A Simulation Method for Product Reliability Growth

Abstract

The invention relates to a simulation method for product reliability growth, which takes product parameters and working parameters as main factors affecting the process of formulating a product reliability growth plan, and simulates the product reliability growth process by means of a mathematical simulation test; wherein the product parameters are inherent properties of the product; the working parameters are factors affecting the change of the MTBF of the product, and whether the product to be tested reaches the expected MTBF index is judged by adjusting the working parameters; a series of problems such as the large difference between the formulation of the product reliability growth plan and the actual product reliability growth test process due to the difficulty in quantitatively evaluating the product reliability growth rate m, which may lead to the delayed delivery of later projects, etc., are solved; the product reliability growth test reaches the expected MTBF index, and the relevant settings that meet the expected parameters are used as the basis for formulating the product reliability growth plan, so that the formulated reliability growth plan can be guaranteed to be consistent with the actual product reliability growth test process.

Description

Technical Field

[0001] The invention belongs to the technical field of product reliability research, and in particular relates to a simulation method for product reliability growth in a reliability growth test process. Background Art

[0002] Reliability growth refers to the process of discovering defects and improving products through "test-analysis-improvement-retest" TAAF (Test Analysis And Fix) during product development. The test can be a field test or a test of simulated users in the laboratory. The indicator for evaluating the improvement effect of TAAF is generally MTBF (Mean Time Between Failure). A common method in the prior art is to formulate and implement a product reliability growth plan through the reliability growth rate M. Specifically, the test time required for the product is calculated through the reliability growth rate M; however, the reliability growth rate M depends on many factors, such as expert decision-making, working hours, etc., which all have an impact on the reliability growth rate M. In other words, the value of M is difficult to quantify in a quantitative way. Since the value of M is difficult to quantify, the M value is usually set based on work experience during product testing. However, this is likely to result in the actual product reliability test time required in the later stage being much longer than the planned test time in the reliability growth plan initially formulated based on the reliability growth rate m, which may lead to a series of problems such as the project being unable to be delivered in time. That is, the conventional technology has the problem that the product reliability growth plan formulated based on the growth rate m is not practical, which may affect the overall efficiency of the project and bring about a series of problems such as project delays. Summary of the invention

[0003] In view of the fact that it is difficult to quantitatively evaluate the product reliability growth rate m, which leads to a large difference between the formulation of the product reliability growth plan and the actual product reliability growth test process, and thus may lead to a series of problems such as delayed delivery of later projects, the present invention proposes a simulation method for product reliability growth, which quantifies the work content in the product reliability growth test process into specific parameters, simulates the entire process of the product reliability growth test through the quantifiability of the parameters and the test means of simulation; by continuously updating and optimizing the quantitative parameters, repeatedly performing the process of "finding product faults, optimizing faults, and continuing testing", finally achieving the expected MTBF index of product reliability growth, and taking the parameter settings involved in the test process to meet the index as the basis for formulating the product reliability growth plan, thereby ensuring that the formulated reliability growth plan is consistent with the actual product reliability growth test process, and improving the accuracy of the plan formulation to improve the efficiency in the subsequent actual application process of the product reliability growth test, thereby reducing a series of problems such as increased testing costs, low communication efficiency, customer loss, and reduced orders caused by the failure to deliver products in a timely manner. This method is applicable to all products and industries that require reliability growth testing, such as electronic components (capacitors, resistors, diodes, integrated circuits, etc.), or mature electronic products (mobile phones, laptops, wind turbines, photovoltaic inverters, automobiles, household appliances, etc.).

[0004] The technical solution of the present invention is:

[0005] A simulation method for product reliability growth, taking product parameters and working parameters as the main factors affecting the process of formulating a product reliability growth plan, simulates the product reliability growth process by means of a mathematical simulation test; wherein the product parameters are inherent properties of the product; the working parameters are factors affecting the change of the MTBF of the product, and whether the product to be tested reaches the expected MTBF index is determined by adjusting the working parameters; the specific steps of the simulation method for product reliability growth are as follows:

[0006] Step 1: Quantify the fault characteristics of the product to be tested, including the following quantitative factors: number of fault categories, number of fault causes, proportion of each fault category, proportion of each fault cause, and use statistical methods to express the concentration of fault categories;

[0007] Step 2: Based on the triggering conditions of the fault analysis, determine whether the product under test needs to be analyzed for faults, and determine whether the requirements for fault improvement are met;

[0008] Step 3: Based on the quantitative factors of the fault characteristics of the product to be tested and the judgment on whether fault analysis and fault improvement are needed, a mathematical simulation test is conducted on the reliability growth of the product to be tested.

[0009] Furthermore, step one includes the following sub-steps:

[0010] Step 1.1: Determine the initial MTBF0 of the whole product to be tested, and calculate the initial failure rate λ0 of the whole product based on the formula: λ=1 / MTBF, where λ is the failure rate of the product to be tested, and MTBF is the initial mean time between failures of the whole product to be tested;

[0011] Step 1.2: Determine the number of fault categories of the whole product to be tested;

[0012] Step 1.3: Determine the proportion of each fault category to the total failure rate of the product: Determine the proportion of the number of failures of each fault category to the total failure rate of the product based on statistical methods. By counting the frequency of occurrence of each fault category, the concentration of the fault category is expressed by the shape parameter β.

[0013] Step 1.4: Determine the number of potential defects under the fault category of each product: Generate random numbers to simulate the number of potential defects under the fault category of a product;

[0014] Step 1.5: Determine the proportion of the failure rate of each potential defect to the failure rate of its corresponding fault category.

[0015] Furthermore, step 1.3 specifically includes the following sub-steps:

[0016] Step 1.3.1: Generate a set of random numbers that satisfy the Weibull distribution with a shape parameter of β. The number of random numbers is the number of fault categories.

[0017] Step 1.3.2: Normalize this set of random numbers. The sum of the probabilities after normalization is 1. Each probability corresponds to the proportion of the failure rate of each fault category.

[0018] Furthermore, step 2 specifically includes the following sub-steps:

[0019] Step 2.1: Determine whether to perform fault analysis: Determine whether the triggering conditions for fault analysis are met. If the triggering conditions for fault analysis are met, start the fault analysis process;

[0020] Step 2.2: Whether the fault cause is analyzed: Determine whether the probability of analyzing the real fault cause at least reaches the expected threshold;

[0021] Step 2.3: Determine whether the requirements for fault improvement are met: After analyzing the real cause of the fault, consider whether there is a corresponding improvement plan for the fault and whether the improvement plan meets the cost requirements;

[0022] Step 2.4: Determine the improvement ratio: reduce the failure rate of each component with potential defects to a preset ratio as the determined improvement ratio;

[0023] Step 2.5: Implement improvement strategy: adopt the overall improvement strategy or batch improvement strategy for the tested products.

[0024] Furthermore, the method further comprises the following steps:

[0025] Step 2.6: Determine the number of prototypes: Determine the number of prototypes used for reliability growth testing;

[0026] Step 2.7: Determine the test time: determine the time when the reliability growth test can be performed, and the test time meets a preset time range;

[0027] Step 2.8: Set probability threshold: Before the mathematical simulation test of reliability growth, determine a probability threshold to characterize the minimum requirement that the mathematical simulation test meets the expected MTBF.

[0028] Furthermore, step three specifically includes the following sub-steps:

[0029] Step 3.1: Fault occurrence: Random faults occur in the product, and the fault interval, fault category code, and failure rate of each potential defect meet the set values ​​in step 1;

[0030] Step 3.2: Fault handling: Decide whether to conduct fault analysis on the product based on the conditions that trigger the fault analysis, and determine whether the product meets the requirements for fault improvement; if the product meets the requirements for fault improvement, implement the final improvement strategy for the product;

[0031] Step 3.3: Update failure rate: After implementing the improvement strategy, update the failure rate of the entire product, fault type, and fault cause;

[0032] Step 3.4: Repeat steps 3.1-3.3. If the calendar time reaches the test time determined in step 2.7, terminate the simulation;

[0033] Step 3.5: Display reliability growth results: Dynamically display the cumulative operating time, instantaneous failure rate, and instantaneous MTBF of each product unit failure recorded in the reliability growth mathematical simulation test, and determine whether the instantaneous MTBF of the product unit can be increased to the expected instantaneous MTBF of the product unit after the test within the expected time;

[0034] Step 3.6: Repeat simulation: Perform multiple mathematical simulation tests, take the results of several simulation tests for probability statistics, and compare them with the probability threshold set in step 2.8 to determine whether the expected MTBF is achieved; if the statistical MTBF probability value is less than the set probability threshold, it is necessary to adjust the simulation related parameters and repeat steps 3.1 to 3.5 until the expected MTBF can be achieved.

[0035] Preferably, the statistical distribution in step one is any one of Weibull distribution, uniform distribution and normal distribution.

[0036] Preferably, the conditions for triggering fault analysis include: performing fault analysis immediately when a fault occurs, or performing fault analysis when the cumulative number of fault occurrences reaches a preset number.

[0037] Preferably, in step 2.7, the maximum value of the time range is the total number of days from the current test time to the day before the product is officially mass-produced; the minimum value of the time range satisfies the formula: N*M*24 / MTBF0≥S, wherein N is the total number of days of the minimum value of the time range, M is the number of samples of the tested product, MTBF0 is the initial MTBF of the whole product to be tested, and S is a threshold determined based on the test risk level.

[0038] Preferably, in step 3.6, the results of several simulation tests are taken for probability statistics and compared with the probability threshold set in step 2.8, specifically: taking the expected MTBF index as a benchmark, the ratio of the number of tests in which the MTBF obtained in several simulation tests is greater than or equal to the expected MTBF index to the total number of simulation tests is taken as the probability statistical value of the simulation test, and the probability statistical value is compared with the probability threshold set in step 2.8. If it is greater than or equal to the set probability threshold, it means that the possibility of the simulation test results reaching the expected MTBF has met the requirements.

[0039] The beneficial effects of the present invention are:

[0040] The present invention simulates the product reliability growth process, quantifies the work content involved in the reliability growth process, and continuously adjusts the quantified parameters to enable the product reliability growth test to achieve the expected MTBF index, and uses the relevant settings of the parameters involved in the test process that meets the expected MTBF index as the basis for formulating a product reliability growth plan, thereby ensuring that the formulated reliability growth plan is consistent with the actual product reliability growth test process, and improving the accuracy of plan formulation to improve the efficiency in the subsequent actual application process of product reliability growth testing, thereby reducing a series of problems such as increased testing costs, low communication efficiency, customer loss, and reduced orders caused by the failure to deliver products in a timely manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a graph showing the relationship between the MTBF growth and the product reliability growth test time of the present invention;

[0042] Figure 2 It is a TAAF process diagram for increasing the reliability of the product of the present invention;

[0043] Figure 3 This is a relationship diagram of the MTBF growth, failure rate and product reliability growth test time of the present invention;

[0044] Figure 4 This is a process diagram of the reliability growth simulation test of the product of the present invention;

[0045] Figure 5 This is a simulation flow chart of the reliability growth of the product of the present invention;

[0046] Figure 6 It is a reliability growth simulation-cumulative failure number diagram of the product of the present invention;

[0047] Figure 7 This is a reliability growth simulation of the product of the present invention - instantaneous MTBF and cumulative failure number diagram;

[0048] Figure 8 This is a graph of reliability growth simulation of the product of the present invention - instantaneous MTBF and test days (50 days);

[0049] Fig. 9 This is a graph of reliability growth simulation of the product of the present invention - instantaneous MTBF and test days (100 days). DETAILED DESCRIPTION

[0050] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0051] Since the growth process of MTBF has a certain trend law, the growth process of MTBF can be described by mathematical formula. The common mathematical method is to use the Duane model or the AMSAA model. The instantaneous MTBF formula of the Duane model is as follows:

[0052]

[0053] Where a is a parameter related to the initial MTBF of the product; m is the reliability growth rate; and T is the total test time of the product.

[0054] Since the MTBF growth curve can be described by the reliability growth model, the reliability growth model can also be used as a basis to determine the reliability growth plan. The growth plan generally includes the number of test samples, the test time, the team that performs fault analysis and defect improvement after testing product failures, and the selection of fault improvement decisions.

[0055] What needs to be quantified in the reliability growth test process of the present invention are the initial MTBF0, the total product test time T and the reliability growth rate m. The relationship between these three parameters and the expected MTBF index is expressed by the Duane model formula as follows:

[0056]

[0057] Among them, MTBF is the instantaneous MTBF of the product after the test, MTBF0 is the initial MTBF, T1 is the test time for testing the initial MTBF0, T is the total test time of the product (including T1), and m is the reliability growth rate of the product; wherein, MTBF0 is determined based on the historical MTBF of existing products, such as selecting the average value of multiple historical MTBFs of a similar product as the MTBF0 of the tested product. It can also be set based on actual needs or actual experience, and there is no limitation here.

[0058] Since MTBF0 is a pre-set parameter during the test phase and cannot be changed, and the test time T1 has little impact on the initial MTBF0 of the test product, the parameters that determine whether the test product can achieve the expected MTBF index are the total product test time T and the reliability growth rate m. Usually, companies evaluate the value of the reliability growth rate m based on existing reliability growth experience or self-evaluation, and then calculate the total test time T (number of prototypes * calendar time) required for the product based on the expected MTBF index, thereby formulating a plan for product reliability growth.

[0059] It can be seen from the Duane model that when the initial MTBF0 and the expected MTBF index are determined, the larger the value of the reliability growth rate m, the less total test time T required for the product to reach the expected MTBF. Taking the initial MTBF0 as 100h as an example, the test time T1 to test the initial MTBF0 is 200h. If the expected MTBF index is 300h, when the reliability growth rate m = 20%, the total test time T required is 15925h; when m = 50%, the total test time T required is 450h.

[0060] According to the description in Section 5.6.2 of the reference "GJB 1407-1992 Reliability Growth Test", in the field of reliability growth related technology, it is generally believed in the industry that the value of the reliability growth rate m is related to factors such as the enterprise's development plan, funding and technical level. The value of the reliability growth rate m generally ranges from 0.3 to 0.6. If the value of the reliability growth rate m of a product is between 0.1 and 0.3, it means that the optimization measures for the product are generally effective. If the value of m is between 0.6 and 0.7, it means that the enterprise has taken efficient fault analysis and corrective measures for the product. This is exactly the problem with the existing technology: since the value of M is difficult to quantify, when testing the product, the M value is usually set based on work experience (for example, based on the generally believed range of the reliability growth rate m), but this is likely to result in the actual product reliability test time required in the later stage being much longer than the planned test time in the reliability growth plan initially formulated based on the reliability growth rate m, which may lead to a series of problems such as the project being unable to be delivered in time.

[0061] Therefore, in view of the above problems, the present invention proposes a simulation method for product reliability growth, which performs mathematical simulation on the product reliability growth process. Through simulation, the product reliability growth process can be simulated to achieve effective observation of the product reliability growth process. In the test process of mathematical simulation, by adjusting the working parameters of the product, it is possible to accurately predict whether the product reaches the expected MTBF index, and then a more accurate and reliable product reliability growth plan can be formulated, so that the product reliability growth results in actual applications are closer to the requirements of the expected MTBF index.

[0062] When actually conducting reliability growth tests, a cycle of "fault occurrence-fault improvement-testing" will occur, such as Figure 2 As shown. The basic principle of reliability growth is to induce product failure through product testing, analyze the failure and find out the cause of the failure, redesign and optimize according to the cause of the failure, and then test again. On the one hand, the effectiveness of the failure improvement is verified, and on the other hand, the potential failures of the product are continuously tested. Therefore, the reliability growth of the product is the result of an iterative design process.

[0063] In the process of constantly discovering product defects and improving them, product defects are becoming fewer and fewer, and the interval between failures in tests is getting longer and longer. That is, the instantaneous failure rate of the product gradually decreases as the reliability increases and the test time increases. Figure 3 shown.

[0064] There are many methods for predicting how long a product needs to be tested to reach the expected MTBF index during the process of product reliability growth. Common methods include prediction based on product production experience and usage experience, or prediction based on external experience such as historical product failure data, information on similar products, etc., or prediction by testing products in the research and development and manufacturing processes, etc. The method adopted in the present invention is to predict whether the product can reach the expected MTBF index by testing the product in a mathematical simulation manner.

[0065] Specifically, the present invention quantifies the defect characteristics and fault analysis process of the product, and uses mathematical simulation methods to simulate the cycle of "fault occurrence-fault improvement-testing" in the reliability growth process. Figure 4 As shown; the following will specifically explain the simulation method for the reliability growth of this product:

[0066] Step 1: Quantify the fault characteristics of the product to be tested, including the following quantitative factors: number of fault categories, number of fault causes, proportion of each fault category, proportion of each fault cause, and express the concentration of fault categories in the form of statistical distribution;

[0067] Step 1 specifically includes the following sub-steps:

[0068] Step 1.1: Determine the initial MTBF0 of the product to be tested, and calculate the initial failure rate λ0 of the product based on the formula: λ=1 / MTBF; the formula λ=1 / MTBF indicates that the failure rate of the product and the mean time between failures (MTBF) of the product are reciprocal. Specifically, substituting λ0 into the above formula, we get λ0=1 / MTBF0;

[0069] Step 1.2: Determine the number of fault categories of the whole product to be tested: The fault category is a classification of product defects. Since the fault category can be divided based on different standards, the number of fault categories can be the number of fault codes, or the number of faults observed and recorded from other angles (such as classifying defects by discipline (electronics, structure, software)), or the number divided by fault mechanism (such as overvoltage, overcurrent, timing error, control deviation and other fault mechanisms, or any other reasonable classification method for product defects;

[0070] Step 1.3: Determine the proportion of each fault category in the total machine failure rate: The present invention preferably uses the statistical method of Weibull distribution to generate the proportion of each fault category. The smaller the shape parameter β of the Weibull distribution is, the more concentrated the faults are. In addition to using the Weibull distribution to generate the probability ratio, the statistical distribution type of the frequency of occurrence of each fault category of similar models can also be counted, and the corresponding random probability ratio can be generated using the statistical distribution type. In addition, in addition to using the statistical method of Weibull distribution, other statistical methods such as uniform distribution and normal distribution can also be used.

[0071] Step 1.3 also includes the following sub-steps:

[0072] Step 1.3.1: Generate a set of random numbers that satisfy the Weibull distribution with a shape parameter of β. The number of random numbers is the number of fault categories.

[0073] Step 1.3.2: Normalize this set of random numbers. The sum of the probabilities after normalization is 1. Each probability corresponds to the proportion of the failure rate of each fault category.

[0074] For example: Assume that a product has 100 fault codes (that is, the number of fault categories is 100), and the shape parameter β is 0.5. If 10,000 faults are recorded continuously, some fault codes appear multiple times, while some fault codes do not appear. Among them, the number of times the top 20% of fault codes with the highest failure rate appear accounts for about 78% of the total number of faults, which basically meets the "80 / 20 principle".

[0075] The proportion of the top 20% fault categories to the total failure rate under different shape parameters β is shown in the following table:

[0076]

[0077] Step 1.4: Determine the number of potential defects under the fault category of each product: Since not all components can report the reminder of "XX component failure" after the failure of the product, in industrial products (such as photovoltaic inverters), it is generally a sensor or control node that reports the fault after multiple components fail. The present invention uses a method of generating random numbers to simulate the number of potential defects under the fault category of a certain product, and this random number conforms to the statistical distribution. Specifically, the number of potential defects conforms to the uniform distribution or Weibull distribution or other statistical distribution. The distribution is related to the logic of product sensor layout and fault reporting. For example, after defects 1, 2, 3, and 4 appear, fault A will be reported, and after defects 6, 7, 8, 9, and 10 appear, fault B will be reported, and so on. The number of defects corresponding to faults [A, B, ...] is [4, 5, ...], and the distribution type satisfied by the defect sequence [4, 5, ...] is the distribution type satisfied by the number of potential defects. The present invention preferentially uses Weibull distribution to characterize the distribution type of the number of potential defects.

[0078] Step 1.5: Determine the proportion of the failure rate of each potential defect to the failure rate of its corresponding fault category: This step uses the statistical method of step 1.3 to determine the failure rate proportion of each potential defect. For example, if the failure rate of a fault category is 0.1 and the number of defects in this fault category is 5, then the failure rate of each defect needs to be determined. Use step 1.3.1 to generate 5 values ​​[3, 6, 4, 8, 6], which are normalized to [11%, 22%, 15%, 30%, 22%]. The failure rates of the 5 defects under this fault category are [0.011, 0.022, 0.015, 0.03, 0.022].

[0079] Step 2: Based on the triggering conditions of the fault analysis, determine whether the product under test needs to be analyzed for faults, and determine whether the requirements for fault improvement are met;

[0080] Step 2 includes the following sub-steps:

[0081] Step 2.1: Determine whether to perform fault analysis: Many faults may occur during the product reliability growth test. Sometimes the company decides to improve all faults that occur, and sometimes only improves high-frequency faults (for example, some companies believe that a fault that only occurs once is an accidental event and does not need to be improved. When a certain defect accumulates and occurs many times, it is determined that it may be a systemic fault and needs to be analyzed and improved). The triggering conditions of fault analysis have a great impact on the subsequent product reliability growth test process. Therefore, the present invention first determines whether the fault of the product to be tested meets the triggering conditions of fault analysis;

[0082] Specifically, the triggering conditions include but are not limited to: performing fault analysis immediately when a fault occurs, or performing fault analysis when the cumulative number of fault occurrences reaches a preset number. In this embodiment, the preset number is preferably 1, however, in the actual product reliability growth test process, the triggering conditions for fault analysis are more than the above two, and the preset number can also be determined according to the actual situation.

[0083] Step 2.2: Whether the real cause of the fault has been analyzed. Not every fault analysis can find the real cause. Therefore, the present invention uses the failure to reach a certain expected optimization threshold (for example, 50%) to analyze the real cause as the standard for judging whether the cause of the fault has been analyzed. For example, some faults due to electromagnetic compatibility or short-term electric shocks disappear naturally after the product is restarted, and the fault cannot be reproduced or analyzed. The cause of the fault; some electric shocks may cause multiple local components to burn, and it is difficult to determine the initial cause of the burn. Therefore, quantifying the proportion of whether the real cause of the fault is found can be more in line with the actual workflow and improve the effect of product reliability growth testing in actual applications.

[0084] Step 2.3: Determine whether the requirements for fault improvement are met. There is an opportunity for improvement only when the real cause is found. However, not all faults meet the requirements for fault improvement after the real cause is found. For example, some fault causes are found but there is no suitable solution, and some fault improvement solutions require a lot of cost. In this case, the company will consider whether to improve the fault as appropriate. The present invention requires that when the real cause of the product failure is found, first consider whether there is a corresponding improvement plan for the fault. If there is a corresponding improvement plan, consider whether the improvement plan meets the cost requirements. If so, make subsequent improvements to the fault.

[0085] Step 2.4: Determine the improvement ratio. In actual work, improvement measures cannot always completely eliminate defects, but only reduce defects to a sufficiently low level. The present invention takes the improvement ratio of defects as a requirement, that is, in the case of improvement measures, the failure rate of the product should be reduced by a preset ratio, that is, the failure rate of the product parts after the defect improvement should be less than or equal to the failure rate of the product parts before the defect improvement multiplied by the preset ratio. For example: For example, the original failure rate of a capacitor is 0.0001. During the test process, the goal is to reduce the existing failure rate of this capacitor by 80%. After optimization, the failure rate of this capacitor is 0.0001*(1-80%)=0.00002. How to determine whether the optimization measures can reduce the original failure rate by 80% is not within the scope of the present invention, but is to be determined based on the specific application of a specific device.

[0086] Step 2.5: Implement improvement measures. If there are improvement measures, determine whether to adopt a full improvement strategy or a batch improvement strategy for the tested product. Different improvement strategies have a certain degree of influence on the product reliability growth process. If improvements are made early, the downtime and failure analysis time will be reduced, and more time will be used for testing.

[0087] After steps 2.1 to 2.5, the preparation stage of the product reliability test has been completed. Before conducting the simulation test of product reliability, in addition to the above steps, the following steps are also included:

[0088] Step 2.6: Determine the number of prototypes: Determine the number of prototypes used for the reliability growth test. Theoretically, the more prototypes there are, the more defects are observed in the same period of time. In this embodiment, the number of prototypes for the reliability growth test is set according to actual needs and is not limited here.

[0089] Step 2.7: Determine the test time: Determine the time when the reliability growth test can be performed. The test time meets a preset time range. The maximum value of the time range is the total number of days from the current test time to the day before the product is officially mass-produced. The minimum value of the time range satisfies the formula: N*M*24 / MTBF0≥S, where N is the total number of days of the minimum value of the time range, M is the number of samples of the tested product, MTBF0 is the initial MTBF of the whole product to be tested, and S is a threshold determined based on the test risk level. The value of S can be determined based on actual experience. The larger the value of S, the lower the reliability growth test risk of the product.

[0090] Step 2.8: Set the probability threshold: the probability that the reliability growth test meets the expected MTBF value requirements. The reliability growth process is random, and not every reliability growth can meet the expected MTBF value. Therefore, it is necessary to set a probability threshold to represent the project team's requirements for the probability of meeting the expected MTBF value after the reliability growth test, such as 80%; the value of the probability threshold is based on the project requirements of the project where the test product is located, or is set for other practical considerations, and no specific restrictions are made here.

[0091] Step 3: Based on the quantitative factors of the fault characteristics of the product to be tested and the judgment on whether fault analysis and fault improvement are needed, a mathematical simulation test is conducted on the reliability growth of the product to be tested.

[0092] When the product parameters and working parameters are determined, a simulation test of product reliability growth is carried out. Figure 5 The simulation steps are as follows:

[0093] Step 3.1: Fault Occurrence: Random faults occur in the entire product, and the fault interval, fault category code, and potentially defective parts meet the set values ​​in steps 1.1 to 1.5;

[0094] Step 3.2: Fault handling: Decide whether to conduct fault analysis based on the conditions that trigger fault analysis, and determine whether the product meets the requirements for fault improvement; if the product meets the requirements for fault improvement, determine that the failure rate of the product must be reduced to at least the preset threshold, and implement the final improvement strategy for the product;

[0095] Step 3.3: Update failure rate: After a fault occurs and is analyzed and improved, the failure rate corresponding to the product, fault type, and fault cause will change. The specific changes are determined based on the situation simulated in step 3.2.

[0096] Step 3.4: Repeat steps 3.1-3.3. If the calendar time reaches the test time determined in step 2.7, terminate the simulation;

[0097] Step 3.5: Display reliability growth results: Dynamically display the cumulative operating time, instantaneous failure rate, and instantaneous MTBF of each product failure recorded in the reliability growth mathematical simulation test, and determine whether the instantaneous MTBF of the product can be improved to the expected instantaneous MTBF of the product after the test within the expected time.

[0098] 3.6: Repeat simulation: Conduct multiple physical simulation tests, take the results of several simulation tests for probability statistics, and compare them with the probability threshold set in step 2.8 to determine whether the expected MTBF is achieved; if the statistical MTBF probability value is less than the set probability threshold, it is necessary to adjust the simulation-related parameters and repeat steps 3.1 to 3.5 until the expected MTBF is achieved. Among them, the results of several simulation tests are taken for probability statistics and compared with the set probability threshold, specifically: taking the expected MTBF index as the benchmark, the ratio of the number of tests in which the MTBF obtained in several simulation tests is greater than or equal to the expected MTBF index to the total number of simulation tests is taken as the probability statistics value of the simulation test, and the probability statistics value is compared with the probability threshold set in step 2.8. If it is greater than or equal to the set probability threshold, it means that the possibility of the simulation test results reaching the expected MTBF has met the requirements. Further, the parameters involved in the simulation test that reaches the expected MTBF are used as the basis for formulating the product reliability growth plan.

[0099] The following uses a certain inverter as an example to illustrate the above method:

[0100] The initial MTBF0 of a certain inverter is evaluated as 100h, and it is expected that the MTBF will be increased to 200h through reliability growth testing.

[0101]

[0102] This case assumes that steps 7-11 in the table above take a very short time and the working time is simplified to 0h.

[0103] By performing mathematical simulation according to the set parameters, the number of fault categories and the instantaneous MTBF of the product can be simulated within the expected reliability growth test time.

[0104] The simulation simulates the process in which multiple prototypes fail in reliability growth tests, the MTBF increases after improved fault analysis, and the failure interval becomes longer.

[0105] Figure 6 It is a trend chart of the cumulative test hours and the cumulative number of failures in each of the 50 simulations.

[0106] Figure 7 This is a trend chart of the cumulative number of failures and instantaneous MTBF in each of the 50 simulations.

[0107] Figure 8 The chart shows the instantaneous MTBF trend corresponding to the calendar days (test days) in 50 simulations. The chart indicates that the probability that the MTBF in multiple simulations can meet the MTBF requirement at the end of the reliability growth test is 55%, that is, when the reliability growth test is actually carried out, the probability that the growth test can reach the MTBF requirement is only 55% when the working capacity in the case is met. If the probability is very low, the number of prototypes or the number of test days can be adjusted or the requirements for failure analysis can be increased.

[0108] If the test days are extended from 50 days to 100 days, 93% of the simulation results in multiple simulations can meet the MTBF requirement. Fig. 9 shown.

[0109] The above-mentioned embodiment only expresses one implementation mode of the present invention, and its description is relatively specific and detailed, but it cannot be understood as limiting the scope of the invention patent. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be based on the attached claims.

Claims

1. A simulation method for product reliability growth, characterized in that: Taking product parameters and working parameters as the main factors affecting the process of formulating product reliability growth plan, the product reliability growth process is simulated by mathematical simulation test; wherein, product parameters are inherent attributes of products; working parameters are factors affecting the change of product MTBF, and whether the product to be tested reaches the expected MTBF index is judged by adjusting the working parameters; the specific steps of the simulation method of product reliability growth are as follows: Step 1: Quantify the fault characteristics of the product to be tested, including the following quantitative factors: number of fault categories, number of fault causes, proportion of each fault category, proportion of each fault cause, and use statistical methods to express the concentration of fault categories; Step 2: Based on the triggering conditions of the fault analysis, determine whether the product under test needs to be analyzed for faults, and determine whether the requirements for fault improvement are met; Step 3: Based on the quantitative factors of the fault characteristics of the product to be tested and the judgment on whether fault analysis and fault improvement are needed, a mathematical simulation test is conducted on the reliability growth of the product to be tested; Step 1 includes the following sub-steps: Step 1.1: Determine the initial MTBF0 of the whole product to be tested, and calculate the initial failure rate λ0 of the whole product based on the formula: λ=1 / MTBF, where λ is the failure rate of the product to be tested, and MTBF is the initial mean time between failures of the whole product to be tested; Step 1.2: Determine the number of fault categories of the whole product to be tested; Step 1.3: Determine the proportion of each fault category to the total failure rate of the product: Based on statistical methods, determine the proportion of the number of failures of each fault category to the total failure rate of the product. By counting the frequency of each fault category, the shape parameter β Express the concentration of fault categories; Step 1.4: Determine the number of potential defects under the fault category of each product: Generate random numbers to simulate the number of potential defects under the fault category of a product; Step 1.5: Determine the proportion of the failure rate of each potential defect to the failure rate of its corresponding fault category; Step 2 specifically includes the following sub-steps: Step 2.1: Determine whether to perform fault analysis: Determine whether the triggering conditions for fault analysis are met. If the triggering conditions for fault analysis are met, start the fault analysis process; Step 2.2: Whether the fault cause is analyzed: Determine whether the probability of analyzing the real fault cause at least reaches the expected threshold; Step 2.3: Determine whether the requirements for fault improvement are met: After analyzing the real cause of the fault, consider whether there is a corresponding improvement plan for the fault and whether the improvement plan meets the cost requirements; Step 2.4: Determine the improvement ratio: reduce the failure rate of each component with potential defects to a preset ratio as the determined improvement ratio; Step 2.5: Implement improvement strategy: adopt all improvement strategies or batch improvement strategies for the tested products; Before step 3, the following steps are also included: Step 2.6: Determine the number of prototypes: Determine the number of prototypes used for reliability growth testing; Step 2.7: Determine the test time: determine the time when the reliability growth test can be performed, and the test time meets a preset time range; Step 2.8: Set probability threshold: Before the mathematical simulation test of reliability growth, determine a probability threshold to indicate that the mathematical simulation test meets the minimum requirement of the expected MTBF; Step 3 specifically includes the following sub-steps: Step 3.1: Fault occurrence: Random faults occur in the product, and the fault interval, fault category code, and failure rate of each potential defect meet the set values ​​in step 1; Step 3.2: Fault handling: Decide whether to conduct fault analysis on the product based on the conditions that trigger the fault analysis, and determine whether the product meets the requirements for fault improvement; if the product meets the requirements for fault improvement, implement the final improvement strategy for the product; Step 3.3: Update failure rate: After implementing the improvement strategy, update the failure rate of the entire product, fault type, and fault cause; Step 3.4: Repeat steps 3.1-3.

3. If the calendar time reaches the test time determined in step 2.7, terminate the simulation; Step 3.5: Display reliability growth results: Dynamically display the cumulative operating time, instantaneous failure rate, and instantaneous MTBF of each product unit failure recorded in the reliability growth mathematical simulation test, and determine whether the instantaneous MTBF of the product unit can be increased to the expected instantaneous MTBF of the product unit after the test within the expected time; Step 3.6: Repeat simulation: Perform multiple mathematical simulation tests, take the results of several simulation tests for probability statistics, and compare them with the probability threshold set in step 2.8 to determine whether the expected MTBF is achieved; if the statistical MTBF probability value is less than the set probability threshold, it is necessary to adjust the simulation related parameters and repeat steps 3.1 to 3.5 until the expected MTBF can be achieved.

2. A product reliability growth simulation method according to claim 1, characterized in that: Step 1.3 specifically includes the following sub-steps: Step 1.3.1: Generate a set of random numbers that satisfy the Weibull distribution with a shape parameter of β. The number of random numbers is the number of fault categories. Step 1.3.2: Normalize this set of random numbers. The sum of the probabilities after normalization is 1. Each probability corresponds to the proportion of the failure rate of each fault category.

3. The method for simulating product reliability growth according to claim 1, characterized in that: The statistical distribution in step 1 is any one of Weibull distribution, uniform distribution and normal distribution.

4. The method for simulating product reliability growth according to claim 1, characterized in that: The conditions for triggering fault analysis include: performing fault analysis immediately when a fault occurs, or performing fault analysis when the cumulative number of fault occurrences reaches a preset number.

5. The method for simulating product reliability growth according to claim 1, characterized in that: In step 2.7, the maximum value of the time range is the total number of days from the current test time to the day before the product is officially mass-produced; the minimum value of the time range satisfies the formula: N*M*24 / MTBF0≥S, where N is the total number of days of the minimum value of the time range, M is the number of samples of the tested product, MTBF0 is the initial MTBF of the whole product to be tested, and S is a threshold determined based on the test risk level.

6. The method for simulating product reliability growth according to claim 1, characterized in that: In step 3.6, the results of several simulation tests are taken for probability statistics and compared with the probability threshold set in step 2.

8. Specifically, taking the expected MTBF index as a benchmark, the ratio of the number of tests in which the MTBF obtained in several simulation tests is greater than or equal to the expected MTBF index to the total number of simulation tests is taken as the probability statistical value of the simulation test, and the probability statistical value is compared with the probability threshold set in step 2.

8. If it is greater than or equal to the set probability threshold, it means that the possibility of the simulation test results reaching the expected MTBF has met the requirements.

Images