Complete machine life prior information extraction method and device, equipment and storage medium

Abstract

The invention provides a prior information extraction method and device for the service life of a complete machine, equipment and a storage medium. The method comprises the following steps: acquiring preset reliability, and determining reliable life of a plurality of component products in a target complete machine under the preset reliability; determining a median absolute deviation of the reliable life of the plurality of component products, and determining a short-life component product of which the reliable life and the median absolute deviation meet a preset short-life relationship condition from the plurality of component products; according to the short-life assembly product, constructing a plurality of similar complete machines similar to the target complete machine; determining a shape parameter and a scale parameter of each similar complete machine in the plurality of similar complete machines; and through linear fitting, according to the shape parameter and the scale parameter, determining life parameter prior information of the target complete machine, and according to the life parameter prior information, carrying out storage period evaluation on the target complete machine.

Description

Technical Field

[0001] The present disclosure relates to the field of data processing technology, and in particular to a method, device, equipment and storage medium for extracting prior information of the life of a whole machine. Background Art

[0002] The storage period evaluation of the whole machine often faces the problems of high specimen cost and long test time, resulting in small sample size and low credibility of evaluation results. Using the prior information of life parameters based on expert experience for storage period evaluation can overcome the shortcomings of traditional statistical evaluation methods in small sample scenarios. However, the prior information of life parameters determined based on expert experience is greatly affected by the subjective factors of experience judgment, which makes the prior information of life parameters insufficient to support the evaluation results.

[0003] In view of this, how to accurately extract the prior information of life parameters has become a technical problem that needs to be solved urgently. Summary of the invention

[0004] In view of this, the purpose of the present disclosure is to propose a method, device, equipment and storage medium for extracting prior information of the life of a whole machine to solve or partially solve the above technical problems.

[0005] Based on the above purpose, the first aspect of the present disclosure proposes a method for extracting prior information of the life of a whole machine, the method comprising:

[0006] Obtaining a preset reliability and determining the reliable life of multiple component products in the target whole machine under the preset reliability;

[0007] Determine the median absolute deviation of the reliable life of the plurality of component products, and determine, from the plurality of component products, a short-life component product whose reliable life and the median absolute deviation satisfy a preset short-life relationship condition;

[0008] constructing a plurality of similar complete machines similar to the target complete machine based on the short-life component product;

[0009] Determining shape parameters and scale parameters of each similar whole machine among the plurality of similar whole machines;

[0010] The life parameter prior information of the target whole machine is determined according to the shape parameter and the scale parameter through linear fitting, and the storage period of the target whole machine is evaluated according to the life parameter prior information.

[0011] Based on the same inventive concept, the second aspect of the present disclosure proposes a device for extracting prior information of the life of a whole machine, comprising:

[0012] A reliable life determination module is configured to obtain a preset reliability and determine the reliable life of multiple component products in the target whole machine under the preset reliability;

[0013] a short-life component product determination module, configured to determine the median absolute deviation of the reliable lifespans of the plurality of component products, and determine, from the plurality of component products, a short-life component product whose reliable lifespan and the median absolute deviation satisfy a preset short-life relationship condition;

[0014] A construction module, configured to construct a plurality of similar complete machines similar to the target complete machine based on the short-life component product;

[0015] A parameter determination module, configured to determine a shape parameter and a scale parameter of each similar whole machine in the plurality of similar whole machines;

[0016] The evaluation module is configured to determine the life parameter prior information of the target whole machine according to the shape parameter and the scale parameter through linear fitting, and perform a storage period evaluation on the target whole machine according to the life parameter prior information.

[0017] Based on the same inventive concept, the third aspect of the present disclosure proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor implements the method as described above when executing the computer program.

[0018] Based on the same inventive concept, a fourth aspect of the present disclosure proposes a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the method as described above.

[0019] As can be seen from the above, the method, device, equipment and storage medium for extracting prior information of the life of the whole machine provided by the present disclosure. Obtain a preset preset reliability, and determine the reliable life of multiple component products in the target whole machine under the preset reliability. Determine the median absolute deviation of the reliable life of multiple component products, and determine the short-life component products that meet the preset short-life relationship conditions between the reliable life and the median absolute deviation from the multiple component products. According to the short-life component products, construct multiple similar whole machines similar to the target whole machine. Determine the shape parameters and scale parameters of each similar whole machine in the multiple similar whole machines. Through linear fitting, determine the life parameter prior information of the target whole machine according to the shape parameters and scale parameters, and evaluate the storage period of the target whole machine according to the life parameter prior information. In this way, by constructing multiple similar whole machines similar to the target whole machine, extracting the life parameter prior information of the target whole machine from the component products in the similar whole machine, the source of the life parameter prior information is more objective, and the extraction process is completely dependent on objective data, which can avoid the problem of unstable and inconsistent storage period evaluation results due to subjective differences in expert experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the present disclosure or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, the drawings described below are only embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 It is a flow chart of a method for extracting prior information of the whole machine life according to an embodiment of the present disclosure;

[0022] Figure 2 It is a schematic diagram of the first parameter fitting result of a similar whole machine according to an embodiment of the present disclosure;

[0023] Figure 3 It is a schematic diagram of the second parameter fitting result of a similar whole machine according to an embodiment of the present disclosure;

[0024] Figure 4 It is a schematic diagram of shape parameters and prior fitting results of similar products according to an embodiment of the present disclosure;

[0025] Figure 5 It is a schematic diagram of scale parameters and prior fitting results of similar products of the embodiment of the present disclosure;

[0026] Figure 6 This is a flow chart of extracting prior information of life parameters based on similar whole machines according to an embodiment of the present disclosure;

[0027] Figure 7It is a structural schematic diagram of a device for extracting prior information of the whole machine life according to an embodiment of the present disclosure;

[0028] Figure 8 It is a schematic diagram of the structure of an electronic device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0029] In order to make the objectives, technical solutions and advantages of the present disclosure more clearly understood, the present disclosure is further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.

[0030] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present disclosure should be understood by people with ordinary skills in the field to which the present disclosure belongs. The "first", "second" and similar words used in the embodiments of the present disclosure do not indicate any order, quantity or importance, but are only used to distinguish different components. "Including" or "comprising" and similar words mean that the elements or objects appearing before the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0031] Based on the description of background technology, the storage period evaluation of whole machine products often faces the problems of high test piece cost and long test time, resulting in small sample size and low credibility of evaluation results. The Bayesian method can make full use of prior information such as similar products, historical data, expert experience, and current test data of the whole machine to give the storage period evaluation results of the whole machine, which can overcome the shortcomings of traditional statistical evaluation methods in small sample scenarios. However, traditional prior information acquisition methods are too dependent on product historical information and similar product information, and are greatly affected by subjective factors such as experience judgment. In scenarios where product historical information and similar products that can be used as reference are scarce, the collected prior information is insufficient to support the evaluation results.

[0032] As mentioned above, how to accurately extract prior information of life parameters has become an important research issue.

[0033] Based on the above description, if Figure 1 As shown, the method for extracting prior information of the whole machine life proposed in this embodiment includes:

[0034] Step 101 , obtaining a preset reliability, and determining the reliable life of multiple component products in a target whole machine under the preset reliability.

[0035] In specific implementation, component products are also called low-level components, which are the components, assemblies and parts at all levels in the target complete machine.

[0036] Obtain product information of multiple component products of the target machine, wherein the product information includes at least one of the following: life distribution type of component products, life distribution parameter of component products (target machine reference temperature), and number of component products in the target machine (target machine installed quantity).

[0037] According to the life distribution type and life distribution parameters of the component product, the reliable life of the component product is determined. The reliable life of each component product in the target whole machine is determined respectively, and the reliable life of multiple component products in the target whole machine is obtained.

[0038] Step 102, determining the median absolute deviation of the reliable life of the plurality of component products, and determining short-life component products from the plurality of component products whose reliable life and the median absolute deviation satisfy a preset short-life relationship condition.

[0039] In specific implementation, when the reliable life of a component product and the median absolute deviation meet the preset short-life relationship conditions, and the reliable life of the component product is less than the median life, the component product is determined to be a short-life component product (extremely short-life low-layer component).

[0040] Step 103: construct a plurality of similar complete machines similar to the target complete machine based on the short-life component product.

[0041] In the specific implementation, first, select the short-lived component products in the target machine. Then, according to the preset similarity, select the component products that are the same as the target machine. Finally, select the component products that are different from the target machine to construct a similar machine. Repeat the above steps to construct multiple similar machines.

[0042] Step 104: determine the shape parameters and scale parameters of each similar whole machine in the plurality of similar whole machines.

[0043] In a specific implementation, the life distribution type of each similar whole machine among multiple similar whole machines is obtained. When the life distribution type of each similar whole machine is a Weibull distribution type, the linear relationship of the life distribution parameters of each similar whole machine is determined according to the Weibull distribution type. Based on the linear relationship of the life distribution parameters of each similar whole machine, the shape parameters and scale parameters of each similar whole machine are determined.

[0044] Step 105 , determining life parameter prior information of the target whole machine according to the shape parameter and the scale parameter through linear fitting, and performing storage period evaluation on the target whole machine according to the life parameter prior information.

[0045] In specific implementation, the degree of fit between the shape parameters and scale parameters of multiple similar machines is determined, and a linear fitting model is determined according to the degree of fit. Based on the linear fitting model, the shape parameters and scale parameters of multiple similar machines are fitted to obtain the life parameters of multiple similar machines. The life parameters of multiple similar machines are used as the prior information of the life parameters of the target machine.

[0046] The target machine is composed of components, assemblies, and parts at various levels. The product information of each level of components also contains the life information of the target machine. The component product information has low acquisition cost, large data volume, and little influence of subjective factors. Therefore, it can be used as a priori information source for the storage period evaluation of the whole machine.

[0047] Through the above-mentioned embodiments, a preset preset reliability is obtained, and the reliable life of multiple component products in the target whole machine under the preset reliability is determined. The median absolute deviation of the reliable life of multiple component products is determined, and short-lived component products that meet the preset short-life relationship conditions between the reliable life and the median absolute deviation are determined from multiple component products. Based on the short-lived component products, multiple similar whole machines similar to the target whole machine are constructed. The shape parameters and scale parameters of each similar whole machine in the multiple similar whole machines are determined. Through linear fitting, the life parameter prior information of the target whole machine is determined according to the shape parameters and scale parameters, and the storage period of the target whole machine is evaluated based on the life parameter prior information. In this way, by constructing multiple similar whole machines similar to the target whole machine, the life parameter prior information of the target whole machine is extracted from the component products in the similar whole machines, so that the source of the life parameter prior information is more objective, and the extraction process is completely dependent on objective data, which can avoid the problem of unstable and inconsistent storage period evaluation results due to subjective differences in expert experience.

[0048] In some embodiments, step 101 includes:

[0049] Step 1011, obtaining a preset reliability.

[0050] Step 1012: for each component product among the plurality of component products, determine the life distribution type of the component product.

[0051] Step 1013, in response to determining that the life distribution type of the component product is an exponential distribution type, based on the exponential distribution type, determining a reliable life of the component product under the preset reliability,

[0052]

[0053] Among them, i R is the reliable life of the component product, λ is the failure rate of exponential distribution, and R is the preset reliability.

[0054] Step 1014, in response to determining that the life distribution type of the component product is a Weibull distribution type, determining a reliable life of the component product at the preset reliability based on the Weibull distribution type,

[0055]

[0056] Among them, t R is the reliable life of the component product, η is the scale parameter of the Weibull distribution, m is the shape parameter of the Weibull distribution, and R is the preset reliability.

[0057] Step 1015, in response to determining that the life distribution type of the component product is a log-normal distribution type, determining a reliable life of the component product at the preset reliability based on the log-normal distribution type,

[0058]

[0059] Among them, t R is the reliable life of the component product, Φ -1 (·) is the inverse function of the standard normal distribution, μ is the logarithmic mean of the lognormal distribution, σ is the logarithmic variance of the lognormal distribution, and R is the preset reliability.

[0060] In a specific implementation, product information of multiple component products of the target machine is obtained, wherein the product information includes at least one of the following: life distribution type of component products, life distribution parameter of component products (target machine reference temperature), and number of component products in the target machine (target machine installed quantity).

[0061] The life distribution type of the component product includes at least one of the following: exponential distribution type, Weibull distribution type and lognormal distribution type. The life distribution parameters of the component product include: failure rate of exponential distribution, scale parameter of Weibull distribution, logarithmic mean of lognormal distribution, shape parameter of Weibull distribution and logarithmic variance of lognormal distribution.

[0062] For example, the collected data on the life distribution type, life distribution parameters and the number of component products (installed number) of multiple component products (low-level components) in the target whole machine are shown in Table 1 below.

[0063] Table 1 Component product information of the target machine

[0064]

[0065]

[0066] According to the life distribution type and life distribution parameters of the component product, the reliable life of the component product is determined. The reliable life of each component product in the target whole machine is determined respectively, and the reliable life of multiple component products in the target whole machine is obtained.

[0067] Considering that when conducting storage period assessment for the target whole machine, the focus is usually on the reliable life of the target whole machine under high reliability (usually 0.8, 0.85 or 0.9 reliability is selected), the reliable life of the component products under the series assumption is higher than that of the target whole machine. Therefore, when evaluating whether the component product is a short-life component product, the focus should be on the reliable life of the component product under high reliability. Here, it is recommended to select 0.9 reliability as the evaluation standard. In actual application scenarios, the corresponding reliability level can also be determined according to the actual situation of the specific product.

[0068] According to engineering practice, the life of component products is usually of exponential distribution type, Weibull distribution type or log-normal distribution type. The specific calculation formula for the reliable life of component products is as follows:

[0069]

[0070] Among them, t R is the reliable life of the component product, R is the reliability of the reliable life, λ is the failure rate of the exponential distribution, η is the scale parameter of the Weibull distribution, m is the shape parameter of the Weibull distribution, Φ -1 (·) is the inverse function of the standard normal distribution, μ is the logarithmic mean of the lognormal distribution, and σ is the logarithmic variance of the lognormal distribution.

[0071] Through the above scheme, the reliable life of the component product can be accurately determined according to the life distribution type and life distribution parameters of the component product. For different life distribution types, the reliable life of the component product is determined according to the corresponding life distribution parameters and algorithms. When the life distribution type of the component product is an exponential distribution type, the reliable life of the component product can be accurately obtained according to the failure rate of the exponential distribution using the exponential function. When the life distribution type of the component product is a Weibull distribution type, the reliable life of the component product can be accurately obtained according to the scale parameter and shape parameter of the Weibull distribution using the Weibull function. When the life distribution type of the component product is a lognormal distribution type, the reliable life of the component product can be accurately obtained according to the logarithmic mean and logarithmic variance of the lognormal distribution using the logarithmic function.

[0072] In some embodiments, step 102 includes:

[0073] Step 1021, determining the median lifespan of the plurality of component products, and determining the median absolute deviation based on the reliable lifespan of each component product and the median lifespan of the plurality of component products,

[0074] MAD=median(|t R,i -median(t R )|)

[0075] Wherein, MAD is the median absolute deviation, t R,i is the reliable life of the i-th component product, median(·) is the median, median(t R ) is the median life span of the multiple component products.

[0076] Step 1022: in response to determining that the reliable life of the i-th component product and the median absolute deviation satisfy a preset short-life relationship condition, and the reliable life of the i-th component product is less than the median life, determining that the i-th component product is a short-life component product.

[0077] |t R,i -median(t R )|>3k·MAD

[0078] Among them, |t R,i -median(t R )|>3k MAD is the condition that the reliable life of the i-th component product and the median absolute deviation meet the preset short-life relationship, t R,i is the reliable life of the i-th component product, median(t R ) is the median of the reliable life of the multiple component products, MAD is the median absolute deviation, and k is the proportional factor constant.

[0079] In specific implementation, the distance between the reliable life of multiple component products and the median life is used to determine whether the reliable life of each component product is an outlier data. The distance benchmark is the median absolute deviation MAD (Median Absolute Deviation).

[0080] For each component product's reliable life, if the reliable life meets the preset short-life relationship condition, then the reliable life of this component product is determined to be an outlier. In the above formula, the proportional factor constant k can correct the median absolute deviation MAD to a consistent estimator of the standard deviation σ. In the normal distribution scenario, Φ -1 (·) is the inverse function of the standard normal distribution.

[0081] When the reliable life of the i-th component product and the median absolute deviation meet the preset short-life relationship condition, and the reliable life of the i-th component product is less than the median life, it is determined that the i-th component product is an extremely small outlier, that is, the i-th component product is a short-life component product (extremely short-life low-level component).

[0082] Based on the life distribution information of the component products of the target complete machine, calculate and analyze the reliable life of each component product when the reliability is 0.9 to identify short-life component products (extremely short-life low-level components). The results are summarized in Table 2 below, and the results show that no short-life component products are found among the component products of the target complete machine.

[0083] Table 2 Reliable life of each component product in the target complete machine and short-life component products

[0084]

[0085]

[0086] Through the above solution, by determining the median life of multiple component products and determining the median absolute deviation based on the reliable life of each component product and the median life of multiple component products, it is possible to accurately judge whether the reliable life of the component product is an outlier data based on the median absolute deviation, and further judge whether the component product is a short-life component product.

[0087] In some embodiments, step 103 includes:

[0088] Step 1031, determine the total number of all component products in the target complete machine and the first number of the short-life component products.

[0089] Step 1032, obtain the similarity degree between the target complete machine and the similar complete machine preset in advance.

[0090] Step 1033, determine the second number of the same component products based on the total number, the similarity degree and the first number, and determine the third number of the different component products based on the total number and the similarity degree.

[0091] Step 1034, select the first number of short-life component products, select the second number of the same component products in a non-replacement sampling manner, select the third number of the different component products in a replacement sampling manner, construct a similar complete machine similar to the target complete machine, and obtain multiple similar complete machines by repeating the construction process of the similar complete machine.

[0092] In specific implementation, the Bootstrap sampling method is adopted, and component products are sampled without repetition according to a ratio as the components of the similar complete machine, and a similar complete machine similar to the target complete machine is constructed based on the series model. If the target complete machine has a total of N component products, numbered i = 1, 2,..., N, that is, the total number is N. Among them, M (M < N) component products are short-life component products, corresponding to the numbers i = 1, 2,..., M, that is, the first number is M. For example, if the similarity degree is A, the process of sampling and constructing a similar complete machine according to the ratio A is specifically as follows:

[0093] Step 1: directly use the short-lived component products of the target whole machine as part of the component products of the similar whole machine, and record the serial number in the constructed similar whole machine as j = 1, 2, ..., M;

[0094] Step 2: From all N component products of the target machine, extract [NA]-M component products without replacement, where [·] represents rounding down, and the serial numbers in the constructed similar machine are recorded as j=M+1,M+2,…,[NA];

[0095] Step 3: From all the N component products of the target machine, N-[NA] component products are extracted with replacement, and their serial numbers in the constructed similar machine are recorded as j=[NA]+1,[NA]+2,…,N;

[0096] According to the above three steps, a similar whole machine can be obtained. Since the component products with a proportion of A are exactly the same as the target whole machine, it can be considered that the similarity between the constructed similar whole machine and the target whole machine is greater than or equal to A. For example, if the proportion A is set to 80%, it can be considered that the constructed similar whole machine has at least 80% similarity with the target whole machine.

[0097] By repeating the above construction process multiple times, multiple similar machines can be obtained. These sampled similar machines are used as input to further extract the prior information of the life parameters of the target machine.

[0098] According to the aforementioned similar product construction steps, K similar products (K=2000) of the target whole machine are constructed according to the similarity degree of 80%. The following Table 3 shows some of the constructed similar whole machines.

[0099] Table 3 Installed quantity of component products of multiple similar complete machines

[0100]

[0101]

[0102] Through the above scheme, by determining the total number of all component products in the target machine and the first number of short-lived component products, based on the total number, the preset similarity and the first number, the second number of identical component products and the third number of different component products can be accurately determined. The first number of short-lived component products are selected, the second number of identical component products are selected by non-replacement sampling, and the third number of different component products are selected by replacement sampling. In this way, the similar machine constructed has exactly the same short-lived components as the target machine, and the similarity between the similar machine and the target machine reaches the preset similarity.

[0103] In some embodiments, step 104 includes:

[0104] Step 1041, determining the shape parameters of each similar whole machine in the plurality of similar whole machines,

[0105]

[0106] Among them, m k is the shape parameter of the kth similar machine, t is the current time, λ ki (t) is the failure rate function of the i-th component product in the k-th similar whole machine, R ki (t) is the failure rate function of the i-th component product in the k-th similar whole machine, and N is the number of component products in the k-th similar whole machine.

[0107] Step 1042, determining the dimension parameters of each similar whole machine in the plurality of similar whole machines,

[0108]

[0109] Among them, η k is the scale parameter of the kth similar machine, R ki (1) is the failure rate function of the i-th component product in the k-th similar whole machine at time t=1.

[0110] In a specific implementation, the life distribution type of each similar whole machine among a plurality of similar whole machines is obtained;

[0111] In response to determining that the life distribution type of each similar whole machine is a Weibull distribution type, a linear relationship of the life distribution parameters of each similar whole machine is determined according to the Weibull distribution type,

[0112] ln[-ln(1-F(t))]=ln[-ln(R(t))]=mlnt-mlnη

[0113] Among them, F(t) is the unreliability of each similar whole machine from time 0 to time t, R(t) is the reliability of each similar whole machine from time 0 to time t, m is the shape distribution parameter in the life distribution parameters of each similar whole machine, and η is the scale distribution parameter in the life distribution parameters of each similar whole machine.

[0114] Assume that the reliability of the kth similar product constructed by sampling is R Zk (t), k = 1, 2, ..., K, the reliability function, failure probability density function, and failure rate function of the lower layer components are denoted as R ki (t), f ki (t), λ ki (t), i = 1, 2, ..., N, then the shape parameter m of the similar product can be obtained based on the linear relationship of the life distribution parameters of each similar whole machine and the series model ksatisfy:

[0115]

[0116] Among them, -f(t) / R(t) is the calculation formula of failure rate, so the shape parameter m k It can also be expressed by the following formula:

[0117]

[0118] In order to obtain the shape parameter m k Based on , the scale parameter can be obtained by calculating the reliability at time t = 1:

[0119]

[0120] Through the above scheme, the linear relationship of the life distribution parameters of each similar whole machine can accurately determine the shape parameters and scale parameters of each similar whole machine, so as to fit the shape parameters and scale parameters of multiple similar whole machines to obtain the life parameters of multiple similar whole machines, and then determine the life prior information of the target whole machine.

[0121] In some embodiments, step 105 includes:

[0122] Step 1051 , determining the degree of fit between shape parameters and scale parameters of a plurality of similar complete machines, and determining a linear fitting model according to the degree of fit.

[0123] Step 1052: Based on the linear fitting model, shape parameters and scale parameters of a plurality of similar complete machines are fitted to obtain life parameters of the plurality of similar complete machines.

[0124] Step 1053: Use the life parameters of the plurality of similar machines as prior information on the life parameters of the target machine.

[0125] In specific implementation, the linear fitting model includes a general linear model and a two-stage linear model. When the linear relationship between the life distribution parameters of each similar whole machine is strong, a general linear model or a two-stage linear model is used to represent it. Among them, the general linear model and the two-stage linear model are specifically as follows:

[0126] y=ax+bGeneral Linear Model

[0127]

[0128] When the goodness of fit between the linear relationships of the life distribution parameters of each similar whole machine is high (>0.95), the general linear model parameter fitting results are used for life distribution parameter estimation. When the goodness of fit between the linear relationships of the life distribution parameters of each similar whole machine is not high enough (≤0.95), the two-stage linear model is used to fit the sample values, and the parameter fitting results of the first stage (x≤c) are selected for life distribution parameter estimation of similar whole machines.

[0129] According to the linear relationship of the life distribution parameters of each similar whole machine and the linear fitting model, the constructed life distribution parameters of the similar whole machine are fitted and estimated using the numerical method. The curves of lnt and ln[-ln(R(t))] of each similar whole machine are as follows: Figure 2 As shown. The shape parameter m can be obtained by linear fitting k and the scale parameter η k ,k=1,2,…,K’s fitting results. Figure 2 It is a schematic diagram of the first parameter fitting result of a similar whole machine according to an embodiment of the present disclosure. Figure 2 It contains the fitting curves of lnt and ln[-ln(R(t))] of all similar whole machines, and each simulation curve represents the fitting curve of lnt and ln[-ln(R(t))] of a similar whole machine.

[0130] As mentioned above, in order to avoid the influence of data correlation on the assumption of parameter independence, half of the data is used to fit the prior information of shape parameters and scale parameters respectively. The fitting results of the parameters of similar products after segmentation are plotted on the two-dimensional coordinate system as Figure 3 shown. Figure 3 It is a schematic diagram of the second parameter fitting results of a similar whole machine according to an embodiment of the present disclosure. Figure 3 Half of the similar whole machine life parameters are used to fit the shape parameter prior information, and half of the similar whole machine life parameters are used to fit the scale parameter prior information.

[0131] Figure 4 It is a schematic diagram of shape parameters and prior fitting results of similar products according to an embodiment of the present disclosure. Figure 4 There is a histogram of the estimation results of shape parameters of similar whole machines in , and it can be seen that the shape parameters obtained from the constructed similar whole machines show a trend of unimodal distribution. Figure 5 It is a schematic diagram of the scale parameters and prior fitting results of similar products of the embodiment of the present disclosure. Figure 5There is a histogram of the estimated results of the scale parameters of similar whole machines in , and the scale parameters obtained from the constructed similar whole machines show a unimodal distribution trend. Therefore, the normal distribution is selected as the prior distribution of the shape parameter and the scale parameter. Thus, the prior distribution of the target whole machine shape parameter m is further fitted to be the normal distribution N(2.2659, 0.0486), and the prior distribution of the scale parameter η is the normal distribution N(12.3048, 0.6428).

[0132] Through the above scheme, the degree of fit between the shape parameters and scale parameters of multiple similar machines is determined, and the linear fitting model determined according to the degree of fit is more suitable for fitting the shape parameters and scale parameters of multiple similar machines. Based on the linear fitting model, the shape parameters and scale parameters of multiple similar machines are fitted, and the life parameters of multiple similar machines can be accurately obtained, thereby determining the life prior information of the target machine.

[0133] In some embodiments, step 105 includes:

[0134] Step 1054, using a Bayesian algorithm, determine the life parameter posterior information according to the life parameter prior information,

[0135]

[0136] Among them, π (2) (θ|x) is the posterior information of the life parameter, L(x|θ) is the likelihood function of θ given the test data x, π (1) (θ) is the prior information of the lifespan parameter, ∫L(x|θ)π (1) (θ)dθ is a constant.

[0137] In specific implementation, the life parameter a posteriori information is determined based on the life parameter a priori information by using the Bayesian algorithm. The life parameter a posteriori information includes the reliability evaluation result of the target whole machine, thereby realizing the storage period evaluation of the target whole machine.

[0138] Through the above scheme, the life posterior information of the target whole machine can be accurately determined according to the life parameter prior information through the Bayesian algorithm.

[0139] Through the above-mentioned embodiments, a preset preset reliability is obtained, and the reliable life of multiple component products in the target whole machine under the preset reliability is determined. The median absolute deviation of the reliable life of multiple component products is determined, and short-lived component products that meet the preset short-life relationship conditions between the reliable life and the median absolute deviation are determined from multiple component products. Based on the short-lived component products, multiple similar whole machines similar to the target whole machine are constructed. The shape parameters and scale parameters of each similar whole machine in the multiple similar whole machines are determined. Through linear fitting, the life parameter prior information of the target whole machine is determined according to the shape parameters and scale parameters, and the storage period of the target whole machine is evaluated based on the life parameter prior information. In this way, by constructing multiple similar whole machines similar to the target whole machine, the life parameter prior information of the target whole machine is extracted from the component products in the similar whole machines, so that the source of the life parameter prior information is more objective, and the extraction process is completely dependent on objective data, which can avoid the problem of unstable and inconsistent storage period evaluation results due to subjective differences in expert experience.

[0140] It should be noted that the embodiments of the present disclosure may be further described in the following manner:

[0141] The Bayes method is to modify and update the prior knowledge by introducing new information to make full use of the prior information and the newly acquired information. Its general form is shown in the following formula:

[0142]

[0143] In the formula, π (2) (θ|x) represents the modified probability density function of parameter θ, i.e., the posterior distribution; L(x|θ) is the likelihood function of θ given the test data x; π (1) (θ) is the prior knowledge of θ, that is, the prior distribution; ∫L(x|θ)π (1) (θ)dθ is a constant. In the present invention, the prior information of the parameters of the whole machine product is extracted from the information of each level of products (components, assemblies, parts, etc.) constituting the whole machine, and the newly acquired information is the test information of the whole machine product.

[0144] The series model is a commonly used reliability model in engineering. It can calculate the reliability function of the whole product based on the reliability function of the lower-level components. The general form is:

[0145]

[0146] However, using all the low-level component information can only obtain a priori information on the reliability of the whole product, and it is impossible to obtain the priori distribution information of the relevant parameters of the whole machine. Therefore, the present invention proposes a similar product construction method to achieve the expansion of the priori information of the whole machine life distribution parameters.

[0147] This method draws on the Bootstrap sampling method, extracts the next level of products in proportion without duplication, and constructs similar components (complete machines) based on the series model to extract the prior distribution π of the life distribution parameter θ of the complete machine from the life distribution information of the lower-level components that constitute the complete machine. (1) (θ). The specific extraction process of the prior distribution is as follows:

[0148] Figure 6 FIG. 1 is a flow chart of extracting prior information of life parameters based on similar complete machines according to an embodiment of the present disclosure. Figure 6 As shown, the method for extracting prior information of life parameters based on similar complete machines includes:

[0149] Step 201: Collecting low-level component information of the whole machine

[0150] The low-level component information that needs to be collected to constitute the whole machine mainly includes: low-level component life distribution type, life distribution parameters (whole machine base temperature) and the number of installed machines.

[0151] Step 202, identification of extremely short-lived critical low-level components

[0152] For the whole machine product, some of its low-level components have a short life span and are key components that affect the life span of the whole machine. In order to ensure that the life span information of the constructed similar products has a certain reference value for the target whole machine, the low-level components of the constructed similar products should also include these key components. To this end, the present invention needs to identify the extremely short-lived key low-level components in the target whole machine before constructing similar products, which is mainly divided into two steps:

[0153] (1) Calculation of the reliable life of the lower layer components

[0154] Considering that when evaluating the storage period of the whole machine, the focus is usually on its reliable life under high reliability (usually 0.8, 0.85 or 0.9 reliable life), the reliable life of the low-level components under the series assumption is higher than that of the whole machine product. Therefore, when evaluating whether the low-level components are short-life products, attention should be paid to their reliable life under high reliability. Here, it is recommended to select 0.9 reliable life as the evaluation standard. In actual application scenarios, the corresponding reliability level can also be determined according to the actual situation of the specific product.

[0155] According to engineering practice, the life of the lower layer components is usually exponential distribution, Weibull distribution or log-normal distribution, and its reliable life can be calculated according to the following formula:

[0156]

[0157] Among them, t Ris the reliable life when the reliability is R; λ is the failure rate of exponential distribution; η and m are the scale parameter and shape parameter of Weibull distribution respectively; μ and σ are the logarithmic mean and logarithmic variance of lognormal distribution respectively, Φ -1 (·) is the inverse function of the standard normal distribution.

[0158] (2) Identify the minimum outliers of the reliable life of low-level components

[0159] The distance between each data and the median is used to determine whether the reliable life data is outlier data. The distance benchmark is the median absolute deviation (MAD), and its calculation formula is as follows:

[0160] MAD=median(|t R,i -median(t R )|) (4)

[0161] Where MAD is the median absolute deviation, t R,i is the reliable life of the ith lower layer component, median(·) is the median of a given sequence, and median(t R ) is the lifetime sequence of the lower layer components t R of the median.

[0162] For the reliable life data of a certain low-level component, if it satisfies the following formula, the data is considered to be an outlier:

[0163] |t R,i -median(t R )|>3k·MAD (5)

[0164] Where k is a proportional factor constant, which can correct MAD to a consistent estimator of the standard deviation σ. In the normal distribution scenario, the k value can be calculated as follows:

[0165]

[0166] Where Φ -1 (·) is the inverse function of the standard normal distribution.

[0167] If the reliable life of a certain lower layer component t R,i Satisfies formula (5) and the value is less than the median median(t R ), then the data can be considered as a very small outlier, and the low-level composition belongs to an extremely short-lived product.

[0168] Step 203: Similar product construction

[0169] Referring to the Bootstrap sampling method, the lower-level components are sampled without replacement in proportion to form part of the components of the similar product, and the similar product of the whole machine is constructed based on the series model. If a certain type of whole machine has N lower-level components, numbered i = 1, 2, …, N, and M (M < N) of them are extremely short-lived critical lower-level components, corresponding to the numbers i = 1, 2, …, M, then the process of sampling and constructing a similar product in proportion A can be carried out according to the following steps:

[0170] Step 1: Directly use the extremely short-lived lower-level components of the target whole machine as part of the lower-level components of the constructed product, and record the serial numbers in the constructed similar product as j = 1, 2, …, M;

[0171] Step 2: Without replacement, draw [NA] - M lower-level components from all N lower-level components of the target whole machine, where [·] represents rounding down, and record the serial numbers in the constructed similar product as j = M + 1, M + 2, …, [NA];

[0172] Step 3: With replacement, draw N - [NA] lower-level components from all N lower-level components of the target whole machine, and record the serial numbers in the constructed similar product as j = [NA] + 1, [NA] + 2, …, N;

[0173] According to the above three steps, a similar product of the target whole machine can be obtained. Since a proportion A of the lower-level components is exactly the same as the target whole machine, it can be considered that the similarity degree of the constructed product to the target whole machine is greater than or equal to A. For example, if the proportion A is set to 80%, it can be considered that the constructed product is at least 80% similar to the target product.

[0174] Repeat the above construction process multiple times to obtain multiple similar products of the target whole machine. Using these sampled and constructed products as inputs, further extract the prior information of the life parameters of the target whole machine.

[0175] Step 204, Estimation of the life distribution parameters of the similar product

[0176] Use the method of linear fitting to estimate the life distribution parameters. If the life of the whole machine product follows the Weibull distribution, the following linear relationship can be obtained according to the form of the Weibull distribution:

[0177] ln[-ln(1 - F(t))] = ln[-ln(R(t))] = m lnt - m lnη (7)

[0178] In the formula, F(t) is the cumulative failure probability (or unreliability) of the product from 0 to time t, R(t) is the reliability, and m and η are the shape parameter and scale parameter of the Weibull distribution respectively.

[0179] Assume that the reliability of the kth similar product sampled and constructed is R Zk(t), k = 1, 2, ..., K, the reliability function, failure probability density function, and failure rate function of the lower layer components are denoted as R ki (t), f ki (t), λ ki (t), i = 1, 2, ..., N, then according to the above formula (7) and the series model, the shape parameter m of the similar product can be obtained k satisfy:

[0180]

[0181] Where -f(t) / R(t) is the calculation formula for failure rate, so the shape parameter m k It can also be expressed by the following formula:

[0182]

[0183] In order to obtain the shape parameter m k Based on , the scale parameter can be obtained by calculating the reliability at time t = 1:

[0184]

[0185] It can be seen from formula (9) that the shape parameter m is a function related to time t. After a large number of numerical simulations, it is found that for products composed of multiple low-level components with different distributions connected in series, the linear relationship shown in the curve shown in formula (7) is relatively strong, and can generally be directly expressed by a general linear model or a two-stage linear model. Among them, the general form of the linear model and the two-stage linear model is shown in the following formula:

[0186]

[0187] The two-stage linear situation may be caused by mixed distribution. In this case, it is feasible to select the parameter estimation result of the first stage. To this end, the following suggestions are made when using the method proposed by the present invention:

[0188] ① When linear fitting is performed on the constructed similar products using formula (7), if the general linear model has a high goodness of fit (>0.95), the general linear model parameter fitting results are used to estimate the life distribution parameters;

[0189] ② When using formula (7) to perform linear fitting on the constructed similar products, if the goodness of fit of the linear model in the first stage is not high enough (≤0.95), the two-stage linear model is used to fit the sample values, and the parameter fitting results of the first stage (x≤c) are selected for the life distribution parameter estimation of similar products.

[0190] When fitting a two-stage linear model, the stage dividing point c can be determined by referring to the following formula. The dividing point represented by this formula is the shape parameter m k The moment of greatest change.

[0191]

[0192] After determining the data set for linear fitting, use equations (7) and (10) to fit the constructed similar products and obtain the shape parameter m k and the scale parameter η k The estimated results.

[0193] Step 205: Determine the prior distribution of the whole machine life distribution parameter

[0194] By statistically analyzing the estimated results of the life distribution parameters of similar products constructed, we can obtain the prior information of the life distribution parameters of the target whole product. If K similar products are constructed and their shape parameters m are estimated respectively k With the scale parameter η k , k=1,2,…,K, then directly perform statistical analysis on the data set m and η, and select a distribution with better fitting effect as the prior distribution of the parameter. In this process, in order to avoid the influence of data correlation on the assumption of parameter independence, half of the data can be taken for fitting separately when determining the prior distribution of shape parameters and scale parameters, that is, the shape parameter can be taken from the data set m k ,k=1,2,…,K / 2, the scale parameter can be taken as the data set η k ,k=K / 2+1,K / 2+2,…,K, respectively. Thus, the prior distributions of the target machine shape parameter m and scale parameter η are π (1) (m) and π (1) (η), complete the extraction of prior information of the life distribution parameters of the whole machine.

[0195] At present, the selection of prior distribution type is mainly based on the conjugate principle. Its advantage is that the prior distribution of the parameter can be consistent with the type of the posterior distribution, and thus the Bayes posterior analysis of the parameter is more convenient. However, the selection of conjugation is mainly based on the convenience of the calculation process, and it is often difficult to give a clearer explanation for the confirmation of prior information from the perspective of physical mechanism or sample space. In addition, it is difficult to determine the conjugate prior of the parameters of most distributions. For example, the shape parameter of the Weibull distribution does not have an optional conjugate distribution, which makes the method difficult to apply in limited scenarios. Moreover, with the continuous development of computer technology, the MCMC simulation method is fully capable of analyzing complex posterior distributions, which leads to the gradual weakening of the advantages of the conjugate prior.

[0196] The Jeffreys prior distribution is an optional solution for parameter prior distribution in an information-free scenario, but its form is complex and difficult to derive, making it difficult to meet the needs of engineering applications.

[0197] Through the above embodiments, in the process of determining the prior distribution of the life distribution parameters of the target whole machine, the prior information extraction process does not rely on subjective factors such as expert experience, but only relies on the low-level component product information that actually constitutes the target whole machine. Therefore, the proposed method can realize the extraction of prior information of newly developed products that lack historical information, and use the Bayes formula to complete the subsequent reliability evaluation work. Low-level product information is usually rich, so the proposed method has strong feasibility and engineering practice value.

[0198] It should be noted that the method of the embodiment of the present disclosure can be performed by a single device, such as a computer or a server. The method of the present embodiment can also be applied in a distributed scenario and completed by multiple devices cooperating with each other. In the case of such a distributed scenario, one of the multiple devices can only perform one or more steps in the method of the embodiment of the present disclosure, and the multiple devices will interact with each other to complete the described method.

[0199] It should be noted that the above describes some embodiments of the present disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the above embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0200] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, the present disclosure also provides a device for extracting prior information of the life of a whole machine.

[0201] refer to Figure 7 The device for extracting prior information of the whole machine life includes:

[0202] The reliable life determination module 301 is configured to obtain a preset reliability and determine the reliable life of multiple component products in the target whole machine under the preset reliability;

[0203] The short-life component product determination module 302 is configured to determine the median absolute deviation of the reliable life of the plurality of component products, and determine the short-life component products from the plurality of component products whose reliable life and the median absolute deviation satisfy a preset short-life relationship condition;

[0204] A construction module 303 is configured to construct a plurality of similar complete machines similar to the target complete machine based on the short-life component product;

[0205] A parameter determination module 304 is configured to determine a shape parameter and a scale parameter of each similar whole machine in the plurality of similar whole machines;

[0206] The evaluation module 305 is configured to determine the life parameter prior information of the target whole machine according to the shape parameter and the scale parameter through linear fitting, and perform a storage period evaluation on the target whole machine according to the life parameter prior information.

[0207] In some embodiments, the reliable life determination module 301 includes:

[0208] A preset reliability obtaining unit, configured to obtain a preset preset reliability;

[0209] a life distribution type determination unit configured to determine, for each component product of the plurality of component products, a life distribution type of the component product;

[0210] The first reliable life determination unit is configured to determine the reliable life of the component product at the preset reliability based on the exponential distribution type in response to determining that the life distribution type of the component product is an exponential distribution type,

[0211]

[0212] Among them, t R is the reliable life of the component product, λ is the failure rate of exponential distribution, and R is the preset reliability;

[0213] The second reliable life determination unit is configured to determine the reliable life of the component product at the preset reliability based on the Weibull distribution type in response to determining that the life distribution type of the component product is a Weibull distribution type,

[0214]

[0215] Among them, t R is the reliable life of the component product, η is the scale parameter of the Weibull distribution, m is the shape parameter of the Weibull distribution, and R is the preset reliability;

[0216] The third reliable life determination unit is configured to determine the reliable life of the component product at the preset reliability based on the lognormal distribution type in response to determining that the life distribution type of the component product is a lognormal distribution type,

[0217]

[0218] Among them, t R is the reliable life of the component product, Φ -1 (·) is the inverse function of the standard normal distribution, μ is the logarithmic mean of the lognormal distribution, σ is the logarithmic variance of the lognormal distribution, and R is the preset reliability.

[0219] In some embodiments, the short-life component product determination module 302 includes:

[0220] a median absolute deviation determining unit configured to determine a median life of the plurality of component products, and determine a median absolute deviation based on the reliable life of each component product and the median life of the plurality of component products,

[0221] MAD=median(|t R,i -median(t R )|)

[0222] Wherein, MAD is the median absolute deviation, t R,i is the reliable life of the i-th component product, median(·) is the median, median(t R ) is the median life span of the plurality of component products;

[0223] a short-life component product determination unit, configured to determine that the i-th component product is a short-life component product in response to determining that the reliable life of the i-th component product and the median absolute deviation satisfy a preset short-life relationship condition, and the reliable life of the i-th component product is less than the median life,

[0224] |t R,i -median(t R )|>3k·MAD

[0225] Among them, |t R,i -median(t R )|>3k MAD is the condition that the reliable life of the i-th component product and the median absolute deviation meet the preset short-life relationship, t R,i is the reliable life of the i-th component product, median(t R ) is the median of the reliable life of the multiple component products, MAD is the median absolute deviation, and k is the proportional factor constant.

[0226] In some embodiments, the construction module 303 includes:

[0227] A component quantity determination unit, configured to determine the total quantity of all component products in the target whole machine and the first quantity of the short-life component products;

[0228] a similarity degree obtaining unit, configured to obtain a preset similarity degree between the target whole machine and the similar whole machine;

[0229] a component quantity determination unit configured to determine a second quantity of the same component products based on the total quantity, the similarity level and the first quantity, and to determine a third quantity of the different component products based on the total quantity and the similarity level;

[0230] The similar whole machine construction unit is configured to select the first number of short-lived component products, select the second number of identical component products by non-replacement sampling, select the third number of different component products by replacement sampling, construct a similar whole machine similar to the target whole machine, and obtain multiple similar whole machines by repeating the construction process of the similar whole machine.

[0231] In some embodiments, the parameter determination module 304 includes:

[0232] a shape parameter determining unit, configured to determine a shape parameter of each of the plurality of similar whole machines,

[0233]

[0234] Among them, m k is the shape parameter of the kth similar machine, t is the current time, λ ki (t) is the failure rate function of the i-th component product in the k-th similar whole machine, R ki (t) is the failure rate function of the i-th component product in the k-th similar whole machine, and N is the number of component products in the k-th similar whole machine;

[0235] a scale parameter determining unit, configured to determine a scale parameter of each similar whole machine in the plurality of similar whole machines,

[0236]

[0237] Among them, η k is the scale parameter of the kth similar machine, R ki (1) is the failure rate function of the i-th component product in the k-th similar whole machine at time t=1.

[0238] In some embodiments, the evaluation module 305 includes:

[0239] A linear fitting model determination unit is configured to determine the degree of fit between shape parameters and scale parameters of a plurality of similar whole machines, and determine a linear fitting model according to the degree of fit;

[0240] A life parameter determination unit is configured to perform fitting processing between shape parameters and scale parameters of a plurality of similar complete machines based on the linear fitting model to obtain life parameters of the plurality of similar complete machines;

[0241] The prior information determining unit is configured to use the life parameters of the plurality of similar whole machines as the life parameter prior information of the target whole machine.

[0242] In some embodiments, the evaluation module 305 includes:

[0243] a posterior information determination unit, configured to determine the life parameter posterior information according to the life parameter prior information by using a Bayesian algorithm,

[0244]

[0245] Among them, π (2) (θ|x) is the posterior information of the life parameter, L(x|θ) is the likelihood function of θ given the test data x, π (1) (θ) is the prior information of the lifespan parameter, ∫L(x|θ)π (1) (θ)dθ is a constant.

[0246] For the convenience of description, the above device is described by dividing it into various modules according to its functions. Of course, when implementing the present disclosure, the functions of each module can be implemented in the same or multiple software and / or hardware.

[0247] The device of the above embodiment is used to implement the corresponding method for extracting prior information of the whole machine life in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0248] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, the present disclosure also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for extracting prior information of the whole machine life described in any of the above embodiments is implemented.

[0249] Figure 8 A more specific schematic diagram of the hardware structure of an electronic device provided in this embodiment is shown, and the device may include: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040 are connected to each other through the bus 1050 in the device.

[0250] The processor 1010 can be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0251] The memory 1020 may be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 may store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program codes are stored in the memory 1020 and are called and executed by the processor 1010.

[0252] The input / output interface 1030 is used to connect the input / output module to realize information input and output. The input / output module can be configured in the device as a component (not shown in the figure), or it can be externally connected to the device to provide corresponding functions. The input device may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output device may include a display, a speaker, a vibrator, an indicator light, etc.

[0253] The communication interface 1040 is used to connect a communication module (not shown in the figure) to realize communication interaction between the device and other devices. The communication module can realize communication through a wired mode (such as USB (Universal Serial Bus), network cable, etc.), or through a wireless mode (such as mobile network, WIFI (Wireless Fidelity), Bluetooth, etc.).

[0254] The bus 1050 includes a path that transmits information between the various components of the device (eg, the processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040).

[0255] It should be noted that, although the above device only shows the processor 1010, the memory 1020, the input / output interface 1030, the communication interface 1040 and the bus 1050, in the specific implementation process, the device may also include other components necessary for normal operation. In addition, it can be understood by those skilled in the art that the above device may also only include the components necessary for implementing the embodiments of the present specification, and does not necessarily include all the components shown in the figure.

[0256] The electronic device of the above embodiment is used to implement the corresponding method for extracting prior information of the whole machine life in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0257] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present disclosure also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the method for extracting prior information of the whole machine life as described in any of the above embodiments.

[0258] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0259] The computer instructions stored in the storage medium of the above embodiment are used to enable the computer to execute the method for extracting prior information of the whole machine life as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0260] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present disclosure is limited to these examples. Based on the concept of the present disclosure, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present disclosure as described above, which are not provided in detail for the sake of simplicity.

[0261] In addition, to simplify the description and discussion, and in order not to make the embodiments of the present disclosure difficult to understand, the known power / ground connections to the integrated circuit (IC) chips and other components may or may not be shown in the provided figures. In addition, the device can be shown in the form of a block diagram to avoid making the embodiments of the present disclosure difficult to understand, and this also takes into account the fact that the details of the implementation of these block diagram devices are highly dependent on the platform on which the embodiments of the present disclosure will be implemented (that is, these details should be fully within the scope of understanding of those skilled in the art). Where specific details (e.g., circuits) are set forth to describe exemplary embodiments of the present disclosure, it is apparent to those skilled in the art that the embodiments of the present disclosure can be implemented without these specific details or with changes in these specific details. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0262] Although the present disclosure has been described in conjunction with specific embodiments of the present disclosure, many replacements, modifications and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may use the embodiments discussed.

[0263] The embodiments of the present disclosure are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the present disclosure. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present disclosure should be included in the scope of protection of the present disclosure.

Claims

1. A method for extracting prior information of whole machine life, characterized in that: The method comprises: Obtaining a preset reliability and determining the reliable life of multiple component products in the target whole machine under the preset reliability; Determine the median absolute deviation of the reliable life of the plurality of component products, and determine, from the plurality of component products, a short-life component product whose reliable life and the median absolute deviation satisfy a preset short-life relationship condition; constructing a plurality of similar complete machines similar to the target complete machine based on the short-life component product; Determining shape parameters and scale parameters of each similar whole machine among the plurality of similar whole machines; The life parameter prior information of the target whole machine is determined according to the shape parameter and the scale parameter through linear fitting, and the storage period of the target whole machine is evaluated according to the life parameter prior information.

2. The method according to claim 1, characterized in that: The step of obtaining a preset reliability and determining the reliable life of multiple component products in the target whole machine under the preset reliability includes: obtaining a predetermined preset reliability; For each component product among the plurality of component products, determining a life distribution type of the component product; In response to determining that the life distribution type of the component product is an exponential distribution type, based on the exponential distribution type, determining a reliable life of the component product at the preset reliability, Among them, t R is the reliable life of the component product, λ is the failure rate of exponential distribution, and R is the preset reliability; In response to determining that the life distribution type of the component product is a Weibull distribution type, based on the Weibull distribution type, determining a reliable life of the component product at the preset reliability, Among them, t R is the reliable life of the component product, η is the scale parameter of the Weibull distribution, m is the shape parameter of the Weibull distribution, and R is the preset reliability; In response to determining that the life distribution type of the component product is a log-normal distribution type, based on the log-normal distribution type, determining a reliable life of the component product at the preset reliability, Among them, t R is the reliable life of the component product, Φ -1 (·) is the inverse function of the standard normal distribution, μ is the logarithmic mean of the lognormal distribution, σ is the logarithmic variance of the lognormal distribution, and R is the preset reliability.

3. The method according to claim 1, characterized in that The step of determining the median absolute deviation of the reliable lifespans of the plurality of component products, and determining, from the plurality of component products, a short-life component product whose reliable lifespan and the median absolute deviation satisfy a preset short-life relationship condition, comprises: determining a median life span of the plurality of component products, and determining a median absolute deviation based on the reliable life span of each component product and the median life span of the plurality of component products, MAD=median(|t R,i -median(t R )|) Wherein, MAD is the median absolute deviation, t R,i is the reliable life of the i-th component product, median(·) is the median, median(t R ) is the median life span of the plurality of component products; In response to determining that the reliable life of the i-th component product and the median absolute deviation satisfy a preset short-life relationship condition, and the reliable life of the i-th component product is less than the median life, determining that the i-th component product is a short-life component product, |t R,i -median(t R )|>3k·MAD Among them, |t R,i -median(t R )|>3k MAD is the condition that the reliable life of the i-th component product and the median absolute deviation meet the preset short-life relationship, t R,i is the reliable life of the i-th component product, median(t R ) is the median of the reliable life of the multiple component products, MAD is the median absolute deviation, and k is the proportional factor constant.

4. The method according to claim 1, characterized in that The step of constructing a plurality of similar complete machines similar to the target complete machine based on the short-life component product comprises: Determine the total quantity of all component products in the target whole machine and the first quantity of the short-life component products; Obtaining a preset similarity degree between the target whole machine and the similar whole machine; determining a second quantity of the same component products based on the total quantity, the degree of similarity, and the first quantity, and determining a third quantity of the different component products based on the total quantity and the degree of similarity; Select the first quantity of short-lived component products, select the second quantity of the same component products by sampling without replacement, select the third quantity of different component products by sampling with replacement, construct a similar whole machine similar to the target whole machine, and obtain multiple similar whole machines by repeating the construction process of the similar whole machine.

5. The method according to claim 1, characterized in that The determining of the shape parameters and scale parameters of each similar whole machine among the plurality of similar whole machines comprises: determining a shape parameter of each similar whole machine among the plurality of similar whole machines, Among them, m k is the shape parameter of the kth similar machine, t is the current time, λ ki (t) is the failure rate function of the i-th component product in the k-th similar whole machine, R ki (t) is the failure rate function of the i-th component product in the k-th similar whole machine, and N is the number of component products in the k-th similar whole machine; determining a dimension parameter of each similar whole machine in the plurality of similar whole machines, Among them, η k is the scale parameter of the kth similar machine, R ki (1) is the failure rate function of the i-th component product in the k-th similar whole machine at time t=1.

6. The method according to claim 1, characterized in that The determining the life parameter prior information of the target whole machine according to the shape parameter and the scale parameter by linear fitting includes: Determining the degree of fit between shape parameters and scale parameters of a plurality of similar complete machines, and determining a linear fitting model according to the degree of fit; Based on the linear fitting model, shape parameters and scale parameters of a plurality of similar complete machines are fitted to obtain life parameters of the plurality of similar complete machines; The life parameters of the plurality of similar machines are used as prior information on the life parameters of the target machine.

7. The method according to claim 1, characterized in that The performing storage period evaluation on the target whole machine according to the life parameter prior information includes: By using a Bayesian algorithm, the life parameter posterior information is determined according to the life parameter prior information. Among them, π (2) (θ|x) is the posterior information of the life parameter, L(x|θ) is the likelihood function of θ given the test data x, π (1) (θ) is the prior information of the lifespan parameter, ∫L(x|θ)π (1) (θ)dθ is a constant.

8. A device for extracting prior information of whole machine life, characterized in that: include: A reliable life determination module is configured to obtain a preset reliability and determine the reliable life of multiple component products in the target whole machine under the preset reliability; a short-life component product determination module, configured to determine the median absolute deviation of the reliable lifespans of the plurality of component products, and determine, from the plurality of component products, a short-life component product whose reliable lifespan and the median absolute deviation satisfy a preset short-life relationship condition; A construction module, configured to construct a plurality of similar complete machines similar to the target complete machine based on the short-life component product; A parameter determination module, configured to determine a shape parameter and a scale parameter of each similar whole machine in the plurality of similar whole machines; The evaluation module is configured to determine the life parameter prior information of the target whole machine according to the shape parameter and the scale parameter through linear fitting, and perform a storage period evaluation on the target whole machine according to the life parameter prior information.

9. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and running on the processor, wherein when the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

10. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to cause a computer to execute the method according to any one of claims 1 to 7.

Images