Reliable forecasts

If only we really had a crystal ball or some other tool to predict the future. Because of the general surprise at the opponent’s next move, the soldier hopes that his equipment will work. From a company board anticipating upcoming moves to a maintenance manager ordering spare parts, we have many applications for knowing the future.

We often look to past successes to predict the future. Does this mutual fund generate enough income on a regular basis? If so, we predict it will move forward. And anyone who has reviewed mutual fund performance has read or heard the advice not to use past returns to predict future returns. Mutual funds, markets, companies and battlefields change and sometimes react unpredictably.

Of course, we often need to anticipate future

 Situations and possible outcomes before making decisions. Whether we’re investing, ordering spare parts, or planning production, we use future projections to determine the right performance.

When I was a young and new reliability engineer at headquarters, I got a call from a senior reliability engineer in the division asking if I could do part-time forecasting for one of their products. In particular, the areas. I said yes, despite what I hadn’t done yet, and I didn’t know what the partial guess was or how useful it was. I just received a demo copy of the Relax Prediction Module this week, and this project is a great way to learn both part-time forecasting and the software.

I soon discovered that the BOM and failure rate databases were used to estimate the base component count to calculate the expected PCB failure rate. The failure frequency of the multilayer ceramic capacitor is 5 FIT (failure after 109 hours) and the analog ASIC is listed at 450 FIT. The software helped match the components to their failure rate and performed calculations that resulted in a final estimate of the customers’ expected product failure.

The forecast took about 2 hours,

 half or more of which was spent learning the software. With no information other than the bill of materials, all settings in the forecasting software are default settings, temperature rating, collapse, quality level, and so on. Questions about prophecies It’s magic. Put it in the parts list and after a few milliseconds of calculation we already know the future. Or are we?

My original research assumed that many of our products failed because of power supplies, connectors, and fans. The prediction results listed the power supplies and connectors for the five expected failures, and the system had no fan, so it was almost accurate. More complex components are expected to fail more often or earlier than a simpler component.

Where do the failure rates in the table come from?

 Become people know enough values ​​to list. After a bit of reading and calling around, I discovered that the become team collects hacking data from various sources from time to time, including GIDEP and major talks. They sort and analyze the data and make historical models of failure rates that include the effects of temperature, decline, quality and more. The devices they studied were mainly used in military and telecommunications infrastructure. The boxes are usually made of printed circuit boards.

The electronics industry has changed a lot in five years, but it’s obviously going to be hard for us to beat the become team if we don’t solve every problem carefully down to the component level and know the operating conditions. The product I  토토사이트 is not quite comparable to the products in the telecom industry, but it still looks close.

Leave a Reply

Your email address will not be published.