Fault rate prediction method and device, equipment and medium

Abstract

The invention provides a fault rate prediction method and device, equipment and a medium. The method comprises the following steps: acquiring factor data of each historical product, a statistical value of an associated use condition, and factor data of a to-be-predicted product; wherein the factor data is used for representing the characteristics of the product; based on the factor data of the to-be-predicted product and the factor data of each historical product, determining a reference product of the to-be-predicted product in each historical product; and determining the failure rate of the to-be-predicted product based on the factor data of the reference product, the statistical value of the use condition and the factor data of the to-be-predicted product. Through the method, the accuracy of fault rate prediction can be improved.

Description

Technical Field

[0001] The present disclosure relates to the field of information processing technology, and in particular to a failure rate prediction method, device, equipment, and medium. Background Art

[0002] The level of failure rate reflects the reliability of product design, which is related to the user experience and brand reputation. Therefore, in the field of consumer electronics, failure rate prediction is one of the important topics in product design and development. Failure rate prediction mainly includes expert experience evaluation method and model-based prediction method. The former is a subjective prediction made by domain experts based on experience, while the latter is based on historical data training statistics or machine learning models for prediction.

[0003] Currently in practice, the expert experience evaluation method has strong interpretability but low accuracy, while the model-based prediction method has high accuracy but low interpretability. Summary of the invention

[0004] The present invention provides a method, device, equipment and medium for predicting a failure rate.

[0005] According to a first aspect of an embodiment of the present disclosure, a failure rate prediction method is provided, comprising:

[0006] Obtaining the factor data of each historical product and the statistical value of the associated usage, as well as the factor data of the product to be predicted; wherein the factor data is used to characterize the characteristics of the product;

[0007] Based on the factor data of the product to be predicted and the factor data of each historical product, determining a reference product for the product to be predicted from each historical product;

[0008] The failure rate of the product to be predicted is determined based on the factor data and usage statistics of the reference product and the factor data of the product to be predicted.

[0009] In some embodiments, the method further comprises:

[0010] Determine a target factor based on the factor data of each historical product and the statistical value of the associated usage; wherein the correlation between the target factor and the failure of each historical product is greater than the correlation between factors other than the target factor and the failure of each historical product;

[0011] The step of determining a reference product for the product to be predicted from among the historical products based on the factor data of the product to be predicted and the factor data of each historical product includes:

[0012] Determine the difference between the product to be predicted and each historical product based on the data of the target factor in the factor data of the product to be predicted and each historical product;

[0013] The historical products whose differences between the product to be predicted and each historical product satisfy a preset difference condition are determined as reference products for the product to be predicted.

[0014] In some embodiments, the usage statistics include: the number of historical usages and the number of historical usage failures within a time period, and determining the target factor based on the factor data of each historical product and the statistics of the associated usage includes:

[0015] For each historical product, determine the failure frequency of the historical product based on the historical usage quantity and the historical usage failure quantity of the historical product in one or more time periods;

[0016] Based on the factor data and failure frequency of each historical product, a causal relationship graph is constructed; wherein in the causal relationship graph, there is a directed edge between the factor causing the frequency of each historical product failure to meet a preset frequency threshold and the historical product failure;

[0017] The factor pointed to by the directed edge in the causal relationship graph is used as the target factor.

[0018] In some embodiments, the failure frequency of the historical product satisfies the formula:

[0019]

[0020] Among them, λ i represents the failure frequency of the i-th historical product; C it A represents the number of historical failures of the i-th historical product in the t-th time period; it A represents the historical usage quantity of the i-th historical product in the t-th time period, i0 =0.

[0021] In some embodiments, determining the failure rate of the product to be predicted based on the factor data and usage statistics of the reference product and the factor data of the product to be predicted includes:

[0022] Based on the historical usage quantity of the reference product within the preset time period, estimating the usage quantity of the product to be predicted within the preset time period;

[0023] Based on the factor data of the product to be predicted and the factor data of the reference product, as well as the usage quantity of the product to be predicted in the preset time period and the failure frequency of the reference product, estimate the failure quantity of the product to be predicted in the preset time period;

[0024] Based on the estimated number of failures and the number of uses of the product to be predicted within the preset time period, the failure rate of the product to be predicted within the preset time period is determined.

[0025] In some embodiments, estimating the number of failures of the product to be predicted within the preset time period based on the factor data of the product to be predicted and the factor data of the reference product, the usage number of the product to be predicted within the preset time period, and the failure frequency of the reference product includes:

[0026] For each factor, determining the factor data difference between the factor data of the product to be predicted and the factor data of the reference product;

[0027] Determine the failure frequency of the product to be predicted based on the difference in each factor data between the product to be predicted and the reference product and the failure frequency of the reference product;

[0028] Based on the failure frequency of the product to be predicted and the usage quantity of the product to be predicted within the preset time period, the failure quantity of the product to be predicted within the preset time period is estimated.

[0029] In some embodiments, determining the failure frequency of the product to be predicted based on the difference in each factor data between the product to be predicted and the reference product and the failure frequency of the reference product includes:

[0030] Using a mapping model between factor data differences and failure frequency differences, each factor data difference between the product to be predicted and the reference product is processed to obtain a failure frequency difference value of the product to be predicted relative to the reference product; wherein the mapping model is obtained based on the factor data and failure frequency of each historical product;

[0031] The failure frequency of the product to be predicted is determined based on the failure frequency of the reference product and the difference value of the failure frequency of the predicted product relative to the reference product.

[0032] In some embodiments, the method further comprises:

[0033] Determine the historical reference product of each historical product among the historical products based on the factor data of each historical product and the statistical value of the associated usage;

[0034] Determine the difference between the failure frequency of each historical product and the failure frequency of the corresponding historical reference product;

[0035] For each factor, determine the historical factor data difference value between the factor data of the historical product and the factor data of the corresponding historical reference product;

[0036] Based on the difference value of the failure frequency of each historical product relative to the corresponding historical reference product and the difference value of the historical factor data corresponding to each factor, fitting is performed to construct a mapping between the difference value of the factor data corresponding to the factor and the difference value of the failure frequency to obtain the mapping model.

[0037] In some embodiments, the fitting is performed based on the difference value of the failure frequency of each historical product relative to the corresponding historical reference product, and the difference value of the historical factor data corresponding to each factor, and a mapping between the difference value of the factor data corresponding to the factor and the difference value of the failure frequency is constructed to obtain the mapping model, including:

[0038] Based on the difference value of the failure frequency of each historical product relative to the corresponding historical reference product, and the difference value of the historical factor data corresponding to the target factor, a first preset function is used for fitting to determine the fitting parameters of the first preset function;

[0039] Determine a fitting difference value of the failure frequency of each historical product relative to the corresponding historical reference product based on the historical factor data difference of each historical product and the first preset function after parameter adjustment;

[0040] Based on the residual between the fitting difference value of the fault frequency corresponding to each historical product and the difference value of the fault frequency, and the factor data corresponding to the factors other than the target factor, the second preset function is used for fitting, and the fitting parameters of the first preset function are updated to obtain the mapping model.

[0041] In some embodiments, if the reference product includes multiple products, determining the failure frequency of the product to be predicted based on the failure frequency of the reference product and the difference value of the failure frequency of the predicted product relative to the reference product includes:

[0042] For each reference product, determining the sum of the failure frequency difference value of the predicted product relative to the reference product and the failure frequency of the reference product;

[0043] The average value of the sum values ​​corresponding to all reference products is determined as the failure frequency of the product to be predicted.

[0044] In some embodiments, the number of failures of the product to be predicted within the preset time period satisfies the following formula:

[0045]

[0046] Among them, A n+1,t is the usage quantity of the product to be predicted in the tth time period, is the failure frequency of the product to be predicted, A n+1,0 =0.

[0047] In some embodiments, if the reference product includes multiple products, estimating the usage quantity of the product to be predicted within the preset time period based on the historical usage quantity of the reference product within the preset time period includes:

[0048] The average of the historical usage quantities of each reference product within the preset time period is determined as the usage quantity of the product to be predicted within the preset time period.

[0049] According to a second aspect of an embodiment of the present disclosure, there is provided a failure rate prediction device, comprising:

[0050] An acquisition module configured to acquire factor data of each historical product and statistics of associated usage conditions, as well as factor data of the product to be predicted; wherein the factor data is used to characterize the characteristics of the product;

[0051] A first determination module is configured to determine a reference product of the product to be predicted from each historical product based on the factor data of the product to be predicted and the factor data of each historical product;

[0052] The second determination module is configured to determine the failure rate of the product to be predicted based on the factor data and usage statistics of the reference product and the factor data of the product to be predicted.

[0053] In some embodiments, the apparatus further comprises:

[0054] A third determination module is configured to determine a target factor based on the factor data of each historical product and the statistical value of the associated usage; wherein the correlation between the target factor and the failure of each historical product is greater than the correlation between factors other than the target factor and the failure of each historical product;

[0055] The first determination module is further configured to determine the difference between the product to be predicted and each historical product based on the data of the target factor in the factor data of the product to be predicted and each historical product; and determine the historical product whose difference between the product to be predicted and each historical product meets the preset difference condition as the reference product of the product to be predicted.

[0056] In some embodiments, the usage statistics include: the historical usage quantity and the historical usage failure quantity within a time period. The third determination module is further configured to determine, for each historical product, the failure frequency of the historical product based on the historical usage quantity and the historical usage failure quantity of the historical product within one or more time periods; construct a causal relationship graph based on the factor data and failure frequency of each historical product; wherein, in the causal relationship graph, there is a directed edge between the factor causing the frequency of each historical product failure to meet a preset frequency threshold and the historical product failure; and the factor pointed to by the directed edge in the causal relationship graph is used as the target factor.

[0057] In some embodiments, the failure frequency of the historical product satisfies the formula:

[0058]

[0059] Among them, λ i represents the failure frequency of the i-th historical product; C it A represents the number of historical failures of the i-th historical product in the t-th time period; it A represents the historical usage quantity of the i-th historical product in the t-th time period, i0 =0.

[0060] In some embodiments, the second determination module is further configured to estimate the usage quantity of the product to be predicted within the preset time period based on the historical usage quantity of the reference product within the preset time period; estimate the number of failures of the product to be predicted within the preset time period based on the factor data of the product to be predicted and the factor data of the reference product, as well as the usage quantity of the product to be predicted within the preset time period and the failure frequency of the reference product; and determine the failure rate of the product to be predicted within the preset time period based on the estimated number of failures and usage quantity of the product to be predicted within the preset time period.

[0061] In some embodiments, the second determination module is further configured to determine, for each factor, the factor data difference between the factor data of the product to be predicted and the factor data of the reference product; determine the failure frequency of the product to be predicted based on the difference in each factor data between the product to be predicted and the reference product and the failure frequency of the reference product; and estimate the number of failures of the product to be predicted within the preset time period based on the failure frequency of the product to be predicted and the number of uses of the product to be predicted within the preset time period.

[0062] In some embodiments, the second determination module is further configured to process each factor data difference between the product to be predicted and the reference product using a mapping model between factor data differences and failure frequency differences to obtain a failure frequency difference value of the product to be predicted relative to the reference product; wherein the mapping model is obtained based on the factor data and failure frequency of each historical product; and the failure frequency of the product to be predicted is determined based on the failure frequency of the reference product and the failure frequency difference value of the predicted product relative to the reference product.

[0063] In some embodiments, the apparatus further comprises:

[0064] The fourth determination module is configured to determine the historical reference product of each historical product among all historical products based on the statistical values ​​of the factor data and associated usage of each historical product; determine the difference value between the failure frequency of each historical product and the failure frequency of the corresponding historical reference product; determine, for each factor, the difference value of the historical factor data between the factor data of the historical product and the factor data of the corresponding historical reference product; perform fitting based on the difference value of the failure frequency of each historical product relative to the corresponding historical reference product and the difference value of the historical factor data corresponding to each factor, construct a mapping between the difference value of the factor data corresponding to the factor and the difference value of the failure frequency, and obtain the mapping model.

[0065] In some embodiments, the fourth determination module is further configured to use the first preset function to perform fitting based on the difference value of the failure frequency of each historical product relative to the corresponding historical reference product, and the difference value of the historical factor data corresponding to the target factor, to determine the fitting parameters of the first preset function; determine the fitting difference value of the failure frequency of each historical product relative to the corresponding historical reference product based on the difference in historical factor data of each historical product and the first preset function after parameter adjustment; use the second preset function to perform fitting based on the residual between the fitting difference value of the failure frequency corresponding to each historical product and the difference value of the failure frequency, and the factor data corresponding to factors other than the target factor, to update the fitting parameters of the first preset function and obtain the mapping model.

[0066] In some embodiments, if the reference product includes multiple products, the second determination module is also configured to determine, for each reference product, the sum of the failure frequency difference value of the predicted product relative to the reference product and the failure frequency of the reference product; and determine the average value of the sum corresponding to all reference products as the failure frequency of the product to be predicted.

[0067] In some embodiments, the number of failures of the product to be predicted within the preset time period satisfies the following formula:

[0068]

[0069] Among them, A n+1,t is the usage quantity of the product to be predicted in the tth time period, is the failure frequency of the product to be predicted, A n+1,0 =0.

[0070] In some embodiments, if the reference product includes multiple products, the second determination module is further configured to determine the average of the historical usage quantities of each reference product within the preset time period as the usage quantity of the product to be predicted within the preset time period.

[0071] According to a third aspect of an embodiment of the present disclosure, there is provided an electronic device, including:

[0072] processor;

[0073] a memory for storing processor-executable instructions;

[0074] Wherein, the processor is configured to execute the method as described in the first aspect above.

[0075] According to a fourth aspect of an embodiment of the present disclosure, there is provided a storage medium, including:

[0076] When the instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the method as described in the first aspect above.

[0077] The technical solution provided by the embodiments of the present disclosure may have the following beneficial effects:

[0078] In the embodiments of the present disclosure, when predicting the failure rate of the product to be predicted, after selecting the reference product of the product to be predicted from various historical products, the prediction is made based on the reference product. Since the reference product is a product that is more similar to the product to be predicted, their target populations and differences in usage habits of the target populations may also be relatively similar. Therefore, making predictions based on the reference product rather than all historical products can make the predictions more comparable, reduce the impact of user-side factors on the prediction results, and improve the accuracy of the predictions. In addition, the method of making predictions based on the reference products in the historical products in the embodiments of the present disclosure is not only applicable to predictions after the product is launched, but also applicable to the selection of auxiliary product solutions in the design stage before the product is launched, and the solution has good universality.

[0079] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0081] Figure 1 It is a flow chart of a failure rate prediction method shown in an embodiment of the present disclosure.

[0082] Figure 2 It is an example diagram of a cause-effect relationship diagram in an embodiment of the present disclosure.

[0083] Figure 3 This is an example diagram of the principle of the failure rate prediction method in the embodiment of the present disclosure.

[0084] Figure 4 The figure is a diagram of a failure rate prediction device according to an exemplary embodiment.

[0085] Figure 5 It is a block diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0086] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0087] The terms "first / second / third" etc. involved in the present disclosure are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It can be understood that "first / second / third" etc. can be interchanged with a specific order or sequence where permitted, so that the embodiments of the present disclosure described herein can be implemented in an order other than that illustrated or described herein.

[0088] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present disclosure belongs. The terms used herein are only for the purpose of describing the present disclosure and are not intended to limit the present disclosure.

[0089] When predicting the failure rate, one solution is to establish a gray failure rate formula based on the gray model, and extract the failure rate characteristic values ​​of different mobile phone models based on the bathtub curve principle in the failure rate curve, and then evaluate the failure rate level and distribution of the new model in the complete life cycle through the historical model failure data. Specifically, according to the failure rate of historical models, the failure rate characteristic value is calculated in combination with the gray model, and the relationship between the failure rate characteristic value and the number of failures is established; for the new model to be predicted, the failure rate characteristic value of the new model is first determined based on the manufacturer, main component model and failure rate analysis, model positioning (flagship model, mid-range model, entry-level model, etc.), model function positioning (camera phone, music phone, elderly phone, etc.), and then the failure number prediction value is obtained based on the relationship between the failure rate characteristic value and the number of failures; finally, the predicted value of the after-sales failure rate is divided by the estimated sales volume in the sales plan according to the predicted value of the number of failures. Another solution is to predict the long-term after-sales failure rate based on short-term after-sales data after the product is launched.

[0090] Of the above two solutions, the first solution does not establish a quantitative model for the impact of factors such as the main component models and failure rate analysis, model positioning (flagship model, mid-range model, entry-level model, etc.), and model function positioning (camera phone, music phone, elderly phone, etc.) on the after-sales failure rate. The model lacks interpretability and cannot quantitatively describe the impact of various factors on the after-sales failure rate. In addition, if the new model is very different from the historical model, the prediction effect is relatively poor. The second solution cannot be applied to the design stage before the product is launched to assist in product solution selection.

[0091] In this regard, the present disclosure provides a method for predicting a failure rate, the execution subject of which can be an electronic device such as a server, wherein the server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. In some possible implementations, the method for predicting a failure rate can be implemented by a processor in an electronic device calling computer-readable instructions stored in a memory.

[0092] Figure 1 is a flow chart of a failure rate prediction method shown in an embodiment of the present disclosure, such as Figure 1 As shown, the failure rate prediction method includes the following steps:

[0093] S11, obtaining the factor data of each historical product and the statistical value of the associated usage, as well as the factor data of the product to be predicted; wherein the factor data is used to characterize the characteristics of the product;

[0094] S12, based on the factor data of the product to be predicted and the factor data of each historical product, determining a reference product for the product to be predicted from each historical product;

[0095] S13. Determine the failure rate of the product to be predicted based on the factor data and usage statistics of the reference product and the factor data of the product to be predicted.

[0096] In the embodiments of the present disclosure, the historical product and the predicted product can be the same type of product, such as hardware products such as mobile phones, laptops, tablet computers, digital cameras, etc., or software products such as application software. The prediction of the predicted product failure rate can be a prediction of software failure rate, or a prediction of hardware failure rate such as screen failure rate, speaker failure rate, camera failure rate, or whole machine hardware failure rate, which is not limited in the embodiments of the present disclosure.

[0097] In step S11, the electronic device obtains the factor data of each historical product. Taking the prediction of the failure rate of the mobile phone screen as an example, the corresponding factor data includes, for example, hardware design factors, hardware simulation factors or hardware reliability test factors, etc.; wherein, the design factors may include the size, material, and weight of the screen; the simulation factors may include, for example, the simulation of the hardness or rigidity of the mobile phone frame; the reliability test factor may include, for example, whether a protective case is used, and the corresponding reliability test factor data is whether the mobile phone screen is broken or water-infiltrated when the protective case is used. Taking the software failure rate test in the tablet computer as an example, the corresponding factor data includes, for example, software design factors, software simulation factors or software reliability test factors, etc.; wherein, the software design factors include, for example, the execution logic of the instructions, the layout design of the system interface, etc.; the software simulation factors include, for example, functional verification, regression testing, etc.; the reliability test factors include, for example, stress testing, concurrency testing, stability testing, etc. In the disclosed facility example, the factor data is the design data, simulation data or reliability test data corresponding to the above factors, etc., and the factor data is used to characterize the characteristics of the product software and / or hardware.

[0098] In the embodiment of the present disclosure, the statistical value of the usage of historical product associations includes, for example, the historical usage quantity, the historical usage failure quantity, the historical usage failure rate, etc. The historical usage failure rate is the ratio of the historical usage failure quantity to the historical usage quantity.

[0099] In step S12, the electronic device determines a reference product for the product to be predicted from each historical product based on the factor data of the product to be predicted and the factor data of each historical product, that is, selects a product that is more similar to the product to be predicted from each historical product. It should be noted that in the embodiment of the present disclosure, the factor data of the product to be predicted also includes data corresponding to the same factors as the historical products. Since the factor data represents the characteristics of the product, the electronic device can determine the reference product for the product to be predicted based on the factor data. The reference product can be a product with similar factor data to the product to be predicted, for example, at least one of the design factor data, simulation factor data, or reliability test factor data has similarity. In the embodiment of the present disclosure, it is assumed that the factor data of the product to be predicted is recorded as X n+1,k , where k = 1, 2,…, p, indicating the number of factors.

[0100] In step S13, after determining the reference product of the product to be predicted, the electronic device can determine the failure rate of the product to be predicted based on the factor data and usage statistics of the reference product and the factor data of the product to be predicted. The failure rate of the product to be predicted represents the probability of failure of the product to be predicted. After determining the failure rate of the product to be predicted, the embodiment of the present disclosure can assist in product design selection, market launch, benefit calculation, etc. based on the result.

[0101] It is understandable that in the implementation of the present disclosure, when predicting the failure rate of the product to be predicted, after selecting the reference product of the product to be predicted from various historical products, the prediction is made based on the reference product. Since the reference product is a product that is more similar to the product to be predicted, their target populations and differences in the usage habits of the target populations may also be relatively similar. Therefore, making predictions based on the reference product rather than all historical products can make the predictions more comparable, reduce the impact of user-side factors on the prediction results, and improve the accuracy of the predictions. In addition, the method of making predictions based on the reference products in the historical products in the embodiments of the present disclosure is not only applicable to predictions after the product is launched, but also applicable to the selection of auxiliary product solutions in the design stage before the product is launched, and the solution has good universality.

[0102] In some embodiments, the method further comprises:

[0103] Determine a target factor based on the factor data of each historical product and the statistical value of the associated usage; wherein the correlation between the target factor and the failure of each historical product is greater than the correlation between factors other than the target factor and the failure of each historical product;

[0104] The step of determining a reference product for the product to be predicted from among the historical products based on the factor data of the product to be predicted and the factor data of each historical product includes:

[0105] Determine the difference between the product to be predicted and each historical product based on the data of the target factor in the factor data of the product to be predicted and each historical product;

[0106] The historical products whose differences between the product to be predicted and each historical product satisfy a preset difference condition are determined as reference products for the product to be predicted.

[0107] When determining the target factor based on the factor data of each historical product and the statistical value of the associated usage, the disclosed embodiment may determine it based on, for example, the causal relationship or correlation between the factor data and the historical usage failure rate. When the electronic device determines the difference between the product to be predicted and each historical product based on the data of the target factor in the factor data of the product to be predicted and each historical product, it may calculate the correlation coefficient or the Euclidean distance, Manhattan distance, etc. based on the data of the target factor of the product to be predicted and each historical product, and the disclosed embodiment does not limit this. Taking the determination of the difference between the product to be predicted and each historical product based on the Euclidean distance as an example, the following formula (1) is the calculation formula for the Euclidean distance:

[0108]

[0109] Among them, x MB is the data of the target factor, X ik With X jk It can be represented as the data of two target factors to be compared. When there are multiple target factors, the Euclidean distance is determined based on the cumulative difference of the data of multiple target factors.

[0110] In the disclosed embodiment, the electronic device uses historical products whose differences between the product to be predicted and the historical products meet a preset difference condition as reference products, for example, historical products whose differences between the predicted product and the historical products are less than a preset difference threshold as reference products, or the first K historical products with the smallest differences between the predicted product and the historical products as reference products. In the disclosed embodiment, the number of reference products for the product to be predicted can be one or more.

[0111] In the disclosed embodiment, when the electronic device selects reference products of the product to be predicted from various historical products based on the factor data, it first determines the target factor, and then calculates the difference between the product to be predicted and each historical product based on the data of the target factor in the factor data of the product to be predicted and each historical product, so as to determine the reference product. Since the target factor is the factor with the greater correlation with the failure of each historical product among all factors, the electronic device determines the difference between the product to be predicted and each historical product based on the data of the target factor, that is, the reference product is selected according to the key factor data, which can reduce the interference of the factor data other than the target factor, thereby improving the accuracy of the reference product determination and further improving the accuracy of the failure rate prediction.

[0112] In some embodiments, the usage statistics include the number of historical usages and the number of historical usage failures within a time period, and determining the target factor based on the factor data of each historical product and the statistics of the associated usages includes:

[0113] For each historical product, determine the failure frequency of the historical product based on the historical usage quantity and the historical usage failure quantity of the historical product in one or more time periods;

[0114] Based on the factor data and failure frequency of each historical product, a causal relationship graph is constructed; wherein in the causal relationship graph, there is a directed edge between the factor causing the frequency of each historical product failure to meet a preset frequency threshold and the historical product failure;

[0115] The factor pointed to by the directed edge in the causal relationship graph is used as the target factor.

[0116] In the embodiment of the present disclosure, the statistical value of the historical product usage includes the historical usage quantity and the historical usage failure quantity within a time period, wherein the time period can be in months, years, weeks or days. The following Table 1 is an example table of the factor data and usage statistics of each historical product, wherein one model corresponds to one historical product, the cumulative usage quantity in the mth month is the historical usage quantity within the time period, and the cumulative failure quantity in the mth month is the historical failure quantity within the time period.

[0117]

[0118] In the disclosed embodiment, the electronic device determines the failure frequency of each historical product based on the historical usage number and the historical usage failure number in multiple time periods. For example, the ratio between the historical usage failure number and the historical usage number in multiple time periods can be determined as the failure frequency of the historical product. In addition, it can also be assumed that the change process of the number of hardware failures of each historical product over time obeys a time homogeneous Poisson process, and the Poisson intensity is calculated based on the Poisson distribution, and the Poisson intensity is the failure frequency of the historical product. In the disclosed embodiment, after determining the failure frequency of each historical product, the electronic device can construct a causal relationship graph based on the factor data and failure frequency of each historical product. The causal relationship graph is a directed acyclic graph, whose nodes represent factors or failure frequencies, directed edges represent direct causal relationships, and directed paths represent possible causal relationships. Since the factors connected by the directed edges in the causal relationship graph cause the frequency of each historical product failure to meet the preset frequency threshold, the factors pointed to by the directed edges belong to factors with greater correlation with the failures of each historical product.

[0119] Figure 2 is an example diagram of a cause-effect relationship diagram in an embodiment of the present disclosure, such as Figure 2 As shown in FIG. 1 , the direct cause of the screen failure is the weight of the whole machine, whether a protective case is used, and the screen material. In the embodiment of the present disclosure, the direct cause of the screen failure can be determined as the target factor. Figure 2 It can be seen that there is a directed edge between the direct cause of the screen failure and the screen failure.

[0120] It should be noted that, when constructing a causal relationship graph based on the factor data and failure frequency of each historical product, the embodiment of the present disclosure can be constructed through expert experience, or can be obtained by learning the factor data and failure frequency of each historical product through a causal structure learning algorithm. Among them, the causal structure learning algorithm is, for example, a PC algorithm, an FCI algorithm, or a NOTEARS algorithm, etc., which is not limited in the embodiment of the present disclosure.

[0121] It can be understood that in the embodiment of the present disclosure, the electronic device constructs a causal relationship graph based on factor data and fault frequency, and uses the factor pointed to by the directed edge in the causal relationship graph as the target factor. Since the causal relationship graph gives a qualitative causal relationship between the factor and the fault frequency, the embodiment of the present disclosure facilitates quantitative description of the impact of each factor on the fault frequency, and the target factor determined based on the causal relationship graph is also more accurate.

[0122] In some embodiments, the failure frequency of the historical product satisfies the formula:

[0123]

[0124] Among them, λ i represents the failure frequency of the i-th historical product; C it A represents the number of historical failures of the i-th historical product in the t-th time period; it A represents the historical usage quantity of the i-th historical product in the t-th time period, i0 =0.

[0125] In the disclosed embodiment, the historical usage number and the historical usage failure number of historical products in multiple time periods are processed based on the Poisson process. It is assumed that the change process of the usage failure number of each product over time obeys the time homogeneous Poisson process model, and each product is independent and identically distributed. According to the properties of the Poisson process, the cumulative failure number of a product at time t (i.e., the historical usage failure number in the time period) obeys a non-time homogeneous Poisson process, and its intensity at time t is proportional to the cumulative usage number at time t (i.e., the historical usage number in the time period). Therefore, based on the above Poisson process, the above formula can be derived using the historical usage number and the historical usage failure number in the time period.

[0126] In the embodiment of the present disclosure, since the Poisson distribution is suitable for describing the number of occurrences of random events per unit time, determining the failure frequency of historical products based on the Poisson process can more accurately determine the failure frequency.

[0127] In some embodiments, determining the failure rate of the product to be predicted based on the factor data and usage statistics of the reference product and the factor data of the product to be predicted includes:

[0128] Based on the historical usage quantity of the reference product within a preset time period, estimating the usage quantity of the product to be predicted within a preset time period;

[0129] Based on the factor data of the product to be predicted and the factor data of the reference product, as well as the usage quantity of the product to be predicted in the preset time period and the failure frequency of the reference product, estimate the failure quantity of the product to be predicted in the preset time period;

[0130] Based on the estimated number of failures and the number of uses of the product to be predicted within the preset time period, the failure rate of the product to be predicted within the preset time period is determined.

[0131] In the embodiment of the present disclosure, when the electronic device determines the failure rate of the product to be predicted based on the reference product, it estimates the usage quantity of the product to be predicted within the preset time period according to the historical usage quantity of the reference product within the preset time period. By way of example, when the reference product of the product to be predicted is unique, the historical usage quantity of the reference product in the tth month may be used as the estimated value of the usage quantity of the product to be predicted in the tth month.

[0132] In some embodiments, if the reference products of the product to be predicted include multiple products, estimating the usage quantity of the product to be predicted within the preset time period based on the historical usage quantity of the reference products within the preset time period includes:

[0133] The average of the historical usage quantities of each reference product within the preset time period is determined as the usage quantity of the product to be predicted within the preset time period.

[0134] In this embodiment, for example, the average of the historical usage quantities of multiple reference products in the tth month is used as the usage quantity of the product to be predicted in the tth month. It can be understood that using the average of multiple reference products in the same time period as the estimated value is more accurate. Of course, the embodiment of the present disclosure is not limited to the method of using the average value, for example, it can also be the maximum value or the median value.

[0135] In the disclosed embodiment, the electronic device also estimates the number of failures of the product to be predicted within a preset time period based on the factor data of the product to be predicted and the factor data of the reference product, as well as the number of uses of the product to be predicted within a preset time period and the failure frequency of the reference product. Since the factor data includes data of factors that have a certain correlation with product failures, such as data of target factors, and the failure frequency of the reference product represents the frequency of failures of the reference product within a time period, based on the above data, combined with the number of uses of the product to be predicted within a preset time period, the number of failures of the product to be predicted within a preset time period can be estimated.

[0136] In an embodiment of the present disclosure, the electronic device can determine the failure rate of the product to be predicted within the preset time period based on the estimated number of failures and the number of uses of the product to be predicted within the preset time period. For example, the ratio of the estimated number of failures within the preset time period to the estimated number of uses within the preset time period is determined as the failure rate of the product to be predicted within the preset time period.

[0137] It can be understood that in the embodiments of the present disclosure, when making predictions based on reference products, combining various data of the reference products, including historical usage statistics, factor data, and failure frequency estimates based on historical usage statistics, can improve the accuracy of the prediction of the failure rate of the product to be predicted.

[0138] In some embodiments, estimating the number of failures of the product to be predicted within the preset time period based on the factor data of the product to be predicted and the factor data of the reference product, the usage number of the product to be predicted within the preset time period, and the failure frequency of the reference product includes:

[0139] For each factor, determining the factor data difference between the factor data of the product to be predicted and the factor data of the reference product;

[0140] Determine the failure frequency of the product to be predicted based on the difference in each factor data between the product to be predicted and the reference product and the failure frequency of the reference product;

[0141] Based on the failure frequency of the product to be predicted and the usage quantity of the product to be predicted within the preset time period, the failure quantity of the product to be predicted within the preset time period is estimated.

[0142] In the disclosed embodiment, the electronic device calculates the difference in factor data corresponding to each factor of the product to be predicted and the reference product, and then determines the failure frequency of the product to be predicted based on the difference in factor data of each factor, so that the electronic device can estimate the number of failures of the product to be predicted within a preset time period based on the failure frequency of the product to be predicted and the number of uses of the product to be predicted within the preset time period.

[0143] In some embodiments, determining the failure frequency of the product to be predicted based on the difference in each factor data between the product to be predicted and the reference product and the failure frequency of the reference product includes:

[0144] Using a mapping model between factor data differences and failure frequency differences, each factor data difference between the product to be predicted and the reference product is processed to obtain a failure frequency difference value of the product to be predicted relative to the reference product; wherein the mapping model is obtained based on the factor data and failure frequency of each historical product;

[0145] The failure frequency of the product to be predicted is determined based on the failure frequency of the reference product and the difference value of the failure frequency of the predicted product relative to the reference product.

[0146] In the disclosed embodiment, the mapping model constructs a mapping between factor data differences and fault frequency differences, so that the electronic device can obtain a fault frequency difference value of the product to be predicted relative to the reference product based on each factor data difference between the product to be predicted and the reference product. Based on the obtained fault frequency difference value of the product to be predicted relative to the reference product, the electronic device can determine the fault frequency of the product to be predicted in combination with the fault frequency of the reference product.

[0147] It should be noted that, in the embodiment of the present disclosure, when using a mapping model between factor data differences and fault frequency differences to process each factor data difference between the product to be predicted and the reference product to obtain the fault frequency difference value of the product to be predicted relative to the reference product, it is assumed that the mapping model is a model constructed in multiple steps based on the factor data of the target factor and the factor data of the non-target factor. Therefore, when using the preset model for processing, it is necessary to specifically bring in the factor data differences of the target factor and the factor data differences of the non-target factor.

[0148] It can be understood that in the embodiments of the present disclosure, since the mapping model is obtained based on the factor data and failure frequency of historical products, and a mapping is established between the factor data difference and the failure frequency difference, the model can quantify the impact of the factor on the failure frequency. On this basis, since the electronic device further estimates the number of failures of the product to be predicted within a preset time period based on the failure frequency of the product to be predicted to obtain an estimate of the failure rate, the model can quantify the impact of the factor on the failure rate of the product to be predicted.

[0149] In some embodiments, when there is only one reference product, when the electronic device determines the failure frequency of the product to be predicted based on the failure frequency value of the reference product and the failure frequency difference value of the predicted product relative to the reference product, the sum of the failure frequency value of the reference product and the failure frequency difference value can be directly used as the failure frequency of the product to be predicted.

[0150] In some other embodiments, if the reference product includes multiple products, determining the failure frequency of the product to be predicted based on the failure frequency of the reference product and the difference value of the failure frequency of the predicted product relative to the reference product includes:

[0151] For each reference product, determining the sum of the failure frequency difference value of the predicted product relative to the reference product and the failure frequency of the reference product;

[0152] The average value of the sum values ​​corresponding to all reference products is determined as the failure frequency of the product to be predicted.

[0153] In the embodiment of the present disclosure, the subscript set of the reference product is denoted as P n+1 , j∈P n+1 , then the sum of the fault frequency difference value and the fault frequency can be determined based on the following formula (3):

[0154]

[0155] in, is the difference in the failure frequency of the predicted product relative to the jth reference product, λ j is the fault frequency value of the jth reference product, That is, the sum of the failure frequency difference value corresponding to the j-th reference product and the failure frequency of the j-th reference product.

[0156] After the sum is obtained based on the above formula, the failure frequency of the product to be predicted is determined by taking the average based on the following formula (4):

[0157]

[0158] In the embodiment of the present disclosure, as described above, the number of failures of the product to be predicted within the preset time period is estimated based on the failure frequency of the product to be predicted and the number of times the product to be predicted is used within the preset time period. n+1,t , then in some embodiments, the number of failures of the product to be predicted within the preset time period satisfies the following formula:

[0159]

[0160] Among them, A n+1,t is the usage quantity of the product to be predicted in the tth time period, is the failure frequency of the product to be predicted, A n+1,0 =0.

[0161] In the above formula, is the value estimated by linear interpolation of the usage quantity in the tth time period. In addition, according to the properties of the aforementioned Poisson process, the above formula is a formula obtained by multiplying the usage quantity and the failure frequency in the tth time period by the mean value of the failure quantity in the tth time period.

[0162] As mentioned above, the number of failures of the product to be predicted can be the ratio of the number of failures of the product to be predicted within a preset time period to the number of uses within the preset time period. Based on the above formula (5), the failure rate within the preset time period can be obtained as follows:

[0163]

[0164] Figure 3 FIG. 1 is an example diagram of the principle of the failure rate prediction method in the embodiment of the present disclosure. Figure 3 As shown, for the product to be predicted, based on the factor data including design factors, simulation factors and test factors that are the same as those of the historical products, the failure rate prediction value can be obtained based on the prediction model formula (6).

[0165] As mentioned above, in the embodiment of the present disclosure, the mapping model is obtained based on the factor data and failure frequency of each historical product. In some embodiments, the method further includes:

[0166] Determine the historical reference product of each historical product among the historical products based on the factor data of each historical product and the statistical value of the associated usage;

[0167] Determine the difference between the failure frequency of each historical product and the failure frequency of the corresponding historical reference product;

[0168] For each factor, determine the historical factor data difference value between the factor data of the historical product and the factor data of the corresponding historical reference product;

[0169] Based on the difference value of the failure frequency of each historical product relative to the corresponding historical reference product and the difference value of the historical factor data corresponding to each factor, fitting is performed to construct a mapping between the difference value of the factor data corresponding to the factor and the difference value of the failure frequency to obtain the mapping model.

[0170] In the disclosed embodiment, the electronic device determines the historical reference product of each historical product by adopting the aforementioned method of determining the reference product of the product to be predicted for each historical product. Exemplarily, after determining the target factor based on the factor data of each historical product and the statistical value of the associated usage, the electronic device may determine the difference between the current historical product and the other historical products based on the data of the target factor of the current historical product and the other historical products, and then determine the historical product whose difference meets the preset difference condition as the historical reference product of the current historical product. It should be noted that the preset difference condition corresponding to the historical reference product for the historical product may be the same as or different from the preset difference condition corresponding to the reference product for the product to be predicted.

[0171] After determining the historical reference product for each historical product, the electronic device further determines the difference between the failure frequency of each historical product and the failure frequency of the corresponding historical reference product, and for each factor, determines the historical factor data difference between the factor data of the historical product and the factor data of the corresponding historical reference product.

[0172] For example, assuming that the set of historical reference products of the i-th historical product is P i , assuming that i is the failure frequency of the i-th historical product, λ m is the failure frequency of the historical reference product of the i-th historical product, then Δλ im =λ i -λ m It represents the difference of fault frequency, where m∈P i .

[0173] For example, assuming that the set of historical reference products of the i-th historical product is P i , assuming X ih is the factor data of the hth factor of the ith historical product, X mh is the factor data of the hth factor of the historical reference product of the i-th historical product, then ΔX imh =X ih -X mh It represents the difference value of historical factor data of the hth factor, where h = 1, 2…, p.

[0174] Table 2 below is an example table of historical factor data difference values ​​and failure frequency difference values ​​between historical products and historical reference products:

[0175] Historical product pairings Factor 1 Difference Factor 2 Difference Factor p difference Fault frequency difference Product 1 - Product 2 Product 1 - Product 5 Product 1 - Product 9 Product 2 - Product 1

[0176] As shown in Figure 2, there are three historical reference products for product 1, namely product 2, product 5 and product 9.

[0177] In the disclosed embodiment, after determining the difference value of the failure frequency of each historical product relative to the corresponding historical reference product, and the difference value of the historical factor data corresponding to each factor, the electronic device can use linear fitting or nonlinear fitting to construct a mapping between the difference value of the factor data corresponding to the factor and the difference value of the failure frequency, thereby obtaining a mapping model.

[0178] It can be understood that in the implementation of the present disclosure, a historical reference product is found for each historical product, and then a model is constructed based on the failure frequency difference and historical factor data difference between the historical product and the historical reference product to predict the failure frequency difference value of the product to be predicted relative to the reference product to finally predict the failure frequency of the product to be predicted. On the one hand, since the failure frequency difference and historical factor data difference are calculated for each historical product and the corresponding historical reference product during the model construction process, the influence of user-side factors on the model construction effect can be reduced and the accuracy of the model can be improved. On the other hand, the model constructs a mapping between the historical factor data difference and the failure frequency difference, and thus the influence of the factor on the failure frequency can also be quantified, making the model interpretable.

[0179] In some embodiments, the fitting is performed based on the difference value of the failure frequency of each historical product relative to the corresponding historical reference product, and the difference value of the historical factor data corresponding to each factor, and a mapping between the difference value of the factor data corresponding to the factor and the difference value of the failure frequency is constructed to obtain the mapping model, including:

[0180] Based on the difference value of the failure frequency of each historical product relative to the corresponding historical reference product, and the difference value of the historical factor data corresponding to the target factor, a first preset function is used for fitting to determine the fitting parameters of the first preset function;

[0181] Determine a fitting difference value of the failure frequency of each historical product relative to the corresponding historical reference product based on the historical factor data difference of each historical product and the first preset function after parameter adjustment;

[0182] Based on the residual between the fitting difference value of the fault frequency corresponding to each historical product and the difference value of the fault frequency, and the factor data corresponding to the factors other than the target factor, the second preset function is used for fitting, and the fitting parameters of the first preset function are updated to obtain the mapping model.

[0183] The embodiment of the present disclosure performs mapping in two stages when fitting and constructing a mapping model based on the difference value of the failure frequency of each historical product relative to the corresponding historical reference product and the difference value of the historical factor data corresponding to each factor. Among them, fitting is performed based on the difference value of the historical factor data of the target factor first, and then the historical factor data other than the target factor is used to update the fitted function. By performing multiple fittings with the target factor as the main factor and the factors other than the target factor as the auxiliary factor, the accuracy of the mapping model can be improved.

[0184] In the embodiment of the present disclosure, it is assumed that the first preset function is: Δλ im =f(ΔX imk ; k∈X MB ,θ)+∈ ij ; where θ is the regression model parameter. The form of f(·) can be determined according to the data characteristics, such as linear, piecewise linear, piecewise constant, etc., but it must satisfy the central symmetry, that is: for any k∈X MB ,f(ΔX imk ,·,θ)=-f(-ΔX imk ,·,θ). For example, the following linear form of formula (7) can be used by default:

[0185]

[0186] If a linear form is used, the least squares method can be used to solve the model parameter θ k , where ∈ im The initial value can be set to 0. If necessary, we can also introduce monotonicity constraints when solving the model parameters, that is, requiring f(·) to be constant with respect to ΔX ijk is monotonic; monotonically increasing or decreasing can be determined by the experience of domain experts, which can improve the interpretability and robustness of the model. There are currently many mature methods for solving model parameters under constraints, which are not limited in the embodiments of this disclosure. Represents the regression model parameters obtained by solving.

[0187] After determining the fitting parameters of the first preset function, the embodiment of the present disclosure determines the fitting difference value of the failure frequency of each historical product relative to the corresponding historical reference product based on the historical factor data difference of each historical product and the first preset function after parameter adjustment, that is, based on θ determined in formula (7): k , calculate the fitted Δλ im Then, the electronic device calculates the residual of the above formula (7), that is, calculates the difference between the fitting difference value of the fault frequency corresponding to each historical product and the difference value of the fault frequency. The residual can be expressed as the following formula (8):

[0188]

[0189] The embodiment of the present disclosure uses the second preset function to fit again based on the residual to update the fitting parameters of the first preset function, thereby obtaining the final mapping model. The second preset function and the first preset function can be functions of the same type or different types, which is not limited by the embodiment of the present disclosure. The embodiment of the present disclosure uses the factor data other than the target factor to fit using the second preset function. Exemplarily, a regression model can be established as shown in formula (9):

[0190]

[0191] Similarly, the form of g(·) can be determined according to the data characteristics, such as linear, piecewise linear, piecewise constant, etc., but it must satisfy the central symmetry. The linear form can be used by default, as shown in the following formula (10):

[0192]

[0193] If a linear form is used, the least squares method can be used to solve the model parameter η k Similarly, when necessary, monotonicity constraints can be introduced when solving model parameters. Represents the trained ∈ im Model.

[0194] As mentioned above, if the mapping model is a model constructed in multiple steps based on the factor data of the target factor and the factor data of the non-target factor, then when using the preset model to process the difference in each factor data of the predicted product and the reference product to obtain the fault frequency difference value, the factor data difference of the target factor and the factor data difference of the non-target factor need to be specifically introduced. For example, the fault frequency difference value of the predicted product relative to the jth reference product It can be represented by the following formula (11):

[0195]

[0196] Figure 4 FIG. 1 is a diagram showing a failure rate prediction device according to an exemplary embodiment. Figure 4 , the device comprises:

[0197] The acquisition module 101 is configured to acquire the factor data of each historical product and the statistical value of the associated usage, as well as the factor data of the product to be predicted; wherein the factor data is used to characterize the characteristics of the product;

[0198] A first determination module 102 is configured to determine a reference product for the product to be predicted from among the historical products based on the factor data of the product to be predicted and the factor data of each historical product;

[0199] The second determination module 103 is configured to determine the failure rate of the product to be predicted based on the factor data and usage statistics of the reference product and the factor data of the product to be predicted.

[0200] In some embodiments, the apparatus further comprises:

[0201] The third determination module 104 is configured to determine a target factor based on the factor data of each historical product and the statistical value of the associated usage; wherein the correlation between the target factor and the failure of each historical product is greater than the correlation between the factors other than the target factor and the failure of each historical product;

[0202] The first determination module 102 is further configured to determine the difference between the product to be predicted and each historical product based on the data of the target factor in the factor data of the product to be predicted and each historical product; and determine the historical product whose difference between the product to be predicted and each historical product meets the preset difference condition as the reference product of the product to be predicted.

[0203] In some embodiments, the usage statistics include: the historical usage quantity and the historical usage failure quantity within a time period. The third determination module 104 is further configured to determine, for each historical product, the failure frequency of the historical product based on the historical usage quantity and the historical usage failure quantity of the historical product within one or more time periods; construct a causal relationship graph based on the factor data and failure frequency of each historical product; wherein, in the causal relationship graph, there is a directed edge between the factor causing the frequency of each historical product failure to meet a preset frequency threshold and the historical product failure; and the factor pointed to by the directed edge in the causal relationship graph is used as the target factor.

[0204] In some embodiments, the failure frequency of the historical product satisfies the formula:

[0205]

[0206] Among them, λ i represents the failure frequency of the i-th historical product; C it A represents the number of historical failures of the i-th historical product in the t-th time period; it A represents the historical usage quantity of the i-th historical product in the t-th time period, i0 =0.

[0207] In some embodiments, the second determination module 103 is further configured to estimate the usage quantity of the product to be predicted within the preset time period based on the historical usage quantity of the reference product within the preset time period; estimate the number of failures of the product to be predicted within the preset time period based on the factor data of the product to be predicted and the factor data of the reference product, as well as the usage quantity of the product to be predicted within the preset time period and the failure frequency of the reference product; determine the failure rate of the product to be predicted within the preset time period based on the estimated number of failures and usage quantity of the product to be predicted within the preset time period.

[0208] In some embodiments, the second determination module 103 is further configured to determine, for each factor, the factor data difference between the factor data of the product to be predicted and the factor data of the reference product; determine the failure frequency of the product to be predicted based on the difference in each factor data between the product to be predicted and the reference product and the failure frequency of the reference product; and estimate the number of failures of the product to be predicted within the preset time period based on the failure frequency of the product to be predicted and the number of uses of the product to be predicted within the preset time period.

[0209] In some embodiments, the second determination module 103 is further configured to process each factor data difference between the product to be predicted and the reference product using a mapping model between factor data differences and failure frequency differences to obtain a failure frequency difference value of the product to be predicted relative to the reference product; wherein the mapping model is obtained based on the factor data and failure frequency of each historical product; and the failure frequency of the product to be predicted is determined based on the failure frequency of the reference product and the failure frequency difference value of the predicted product relative to the reference product.

[0210] In some embodiments, the apparatus further comprises:

[0211] The fourth determination module 105 is configured to determine the historical reference product of each historical product among all historical products based on the statistical values ​​of the factor data and associated usage of each historical product; determine the difference value between the failure frequency of each historical product and the failure frequency of the corresponding historical reference product; determine, for each factor, the difference value of the historical factor data between the factor data of the historical product and the factor data of the corresponding historical reference product; perform fitting based on the difference value of the failure frequency of each historical product relative to the corresponding historical reference product and the difference value of the historical factor data corresponding to each factor, and construct a mapping between the difference value of the factor data corresponding to the factor and the difference value of the failure frequency to obtain the mapping model.

[0212] In some embodiments, the fourth determination module 105 is further configured to perform fitting using a first preset function based on the difference value of the failure frequency of each historical product relative to the corresponding historical reference product, and the difference value of the historical factor data corresponding to the target factor, to determine the fitting parameters of the first preset function; determine the fitting difference value of the failure frequency of each historical product relative to the corresponding historical reference product based on the difference in historical factor data of each historical product and the first preset function after parameter adjustment; perform fitting using a second preset function based on the residual between the fitting difference value of the failure frequency corresponding to each historical product and the difference value of the failure frequency, and the factor data corresponding to factors other than the target factor, to update the fitting parameters of the first preset function, and to obtain the mapping model.

[0213] In some embodiments, if the reference product includes multiple products, the second determination module 103 is also configured to determine, for each reference product, the sum of the failure frequency difference value of the predicted product relative to the reference product and the failure frequency of the reference product; and determine the average value of the sum corresponding to all reference products as the failure frequency of the product to be predicted.

[0214] In some embodiments, the number of failures of the product to be predicted within the preset time period satisfies the following formula:

[0215]

[0216] Among them, A n+1,t is the usage quantity of the product to be predicted in the tth time period, is the failure frequency of the product to be predicted, A n+1,0 =0.

[0217] In some embodiments, if the reference product includes multiple products, the second determination module 103 is further configured to determine the average of the historical usage quantities of each reference product within the preset time period as the usage quantity of the product to be predicted within the preset time period.

[0218] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0219] Figure 5 900 is a block diagram of an electronic device according to an exemplary embodiment. For example, the electronic device (device 900) may be a server for executing a failure rate prediction method.

[0220] Reference Figure 5, the apparatus 900 includes a processing component 922, which further includes one or more processors, and a memory resource represented by a memory 932 for storing instructions, such as an application, that can be executed by the processing component 922. The application stored in the memory 932 may include one or more modules, each corresponding to a set of instructions. In addition, the processing component 922 is configured to execute the instructions to perform the above method.

[0221] The device 900 may also include a power supply component 926 configured to perform power management of the device 900, a wired or wireless network interface 950 configured to connect the device 900 to a network, and an input / output (I / O) interface 958. The device 900 may operate based on an operating system stored in the memory 932, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, or the like.

[0222] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 932 including instructions, and the instructions can be executed by the processing component 922 of the device 900 to perform the above method. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.

[0223] A non-transitory computer-readable storage medium, when instructions in the storage medium are executed by a processor of an electronic device, enables the electronic device to perform a failure rate prediction method, the method comprising:

[0224] Obtaining the factor data of each historical product and the statistical value of the associated usage, as well as the factor data of the product to be predicted; wherein the factor data is used to characterize the characteristics of the product;

[0225] Based on the factor data of the product to be predicted and the factor data of each historical product, determining a reference product for the product to be predicted from each historical product;

[0226] The failure rate of the product to be predicted is determined based on the factor data and usage statistics of the reference product and the factor data of the product to be predicted.

[0227] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present disclosure is intended to cover any variations, uses or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The description and examples are to be considered exemplary only, and the true scope and spirit of the present disclosure are indicated by the claims above.

[0228] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A failure rate prediction method, characterized in that: The method comprises: Obtaining the factor data of each historical product and the statistical value of the associated usage, as well as the factor data of the product to be predicted; wherein the factor data is used to characterize the characteristics of the product; Based on the factor data of the product to be predicted and the factor data of each historical product, determining a reference product for the product to be predicted from each historical product; The failure rate of the product to be predicted is determined based on the factor data and usage statistics of the reference product and the factor data of the product to be predicted.

2. The method according to claim 1, characterized in that: The method further comprises: Determine a target factor based on the factor data of each historical product and the statistical value of the associated usage; wherein the correlation between the target factor and the failure of each historical product is greater than the correlation between factors other than the target factor and the failure of each historical product; The step of determining a reference product for the product to be predicted from among the historical products based on the factor data of the product to be predicted and the factor data of each historical product includes: Determine the difference between the product to be predicted and each historical product based on the data of the target factor in the factor data of the product to be predicted and each historical product; The historical products whose differences between the product to be predicted and each historical product satisfy a preset difference condition are determined as reference products for the product to be predicted.

3. The method according to claim 2, characterized in that The usage statistics include: the number of historical uses and the number of historical use failures within a time period. The target factor is determined based on the factor data of each historical product and the statistics of the associated usage, including: For each historical product, determine the failure frequency of the historical product based on the historical usage quantity and the historical usage failure quantity of the historical product in one or more time periods; Based on the factor data and failure frequency of each historical product, a causal relationship graph is constructed; wherein in the causal relationship graph, there is a directed edge between the factor causing the frequency of each historical product failure to meet a preset frequency threshold and the historical product failure; The factor pointed to by the directed edge in the causal relationship graph is used as the target factor.

4. The method according to claim 3, characterized in that The failure frequency of the historical product satisfies the formula: Among them, λ i represents the failure frequency of the i-th historical product; C it A represents the number of historical failures of the i-th historical product in the t-th time period; it A represents the historical usage quantity of the i-th historical product in the t-th time period, i0 =0.

5. The method according to claim 3, characterized in that: The determining the failure rate of the product to be predicted based on the factor data and usage statistics of the reference product and the factor data of the product to be predicted includes: Based on the historical usage quantity of the reference product within the preset time period, estimating the usage quantity of the product to be predicted within the preset time period; Based on the factor data of the product to be predicted and the factor data of the reference product, as well as the usage quantity of the product to be predicted in the preset time period and the failure frequency of the reference product, estimate the failure quantity of the product to be predicted in the preset time period; Based on the estimated number of failures and the number of uses of the product to be predicted within the preset time period, the failure rate of the product to be predicted within the preset time period is determined.

6. The method according to claim 5, characterized in that The estimating the number of failures of the product to be predicted within the preset time period based on the factor data of the product to be predicted and the factor data of the reference product, the usage number of the product to be predicted within the preset time period, and the failure frequency of the reference product, comprises: For each factor, determining the factor data difference between the factor data of the product to be predicted and the factor data of the reference product; Determine the failure frequency of the product to be predicted based on the difference in each factor data between the product to be predicted and the reference product and the failure frequency of the reference product; Based on the failure frequency of the product to be predicted and the usage quantity of the product to be predicted within the preset time period, the failure quantity of the product to be predicted within the preset time period is estimated.

7. The method according to claim 6, characterized in that The determining the failure frequency of the product to be predicted based on the difference in each factor data between the product to be predicted and the reference product and the failure frequency of the reference product comprises: Using a mapping model between factor data differences and failure frequency differences, each factor data difference between the product to be predicted and the reference product is processed to obtain a failure frequency difference value of the product to be predicted relative to the reference product; wherein the mapping model is obtained based on the factor data and failure frequency of each historical product; The failure frequency of the product to be predicted is determined based on the failure frequency of the reference product and the difference value of the failure frequency of the predicted product relative to the reference product.

8. The method according to claim 7, characterized in that The method further comprises: Determine the historical reference product of each historical product among the historical products based on the factor data of each historical product and the statistical value of the associated usage; Determine the difference between the failure frequency of each historical product and the failure frequency of the corresponding historical reference product; For each factor, determine the historical factor data difference value between the factor data of the historical product and the factor data of the corresponding historical reference product; Based on the difference value of the failure frequency of each historical product relative to the corresponding historical reference product and the difference value of the historical factor data corresponding to each factor, fitting is performed to construct a mapping between the difference value of the factor data corresponding to the factor and the difference value of the failure frequency to obtain the mapping model.

9. The method according to claim 8, characterized in that The method of fitting based on the difference value of the failure frequency of each historical product relative to the corresponding historical reference product and the difference value of the historical factor data corresponding to each factor, constructing a mapping between the difference value of the factor data corresponding to the factor and the difference value of the failure frequency, and obtaining the mapping model includes: Based on the difference value of the failure frequency of each historical product relative to the corresponding historical reference product, and the difference value of the historical factor data corresponding to the target factor, a first preset function is used for fitting to determine the fitting parameters of the first preset function; Determine a fitting difference value of the failure frequency of each historical product relative to the corresponding historical reference product based on the historical factor data difference of each historical product and the first preset function after parameter adjustment; Based on the residual between the fitting difference value of the fault frequency corresponding to each historical product and the difference value of the fault frequency, and the factor data corresponding to the factors other than the target factor, the second preset function is used for fitting, and the fitting parameters of the first preset function are updated to obtain the mapping model.

10. The method according to claim 7, characterized in that If the reference product includes multiple products, determining the failure frequency of the product to be predicted based on the failure frequency of the reference product and the difference value of the failure frequency of the predicted product relative to the reference product includes: For each reference product, determining the sum of the failure frequency difference value of the predicted product relative to the reference product and the failure frequency of the reference product; The average value of the sum values ​​corresponding to all reference products is determined as the failure frequency of the product to be predicted.

11. The method according to claim 6, characterized in that The number of failures of the product to be predicted within the preset time period satisfies the following formula: Among them, A n+1,t is the usage quantity of the product to be predicted in the tth time period, is the failure frequency of the product to be predicted, A n+1,0 =0.

12. The method according to claim 5, characterized in that If the reference product includes multiple products, estimating the usage quantity of the product to be predicted within the preset time period based on the historical usage quantity of the reference product within the preset time period includes: The average of the historical usage quantities of each reference product within the preset time period is determined as the usage quantity of the product to be predicted within the preset time period.

13. A failure rate prediction device, characterized in that: The device comprises: An acquisition module configured to acquire factor data of each historical product and statistics of associated usage conditions, as well as factor data of the product to be predicted; wherein the factor data is used to characterize the characteristics of the product; A first determination module is configured to determine a reference product of the product to be predicted from each historical product based on the factor data of the product to be predicted and the factor data of each historical product; The second determination module is configured to determine the failure rate of the product to be predicted based on the factor data and usage statistics of the reference product and the factor data of the product to be predicted.

14. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to execute the method as claimed in any one of claims 1 to 12.

15. A non-transitory computer-readable storage medium, characterized in that: When the instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to perform the method as claimed in any one of claims 1 to 12.

Images