Circuit board reliability prediction method and circuit board stress accelerated life experiment method

Abstract

The present invention discloses a method for predicting the reliability of a circuit board and a method for stress accelerated life experiment of the circuit board, which includes collecting circuit board images; using a component model to detect electronic components in the circuit board images. If all types of electronic components are detected, the types and quantities of the electronic components are saved. If unknown electronic components are detected, the pictures of the unknown electronic components are supplemented into the sample set of the component model, and at the same time, the parameters of the new components are supplemented in the format of the database list, and the component model is retrained to obtain a new component model until all types of electronic components are detected; according to the saved types and quantities of the electronic components and the general failure rate of the electronic components, the reliability parameters of the circuit board are calculated. Advantages: Automatically obtain the types and quantities of components affecting the reliability index on the circuit board through the component model, can quickly and directly complete the reliability prediction and evaluation, and realize the recommendation of reliability test parameters.

Description

Technical Field

[0001] The present invention relates to a method for predicting the reliability of a circuit board and a method for stress accelerated life experiment of a circuit board, belonging to the technical fields of image recognition and electronic aided design. Background Art

[0002] The purpose of reliability prediction is to provide a reliable basis for selecting the best design scheme for reliability design, and the reliability of a new product can be estimated before the product is finalized and put into production. If the predetermined index is not reached, the design can be improved in time for the problems found. The reliability index usually selects the failure rate or the reciprocal of the failure rate, that is, the mean time between failures (MTBF). At present, most of the reliability predictions of circuit boards use existing national standards, military standards and special standards of various industries to obtain prediction values (the common index parameters are the mean time between failures MTBF and the failure rate). The main prediction methods are the component counting method or the stress test analysis method. Among them, the component counting method mainly estimates the reliability of a product based on the failures of electronic components and devices, the number of components and devices used in product design, and the usage conditions. The stress analysis method is based on the basic failure rate of components. According to the working environment, quality grade and working environment stress of components, multiplying by the corresponding correction factor can predict the working failure rate of components in the product, and further the failure rate of components can be obtained, and finally the failure rate of the product can be obtained. The basis of these two reliability prediction methods is that the application conditions are that the types, quantities, quality grades, etc. of components have been basically determined, especially on the basis of a detailed Bill of Material (BOM) of the circuit board. However, in the actual development process, especially in the design of some self-developed circuit boards completed on the basis of the secondary development of finished circuit boards, it is often impossible to obtain the detailed types and quantities of components.

[0003] The current prediction method is supported based on the clear BOM of the overall circuit board, and mainly through the reliability index or other indexes of the overall electricity, which cannot take into account the product design developed on the basis of the circuit board in secondary development, that is, the reliability prediction of a part lacking the BOM list. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to overcome the defects of the prior art and provide a method for predicting the reliability of a circuit board and a method for stress accelerated life experiment of a circuit board.

[0005] To solve the above technical problem, the present invention provides a method for predicting the reliability of a circuit board, including:

[0006] Collecting an image of the circuit board;

[0007] Use a pre-trained component recognition model to detect electronic components in the circuit board image. If all types of electronic components are detected, the types and numbers of electronic components are saved. If unknown electronic components are detected, the image of the unknown electronic component is added to the sample set of the component model. At the same time, the parameters of the unknown electronic component are added according to the database list format. Subsequently, the component recognition model is retrained to obtain a new component recognition model. The new component recognition model is then used to detect electronic components in the circuit board image until all types of electronic components are detected.

[0008] The reliability parameters of the circuit board are calculated based on the types and numbers of the stored electronic components and the pre-acquired general failure rates of the electronic components.

[0009] Furthermore, the training of the component recognition model includes: the sample set of the component model is a three-dimensional picture of the component or a front picture of the component; 2 / 3 of it is extracted as a training sample set and the other 1 / 3 is extracted as a test sample set; the training sample set is used as the input of the convolutional neural network model, and the training and iteration are continuously performed to obtain a trained component recognition model, and the trained recognition model is tested using the test sample set. If the accuracy meets the preset requirements, the trained component recognition model is obtained.

[0010] Furthermore, the model for calculating the reliability parameters of the circuit board is:

[0011]

[0012]

[0013] Where t0 is the specific time of fault-free operation actually calculated for the circuit board; λ0 is the total failure rate actually calculated for the circuit board; λ ci is the general failure rate of the i-th component; π Qi is the general quality coefficient of the i-th component; N i is the number of components of type i; n is the number of types of components used in the circuit board.

[0014] Furthermore, the general failure rate model of the i-th component is:

[0015] λ ci ==λ b λ f1 λ f2 λ E (3)

[0016] Among them, λ b is the basic failure rate; f1 is the circuit complexity failure rate; f2is the failure rate due to encapsulation complexity; λ E is the failure rate due to environmental complexity.

[0017] Furthermore, the general failure rate and the general quality factor are pre-stored in a database and are used to be called when calculating the reliability parameters of the circuit board.

[0018] A method for stress accelerated life experiment of a circuit board includes:

[0019] Calculating the reliability parameters of the circuit board by using a circuit board reliability prediction method, where the reliability parameters of the circuit board are the specific time of fault-free operation actually calculated for the circuit board;

[0020] Converting the specific time of fault-free operation actually calculated for the circuit board and the pre-obtained failure rate of the circuit board under usage conditions into an acceleration factor for the failure rate under the accelerated test, and calculating the final test time for reliability verification under the accelerated test conditions.

[0021] Furthermore, the model for calculating the final test time for reliability verification under the accelerated test conditions is:

[0022]

[0023]

[0024]

[0025] where, t test is the final test time for reliability verification under the accelerated test conditions; t0 is the specific time of fault-free operation actually calculated for the circuit board; λ0 is the total failure rate actually calculated for the circuit board; A test is the acceleration factor for converting the failure rate of the circuit board under usage conditions into the failure rate under the accelerated test; A k is the k-th acceleration factor for each increased stress in the test; λ ci is the general failure rate of the i-th component; π Qi is the general quality factor of the i-th component; N i is the quantity of the i-th component; n is the number of types of components used in the circuit board, i is the i-th component; (∏ k A k ) j is the product of the acceleration factors of the k failure modes under the stress j; N s is the number of stresses.

[0026] Furthermore, the stress includes one or more of the three stresses of temperature, vibration, and shock.

[0027] A computer-readable storage medium storing one or more programs, wherein the one or more programs include instructions that, when executed by a computing device, cause the computing device to execute any one of the methods described above.

[0028] A computing device, comprising,

[0029] One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for executing any one of the methods described above.

[0030] Advantages achieved by the present invention:

[0031] The present invention automatically obtains the types and quantities of components on the circuit board that affect the reliability indicators through component models, and directly performs reliability prediction calculations using the types and quantities of components and the general failure rates of electronic components, enabling rapid and direct completion of reliability prediction evaluations and recommending reliability test parameters for this parameter. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 Is the hardware connection diagram for the reliability prediction evaluation of the present invention;

[0033] Figure 2 Is the software system structure diagram for the reliability prediction evaluation of the present invention;

[0034] Figure 3 Is the flowchart of the reliability prediction evaluation system;

[0035] Figure 4 Is the flowchart of the image recognition component type and number list. DETAILED DESCRIPTION OF THE INVENTION

[0036] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and should not be used to limit the protection scope of the present invention.

[0037] A circuit board reliability prediction method is based on relatively accurate image recognition. This reliability evaluation process is based on automatically recognizing components in images and completing number statistics, and the statistical results are passed to the reliability prediction system software component to call the failure rate parameters and models, finally completing the circuit board reliability prediction evaluation.

[0038] If Figure 1 And Figure 2 As shown, the hardware part of the present invention includes a camera element and a computer integrated with software, and judges the type and quantity of components through an electronic component recognition system, and completes the reliability prediction through a reliability analysis system.

[0039] Based on the circuit board reliability prediction method, a circuit board stress accelerated life experiment method is also provided. By inputting the test type, the built-in test parameter calculation conversion model automatically completes the matching of reliability test parameter data.

[0040] As Figure 3 shown, the steps of reliability prediction are as follows:

[0041] 1. Start the camera to obtain an image;

[0042] 2. Use the component recognition model to recognize the image data;

[0043] 3. Display the component types and corresponding quantities;

[0044] 4. Transmit the data to the component reliability prediction model;

[0045] 5. Calculate the reliability parameters according to the reliability prediction model.

[0046] In this example, the circuit board reliability prediction model selects the component count method model, and the mean time between failures MTBF:

[0047]

[0048]

[0049] Among them,

[0050] t0 is the specific time of the circuit board's actual calculated trouble-free operation, with the unit h;

[0051] λ0 is the total failure rate of the circuit board's actual calculation, with the unit 10 -6 / h;

[0052] λ ci is the general failure rate of the i-th component, with the unit 10 -6 / h;

[0053] π Qi is the general quality factor of the i-th component;

[0054] N i is the quantity of the i-th component;

[0055] n is the number of types of components used in the circuit board;

[0056] The general failure rate λ ci and the general quality factor π Qi of the i-th component are taken from the standard MIL-HDBK-2l7F, and the database information is stored in the reliability model prediction database.

[0057] The general failure rate λ of the i-th component ci , and the corresponding database information is stored in the reliability model prediction database according to categories.

[0058] The operating failure rate model of a single component is λ ci = λ b λ f1 λ f2 λ E , where λ b is the basic failure rate, which is related to the type of component. Specifically, it is divided into integrated circuits and discrete components: integrated circuits include monolithic digital integration, monolithic analog integrated circuits, and hybrid integrated circuits; discrete components include diodes, triodes, resistors, capacitors, inductors, etc.

[0059] The circuit complexity failure rate λ f1 : Affected by factors such as the number of gates, transistors, and digits in the circuit, its value is equal to the sum of the complexity failure rates of each factor.

[0060] The package complexity failure rate λ f2 : Determined jointly by factors such as packaging processes such as BGA, SOP, FGA, and the number of pins, and its value is equal to the sum of the complexity failure rates of each factor.

[0061] The environmental complexity failure rate λ E : Affected by other factors such as the working environment, temperature stress, voltage stress, and working power, and its value is equal to the sum of the complexity failure rates of each factor.

[0062] By calling the matching in the database, selecting the type of component, packaging method, working parameters, and the specific parameters of the component, the model library of the expected component or similar components can be quickly matched, and the corresponding operating failure rate can be obtained.

[0063] The model library information of the component failure rate model electronic component database storing information classes is written in XML language, and the specific elements include: class, package, number of gates, number of pins, working power, classified as follows:

[0064] Component

[0065] Name: A certain integrated chip / resistor / capacitor / inductor

[0066] Class: Digital circuit, micro, DSP

[0067] Package class: SO, SOP, BGA

[0068] Number of gates (optional): Number of gates

[0069] Working power: 1 / 4W of ordinary resistor, others for capacitor and diode

[0070] Pin number (optional): If the chip name is recognized, the pin number and range can be automatically matched.

[0071] VI. Reliability data MTBF test results show;

[0072] Figure 4 It is a flowchart of the list of types and numbers of image recognition components. The main steps are as follows:

[0073] (I). Turn on the camera to obtain an image of the electronic device;

[0074] (II). Based on the pre-trained component model, complete the feature extraction of the picture data and label data, and detect the components in the electronic device image;

[0075] (III). Determine whether there are unknown electronic components;

[0076] (IV). If so, supplement the three-dimensional and physical information pictures of the picture, and execute the model fit method (a method in Python to directly train new data based on the existing model) on the original model to complete the updated training model, and return to the second step; if not, identify the types and numbers of electronic components in the image data. The component recognition model of the present invention adopts a convolutional neural network (Inception-v4) model.

[0077] (V). Match the information in the component database, and obtain the component working failure rate according to the information stored in the electronic component database.

[0078] VII. Select the test type, the test parameter calculation conversion model, and determine the reliability test parameters;

[0079] In this example, the test parameter calculation conversion model selected for the circuit board reliability test parameter matching is the stress accelerated life experiment model:

[0080]

[0081]

[0082] Among them,

[0083] t test is the final test time of reliability verification under the accelerated test conditions;

[0084] λ0 is the total failure rate under the actual conditions of the circuit board, with the unit of 10 -6 / h;

[0085] A test is the acceleration factor for converting the failure rate of the circuit board under the use conditions into the failure rate under the accelerated test;

[0086] A k is the k-th acceleration factor for each increased stress in the test;

[0087] (∏ k A k ) j is the product of the acceleration factors of the k failure modes under stress j;

[0088] N s is the number of stresses. Usually, three stresses of temperature, vibration, and shock are adopted, where T max , T min , W max , G max are the limit stresses of the three parameters of temperature, vibration, and shock for the normal operation of the circuit board. According to the limit stresses of the three parameters of temperature, vibration, and shock, the acceleration factor of the specific failure mode in the specific application environment can be determined from the stress database (generated from the stress data table in standard GB / T29309). At the same time, a single stress is allowed, and only temperature or vibration or shock is selected to implement the reliability verification test.

[0089] Figure 4 is the flowchart of the list of types and numbers of image recognition components. Its main steps are as follows:

[0090] I. Turn on the camera to obtain the circuit board image;

[0091] II. Based on the pre-trained component model, complete the feature extraction of the picture data and label data, and detect the components on the circuit board image;

[0092] III. Determine whether there are unknown electronic components;

[0093] IV. If there are, supplement the three-dimensional and physical information pictures of the pictures, and execute the model fit method on the original model (the method of directly training new data based on the existing model in python) to complete the update training of the model. Return to the second step;

[0094] If not, identify the types and numbers of electronic components in the current image data.

[0095] The present invention takes into account the usage scenarios of reliability prediction data. The system is based on relatively accurate image recognition, and the reliability evaluation process is based on automatically recognizing components in images and completing the number statistics. The statistical results are passed into the reliability prediction system software component to call the failure rate parameters and models, and finally complete the reliability prediction and evaluation of the circuit board. By manually inputting the test type, the built-in test parameter calculation conversion model automatically completes the matching of the reliability test test parameter data.

[0096] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0097] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one or more of the processes Figure 1 or multiple processes and / or blocks

[0098] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means implements the functions specified in Figure 1 one or more of the processes Figure 1 or multiple processes and / or blocks

[0099] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one or more of the processes Figure 1 or multiple processes and / or blocks

[0100] The above is only the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and modifications can be made, and these improvements and modifications should also be regarded as the protection scope of the present invention.

Claims

1. A method for stress accelerated life experiment of a circuit board, characterized in that, Including: Calculating the reliability parameters of the circuit board by using a circuit board reliability prediction method, where the reliability parameters of the circuit board are the specific time of the circuit board's actual calculated fault-free operation; Converting the specific time of the circuit board's actual calculated fault-free operation and the pre-obtained failure rate of the circuit board under operating conditions into an acceleration factor for the failure rate under the accelerated test, and calculating the final test time for reliability verification under the accelerated test conditions; The model for calculating the final test time for reliability verification under the accelerated test conditions is: ； ； ； Among them, is the final test time for reliability verification under accelerated test conditions; is the specific time of the circuit board's actual calculated trouble-free operation; is the total failure rate of the circuit board's actual calculation; is the acceleration factor that converts the failure rate of the circuit board under use conditions to the failure rate under accelerated test; is the k th acceleration factor for each increased stress in the test; is the general failure rate of the i-th component; is the general quality factor of the i-th component; is the quantity of the i-th component; n is the number of types of components used in the circuit board, and i is the i-th component; is the product of the acceleration factors of the j th failure mode under the action of stress k ; is the number of stresses; The calculating the reliability parameters of the circuit board by using a circuit board reliability prediction method includes: Collecting circuit board images; Using a pre-trained component recognition model to detect the electronic components in the circuit board image. If all types of electronic components are detected, save the types and quantities of the electronic components. If unknown electronic components are detected, supplement the pictures of the unknown electronic components to the sample set of the component model, and at the same time supplement the parameters of the unknown electronic components in the database list format. Subsequently, perform re-training on the component recognition model to obtain a new component recognition model, and then use the new component recognition model to detect the electronic components in the circuit board image until all types of electronic components are detected; Calculating the reliability parameters of the circuit board according to the saved types and quantities of the electronic components and the pre-obtained general failure rates of the electronic components.

2. The method for stress accelerated life experiment of a circuit board according to claim 1, wherein The stress includes one or more of the three stresses of temperature, vibration, and shock.

3. The method for stress accelerated life experiment of a circuit board according to claim 1, characterized in that, The training of the component recognition model includes: the sample set of the component model is three-dimensional pictures of components or front pictures of components; extract 2 / 3 of them as the training sample set and the other 1 / 3 as the test sample set; use the training sample set as the input of the convolutional neural network model, continuously train and iterate to obtain a trained component recognition model, and use the test sample set to test the trained recognition model. When the accuracy meets the preset requirements, the trained component recognition model is obtained.

4. The method for stress accelerated life experiment of a circuit board according to claim 1, wherein The model for the general failure rate of the i-th component is: = (3) Among them, is the basic failure rate; is the failure rate due to circuit complexity; is the failure rate due to packaging complexity; is the failure rate due to environmental complexity.

5. The method for stress accelerated life test of a circuit board according to claim 1, wherein The general failure rate and the general quality coefficient are pre-stored in the database and are used to be called when calculating the reliability parameters of the circuit board.

6. A computer-readable storage medium storing one or more programs, characterized in that, The one or more programs include instructions that, when executed by a computing device, cause the computing device to execute any one of the methods according to claims 1 to 5.

7. A computing device, characterized in that, Including, One or more processors, a memory, and one or more programs, where the one or more programs are stored in the memory and are configured to be executed by the one or more processors. The one or more programs include instructions for executing any one of the methods according to claims 1 to 5.

Images