation test, if the quality evaluation test result indicates that it is qualified, the quality-related information d 3,k is used as a value z i obtained by quantifying it, and “1” is set to z i . If the quality index correction unit 14 indicates that the quality evaluation test result is unqualified, the quality-related information d 3,kThe quantified value z i is set to "0" for z i when calculating.

[0054] When the quality index correction unit 14 includes information indicating the usage record in the quality-related information d3, the usage record number y i is substituted into the following formula (14) or formula (15) to calculate the quantified value z i '.

[0055] In formula (5) of TIFF0007876351000008.tif29166, α is a non-zero constant.

[0056] The usage record number y i is used to calculate the value z i corresponding to the weight coefficient w i '. When calculating the value z i corresponding to the weight coefficient w i from the usage record number y i it is preferable that the usage record number y i is a symmetric normal distribution. However, since the usage record number y i generally follows a Poisson distribution, it has a long right tail and does not necessarily form a normal distribution. Therefore, the quality index correction unit 14 substitutes the usage record number y i into formula (14) or formula (15) to calculate z

[0057] The quality index correction unit 14 scales the quantified value z i ' so that the scale of the quantified value z i ' is within the range of 0 to 1, as shown in the following formula (16).

[0058] In formula (16) of TIFF0007876351000009.tif19166, z<00> i is the quantified value after scaling.

[0059] The quality index correction unit 14 uses z i where "1" or "0" is set, or z shown in formula (16)i By substituting this into the following equation (17), the weight coefficient w i Calculate.

[0060] In formula (17) of TIFF0007876351000010.tif13166, j is a parameter. Quality-related information d 3,k When emphasis is placed on the weight coefficient w i However, the parameter j is adjusted so that its value falls within the range of, for example, 0.2 to 1.8.

[0061] The quality index correction unit 14 uses the following equation (18) as the weight coefficient w i The quality index P3 is corrected by multiplying it by this value. The quality index correction unit 14 displays the corrected quality index P4 on a display, for example.

[0062] TIFF0007876351000011.tif14166

[0063] In the above embodiment 2, the quality index calculation device 2 shown in Figure 6 is configured to include a quality index correction unit 14 that corrects the quality index P3 calculated by the quality index calculation unit 13 using the quality-related information d3 of each component. Therefore, the quality index calculation device 2 shown in Figure 6 can improve the accuracy of quality index calculation compared to the device disclosed in Patent Document 1, and can also calculate quality indexes for newly adopted components or components for which actual lifespan data d1 is unavailable.

[0064] Embodiment 3. Embodiment 3 describes a quality index calculation device 2 that includes a ranking unit 15 that performs ranking of the quality index after correction by the quality index correction unit 14.

[0065] Figure 8 is a configuration diagram showing the quality index calculation device 2 according to Embodiment 3. In Figure 8, the same reference numerals as in Figures 1 and 6 indicate the same or corresponding parts, so their explanation is omitted. Figure 9 is a hardware configuration diagram showing the hardware of the quality index calculation device 2 according to Embodiment 3. In Figure 9, the same reference numerals as in Figures 2 and 7 indicate the same or corresponding parts, so their explanation is omitted. The quality index calculation device 2 shown in Figure 8 comprises a parameter calculation unit 11, a lifetime probability distribution estimation unit 12, a quality index calculation unit 13, a quality index correction unit 14, and a ranking unit 15.

[0066] The ranking unit 15 is implemented, for example, by the ranking circuit 25 shown in Figure 9. The ranking unit 15 obtains the corrected quality index P4 from the quality index correction unit 14. The ranking unit 15 performs ranking based on the corrected quality index P4. The ranking unit 15 displays the ranked and corrected quality index P4 on a display, for example.

[0067] In Figure 8, the parameter calculation unit 11, lifetime probability distribution estimation unit 12, quality index calculation unit 13, quality index correction unit 14, and ranking unit 15, which are components of the quality index calculation device 2, are assumed to be implemented by dedicated hardware as shown in Figure 9. Specifically, the quality index calculation device 2 is assumed to be implemented by a parameter calculation circuit 21, a lifetime probability distribution estimation circuit 22, a quality index calculation circuit 23, a quality index correction circuit 24, and a ranking circuit 25. Each of the parameter calculation circuit 21, lifetime probability distribution estimation circuit 22, quality index calculation circuit 23, quality index correction circuit 24, and ranking circuit 25 can be, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC, an FPGA, or a combination thereof.

[0068] The components of the quality index calculation device 2 are not limited to those implemented by dedicated hardware; the quality index calculation device 2 may also be implemented by software, firmware, or a combination of software and firmware. When the quality index calculation device 2 is implemented by software or firmware, a program for causing a computer to execute the respective processing procedures in the parameter calculation unit 11, lifetime probability distribution estimation unit 12, quality index calculation unit 13, quality index correction unit 14, and ranking unit 15 is stored in the memory 31 shown in Figure 3. Then, the processor 32 shown in Figure 3 executes the program stored in the memory 31.

[0069] Furthermore, Figure 9 shows an example in which each component of the quality index calculation device 2 is implemented by dedicated hardware, while Figure 3 shows an example in which the quality index calculation device 2 is implemented by software or firmware. However, this is merely one example, and some components of the quality index calculation device 2 may be implemented by dedicated hardware, while the remaining components may be implemented by software or firmware.

[0070] Next, the operation of the quality index calculation device 2 shown in Figure 8 will be explained. Except for the ranking unit 15, it is the same as the quality index calculation device 2 shown in Figure 6. Therefore, only the operation of the ranking unit 15 will be explained here.

[0071] The ranking unit 15 obtains the corrected quality index P4 from the quality index correction unit 14. The ranking unit 15 performs ranking based on the corrected quality index P4. The ranking unit 15 displays the ranked and corrected quality index P4 on a display, for example. The ranking process performed by the ranking unit 15 will be explained in detail below.

[0072] Figure 10 is an explanatory diagram showing the probability of the ranked quality index P4 occurring. In Figure 10, the horizontal axis represents the quality index P4, and the vertical axis represents the probability of quality index P4 occurring. When the ranking unit 15 ranks the corrected quality index P4 into U levels, the internal memory of the ranking unit 15 contains U-1 threshold values ​​Th uThe following is stored: u = 1, ..., U-1. U is an integer greater than or equal to 2. Threshold Th u This is not limited to what is stored in the internal memory of the ranking unit 15, but may also be provided from outside the quality index calculation device 2. For example, Th1 <Th2<Th3<···<Th U-1 That is the case. Note that the threshold Th u This may be a percentile value obtained from the lifetime probability distribution f(t), or it may be a value representing the mean ± kσ.

[0073] The ranking section 15 uses the corrected quality index P4 and the threshold Th. u Compare this with (u=1,···,U-1). The ranking section 15 uses a quality index P4 and a threshold Th. u Based on the comparison results, the quality indicator P4 is ranked as follows. P4 <Th1→ランク1 Th1≦P4 <Th2→ランク2 Th2≦P4 <Th3→ランク3 : Th U-2 ≤P4 <Th U-1 → rank U-1 Th U-1 ≤P4→Rank U

[0074] In the above embodiment 3, the quality index calculation device 2 is configured to include a ranking unit 15 that performs ranking of the quality index after correction by the quality index correction unit 14. Therefore, the quality index calculation device 2 shown in Figure 8 can improve the accuracy of quality index calculation compared to the device disclosed in Patent Document 1, and the degree of quality can be easily confirmed.

[0075] Furthermore, this disclosure allows for free combination of each embodiment, modification of any component in each embodiment, or omission of any component in each embodiment. [Explanation of symbols]

[0076] 1 Storage device, 2 Quality index calculation device, 11 Parameter calculation unit, 12 Lifetime probability distribution estimation unit, 13 Quality index calculation unit, 14 Quality index correction unit, 15 Ranking unit, 21 Parameter calculation circuit, 22 Lifetime probability distribution estimation circuit, 23 Quality index calculation circuit, 24 Quality index correction circuit, 25 Ranking circuit, 31 Memory, 32 Processor.

Claims

1. A parameter calculation unit obtains actual lifespan data for multiple components, including components with different specifications from the aforementioned integrated components, where factors affecting the lifespan of the components are common to the integrated components, and factor information indicating factors affecting the lifespan of the multiple components, and calculates parameters relating to the lifespan probability distribution of the multiple components based on the actual lifespan data and factor information for each of the multiple components. A lifetime probability distribution estimation unit estimates the lifetime probability distribution using the factor information and the parameters calculated by the parameter calculation unit, A quality index calculation unit calculates a quality index indicating the quality of the embedded component from the lifetime probability distribution estimated by the lifetime probability distribution estimation unit. A quality index calculation device equipped with the following features.

2. The quality index calculation device according to claim 1, further comprising a quality index correction unit that corrects the quality index calculated by the quality index calculation unit using quality-related information related to the quality of each of the plurality of parts.

3. The quality index calculation device according to claim 2, further comprising a ranking unit that performs ranking of the quality index after correction by the quality index correction unit.

4. The parameter calculation unit obtains actual lifespan data for multiple components, including components with different specifications from the embedded components, where factors affecting the lifespan of the components are common to the embedded components that are incorporated into the finished product, and factor information indicating the factors affecting the lifespan of the multiple components. Based on the actual lifespan data and factor information for each of the multiple components, the unit calculates parameters relating to the lifespan probability distribution of the multiple components. The lifetime probability distribution estimation unit estimates the lifetime probability distribution using the factor information and the parameters calculated by the parameter calculation unit. The quality index calculation unit calculates a quality index indicating the quality of the embedded component from the lifetime probability distribution estimated by the lifetime probability distribution estimation unit. Quality index calculation method.

Images