Low Speed Radar/Train Speed Detection

What about some sort of rotary encoder or similar mounted on the train to the wheels (or similar) and transmitting the speed wirelessly to a receiver?

This sort of onboard measurement system would appear to be ideal for such a low speed application.

The problem is that the trains are not owned by us and therefore we wouldn't have the ability to make any direct implementations

Heck, at those speeds, he could use a bucket (of water) dumper, or a tape measure, or sticks stuck in the ground with orange paint on them.

The trains that come through unload coal and so we have a set of hydraulic triggers that open the belly of the wagons so that the coal fulls into the hopper. Our control requires automation of locomotive/wagon detection so that the triggered can be automated to not only open when a wagon is over and close when it is not, but also control the flow rate of coal through the hopper and hopperlets on to the conveyor.

What about 2 sensors at a defined distance? the target is on the moving component . distance divided by time equals speed.

we already use such devices however travelling at such a slow speed means that it isn't as accurate as we would like it to be.

What sort of accuracy do you need? Maybe we should start there first and work backward.

down to 0km/h desired.

.4 needed

Think about it: 0 km/h will take infinite time to measure accurately, and so what practical minimum speed is acceptable? 0.4 km/h is the minimum speed of the radar he looked at, but he wants less. How much less and how accurately measured? To what resolution? To how many decimal places?

What we have here is

The need for a simple solution to what should be a non-problem.

We're talking instantaneous speed here so there should be no "infinite" measuring time.

At 0 km/hr there is. To wit:

Let's say we have a rotary encoder mounted to one of the wheels. Rotary encoders contain a slotted wheel which generates a sequence of pulses whose rate is a function of shaft speed and encoder resolution.

At 0 km/hr the pulse sequence will stop and the output will remain either high or low, but you won't know that it has stopped. All that you know is that the output has not yet changed state. For all you know, the wheel is still moving, just creeping along. But if it has stopped, you won't know it has stopped until you have waited an infinite time and noted that the output level is still the same. This is true no matter how fine the resolution.

This is a problem having nothing to with encoders; it doesn't matter how you measure it because it always comes down to the same problem: time versus accuracy, and it comes from a surprising corner: Information Theory: arbitrarily-low speeds (and frequencies) take a likewise arbitrarily long time to measure to a given accuracy. There is no getting around it. It is a fundamental property of the universe we live in. The converse is: the less time you take to make a speed (or frequency) measurement, the lower the accuracy.

So we're back to the question: what accuracy do you require?

Gutmonarch, you mentioned elsewhere that you actually enjoy seeing your posts marked Off-Topic, and so here's a suggestion: mark all of your comments Off-Topic. This gives everyone else a five-point head start in helping you meet your needs.

Win-win.

Amen!

Seriously?

Intelligent people with intelligent rational answers just scare the crap out of you that badly?

Is posting well thought out answers that much of a perceived insult or threat or insult to you its something you feel is ban worthy?

Even you must have been able to figure out who that was directed toward, unless you're just looking for any excuse to insult me . After all, you always remind us just how intelligent you are.

It seems CR4's code re-numbers comments (and their 'reply to' references) following a deleted comment, potentially leading to confusion and misunderstanding. This could be considered a functional bug in CR4's code. A workaround is for the admins/mods to leave the comment box intact but replace the offending content with "Deleted by Admin. See CR4 rules" or whatever.

-----

Note: When I make stupid comments I just cut to the chase and report myself to Admin with suggestions of banishment. This comment for example. A huge time-saver for all concerned.

Sorry, that was meant for the gutmag troll.

No worries.

Maybe that is gutmaggets goal, to cause dissension in the CR4 ranks.

I thought that was Lyns job.

Seriously? You mean, I have goals? Yehey, I am happy.

I sorta like the name "gutmag". Please disregard the troll would you?

_{I am just a guy from Kulas Polytechnic (whisper). oh man, would you ever change Lyn?}

Keep in mind the slower you go, the longer the measurement will take for a given accuracy, so perhaps we should start with the required accuracy and work backward.

Machine vision?

Here's a bit of information from the Society of Robots. They also have a forum , but they might have more roboticist than over here.. plus you can shop for parts there!

If you are dealing with more than one train / track / distance etc. This direction could give real time information on all movements as well as detect hopper movement.. people. out of place etc. I don't see any other way. (ok.. well I do)

I could be way off base, but it might be worth looking into

Just a shameless plug for Telemecanique OsiSense Vision sensors, easy to use, inexpensive, but only a 0.5+ meter sensing distance. Very clever programming features, like teaching it to see the leading edge of a railroad car wheel and ignore anything else that is not that curve. http://www.tesensors.com/us/en/product/vision/xuw-ref/

Vision and light sensing in a coal yard will require environmental conditioning (cleaning) to keep it reliable and operational...

This guy probably knows....

http://oppdthewire.com/nebraska-city-station-thaw-shed/

I would think there would be contacts on the track itself to sense the movement and speed and position of the cars....or infra red emitters and receptors....

http://www.raceamerica.com/index.shtml?http&&&www.raceamerica.com/gbeam.html

..." Technology choices for linear velocity sensors include cable extension, magnetic induction, microwave, optical or laser, piezoelectric, radar, or radio frequency, strain gauge, and ultrasonic sensors [1]. However, the use of SpeedMed-like devices, (i.e optical sensor timing gates) may provide a convenient and, most importantly, rapidly accessible source of information on velocity [2]. For these reasons, photocells timing systems are used routinely to measure speeds in athletic performance testing [3]."...

http://www.scielo.org.co/pdf/rinbi/v1n1/v1n1a07.pdf

Use a wheel detector. Surely you can devise a way to trigger the hopper controller at the right time.

https://www.frauscher.com/assets/media/Fachartikel/2006_SD_Vom_Schienenschalter_zum_Radsensor_EN.pdf

I know, right?

I'm wondering why such excruciating accuracy is needed for aligning railroad cars with a coal conveyor? Truly. It's not exactly LIGO we're speaking of here. Perhaps the OP could enlighten us as to the reasoning behind the need for such accuracy. For what it's worth, I have a hunch we're solving the wrong problem here. Just a hunch.

Because with a minimum value of 0.4k/hr that's 20% of the range we expect to see from the train. I'm not just measuring speed but also distance and carriage type. this 2k/hr range runs an algorithm through the PLCs to determine which trigger to activate in order to distribute coal appropriately across hopperletts. I can assure you that in the system I intend to use it, accuracy is imperative.

You could use doppler and ultrasonic schemes also.

I wonder what the increment is where the controller declares zero speed, they don't mention that in your link. That is one of the $9,000 dollar instruments we couldn't pay for with complaints from customers who had short rolls of paper (buy by length)

you have to use this in outdoor dusty environment (how the vibration of the cars would affect the results)

almost of the same is iR cam scene translating -- someone still has to verify the optical path has been not polluted

Well, isn't that instrument special (and expensive?)?

I would suggest that you measure "distance" using an optical device. (Simple laser distance devices are common.)

Your electricians (and the PLC) will be readily able to convert the distance signal into an instantaneous speed signal, with the added benefit that you also have indication of actual position relative to the hoppers.

You then have control over the absolute resolution through the combination of your distance accuracy and the clock speed used in the PLC.

Your trains seem to be going VERY slowly for anautomatic unload cycle. At those speeds (Say 1km/hr average), a 500m long train would take 30 minutes to pass. Nearby facility here handles trains at around 12km/hr in 100 wagon sets at typically 80T per wagon.

Most facilities here work using optical recognition for wagon filling to achieve balanced load profile. Lots of challenges in control theory.

The practical thing here actually is do you have the budget. I'll make you one. It can go simpliest as a proximity switch, target contacts on the wheels, pulse counter+converter and display.

If the most appropriate device cannot be found, make the PCB yourself and seal it for harsh conditions. As far as I have seen, this is what we do in the Railway Industry.

As others said, use some kind of laser sensor for the train detection, DSP for times calculation, convert it to an analog signal with the DAC, and feed it into an OPAMP to obtain your 4-20mA.

If you are not familiar with this type of designs, go to the electronics department and ask them to design it for you.

Commercial laser systems seem to only go down to 0.1meters per minute (Proton Products) looked at these for web speed measurement, pricy for paper mills, but much better than typical measures where the web slips on optical/inductive encoders on carry rolls.

Instead of calculating/predicting position based on speed, if your overall measure range is not excessive, why not digitally detect the car position, maybe inductively on the flange of the wheel? An array of sensors strategically placed will tell you speed, and when & where your product will be dropping. Use your radar sensor to locate the car of interest, then synchronize your wheel detectors to that signal.

I would go with a distance detector, ultrasonic perhaps, and calculate the speed based on change in absolute distance. The detector would need to be along side the track looking in at oncoming car ends and reading a distance over a range of a car length. Radar speed detectors use Doppler shift and at very low speeds, the ability to detect the shift falls off. A ranging detector gets better as the oncoming velocity goes down.

no optics to foul with an ultrasonic sensor, plenty of range, 30 meters for some.

( and cheap)

Put some load cells under the track, and logger the readings, as each axle rolls by, that is 1/2 car, unless they have four axles, which they do, so then you need it to record click-clack, click-clack, and that is one car in X time. Do that over and over, there is your train speed.

Various analyses of the waveforms will tell me plenty more, such as the weight load on each car.

Maybe even tell you if the coal is frozen in the car, no weight change...

That is why one wants to have the coal car tipper, that turns each car over and dumps it without disconnecting the train... the thing is a wonder to behold.

Have you found a solution for this?