S-300MPU and S-300 V videos!!!

PiSigma said:

the early patriots suck due to an numerical error in its calculations:

" A report of the General Accounting office, GAO/IMTEC-92-26, entitled Patriot Missile Defense: Software Problem Led to System Failure at Dhahran, Saudi Arabia reported on the cause of the failure. The Patriot missile failures in Desert Storm were due to roundoff error because the system was designed to be on alert for only a few minutes at a time. Instead, they ran continuously for days. The clock, which was storing time in floating-point, began to have too many digits to the left of the decimal place and not enough after the decimal place to do accurate timings of incoming SCUD velocities."

now don't all of you just want to take Numerical Methods for Engineering??

Click to expand...

This is the most documented one. It's actually taught in every Software Engineering class. In reality, most of PMD encounters with Scud actually went according to the plan. The only problem is that PMD did not destroy the scud.

hey tphuang, are you a comp E??? there doesn't seem to be too many engineers in this forum... we had a bonus project for my numerical methods class to correct the error found in patiot, of course, not use the actual code they had, but something my prof made, wow, that was a barrel of fun. just thinking about it make me want to cry, but i needed the marks.

I'm actually a Software Engineer, that's why I got interested in the Patriot missile and control software of planes. The particular incident you listed is really quite an extraordinary circumstance. It tells me that the people who were designing this software really had no idea of how numbers are stored. Anyone with a half brain can figure out that it's dangerous trying to accumulate fractions that can't be represented exactly by the computer hardware.

BrotherofSnake said:

The S-300 is quite impressive.

The Patriot was the first air defense system in the world to successfully intercept a theater ballistic missile.

Click to expand...

Too bad the Patriot can't shoot down an obsolete Seersucker that hit Kuwait City getting past AWACS, Aegis, and Patriot sensors without being seen.

AssassinsMace said:

Too bad the Patriot can't shoot down an obsolete Seersucker that hit Kuwait City getting past AWACS, Aegis, and Patriot sensors without being seen.

Click to expand...

The last one in 2003, right?

to tphuang
o cool, i'm just glad numerical methods is required learning for most engineering disciplines now. the designers must be pretty dumb to forget something so important. especially when it's the basics.
for those of you that don't know what we are talking about. accumulating fractions by computer system is saying that instead of keeping all decimals in the computer memory( because sometimes you can't), the computer round off or cut off numbers it deemed insignificant, usually after several decimal places. and after awhile, the error builds on itself. it's like building a building, and the materials used are always 1 mm shorter than specified, by the end you get a building that's 2 meters shorter than you wanted, and the people that build it don't know why.

PiSigma said:

for those of you that don't know what we are talking about. accumulating fractions by computer system is saying that instead of keeping all decimals in the computer memory( because sometimes you can't), the computer round off or cut off numbers it deemed insignificant, usually after several decimal places. and after awhile, the error builds on itself. it's like building a building, and the materials used are always 1 mm shorter than specified, by the end you get a building that's 2 meters shorter than you wanted, and the people that build it don't know why.

Click to expand...

Is it a viable option to have an algorithm which would monitor the data and then try to guess the figures that have been cut off? Then periodically just add that guesswork sum to the data, trying to make it more accurate. Would that require more memory/cpu processing than not cutting off and rounding the figures?

Totoro said:

Is it a viable option to have an algorithm which would monitor the data and then try to guess the figures that have been cut off? Then periodically just add that guesswork sum to the data, trying to make it more accurate. Would that require more memory/cpu processing than not cutting off and rounding the figures?

Click to expand...

that's what is suppose to happen. Even the computer clocks are not exactly on time, so you have to continually synchronize and make up the differences.