[Zhu Rong], who represents the latest achievements of China’s aerospace industry, has successfully landed on Mars and began to perform his own Mars exploration mission. With the development of science and technology, the continuous success of various complex engineering systems represented by spacecraft is truly breathtaking.
However, according to the current conventional practice, if we calculate the failure rates of these complex engineering systems before they are implemented, very high results will be obtained; if we really make decisions based on the calculated failure rates, it is estimated Mankind will not see these miracles of technological progress.
For example, if we want to predict the reliability of [Zhu Rong]’s electronic products, in accordance with the current industry specifications, we generally use the US military standard 217F issued in the 1990s and the Chinese military standard 299C issued in 2006, and use these standards Provide the failure rate data and the corresponding prediction model to calculate the failure rate of [Zhu Rong].
The general steps given in this type of standard to predict the reliability of an electronic product are: First, find the basic failure rate calculation model of each type of component in the standard, and calculate a basic failure rate data according to this model; Secondly, considering the manufacturing and use conditions of this component, select the corresponding correction factor (such as manufacturing quality level factor and environmental condition factor) to correct the basic failure rate, and obtain a work failure rate data; finally, the value of all electronic components Add the work failure rate data to get the failure rate data of this electronic product.
Every engineer who does reliability prediction work for the first time will complete the above steps very carefully and religiously. These steps are a time-consuming and laborious task for slightly more complicated electronic products, whether you choose software tools or not. However, what makes engineers even more desperate is that the high failure rate of electronic products calculated in this way makes them doubt life. When faced with doubtful eyes from leaders and customers, I can only say: According to the manual, I did not make a mistake, and this is the result.
What are the consequences of this incredible expected result? Even those who count don’t believe it, and those who see it don’t believe it either. Reliability prediction, as an important part of evaluating product design schemes and supporting R&D decisions, loses professional authority and is skipped or ignored by decision makers.
The reliability prediction work is skipped directly by the decision-maker, which is a relief for the engineer who holds the manual to calculate the calculation; if this step cannot be omitted according to the development procedures, it is necessary to calculate a result and issue a report. The engineer who calculates in the manual must have 10,000 grass-and-mud horses galloping… (10,000 words are omitted here).
Every time I think of the above scenes, it really makes me, a so-called reliability professional, feel embarrassed, either because of insomnia or waking up in the middle of the night. I typed this text when I was awake in the middle of the night.
Why did it cause such a situation? There are nothing more than the following reasons.
One is that the basic failure rate prediction model for each type of electronic components provided in the manual is not updated in time. The development of components changes with each passing day. New technologies and new processes continue to emerge. The models in the manual cannot be updated in a timely and effective manner. Some new components cannot even find the expected models. The above-mentioned 217F manual has not been updated for more than 20 years, and it is said that it is not planned to be updated; the 299C manual has been used for 15 years, and it is said that there is a 299D version, which has not been announced for various reasons.
Second, the correction factor for calculating the failure rate given in the manual is not considered comprehensively. The failure rate of a component depends on two factors. On the one hand, it is related to the R&D and manufacturing capabilities of the component manufacturer. For example, the same type of electronic components are designed and produced by different manufacturers. Obviously, the failure rate should not be The same; on the other hand, the failure rate of components is related to the R&D and manufacturing capabilities of the companies that select the components. The failure rates of the same components in different companies are different. This difference is not considered much in manuals such as 217F and 299C.
The third is that the calculation of the failure rate of the components used in the manual to directly add the failure rate of the product is unscientific. The underlying mathematical logic of such an algorithm is the probability product theorem, and the premise of the probability product theorem is to multiply the probability of occurrence of independent events. In engineering practice, such an assumption does not exist, that is, each component is connected through the circuit board to form the whole of the interconnection of each component-electronic product. This connection is not only a functional connection, but also a failure connection. . But the influence of this connection on the overall failure rate of electronic products is directly ignored by the “addition” algorithm of component failure rates!
For the first question, my team designed a shared business model to attract and incentivize component manufacturers or third-party organizations to directly provide basic failure rate data for components. This business model will be launched soon;
for the second question , My team is about to launch a new component studio efficiency algorithm and supporting software tools. This algorithm can not only consider the impact of component manufacturers’ R&D and manufacturing capabilities on the failure rate of components, but also consider components The impact of the R&D and manufacturing capabilities of integrators on the failure rate;
for the third question, my team is studying a new algorithm that considers the failure rate of electronic products that are associated with failures.
The basic theory to solve the above three problems is to believe in reliability theory, which is a new set of reliability scientific methodology blessed by system science and mathematics.
