Portable DAS uses AI to monitor pipeline leaks and blockages.

Subtitles in the video

Hello everyone, we will now proceed with this application:

Distributed Fiber Optic Vibration Sensing

for pipeline leakage monitoring.

First,

our experimental setup is as follows:

From this side,

we have a 20mm pipe

supplying water to a 50mm PVC pipe.

This pipeline consists of two sections

with a connection point in the middle.

There is a control valve here,

currently in the open position.

When open,

water flows freely through the pipe

once supplied.

However, when closed,

due to a crack present here,

water will seep out

through this minute fissure.

At this point, we need to use this fiber optic cable

to perform sensing

and determine if we can detect

the leakage state.

Our trained AI model

is used for this discrimination experiment.

Alright, currently

since no water is flowing,

our pipe is in this

schematic diagram shows a gray state—

gray indicates

a waterless condition.

Alright.

Now, as we open this valve,

let's first attempt to supply water. Please supply water.

Alright, at this moment,

we observe

it switches to a flowing water state.

Yes.

At this point,

water can naturally flow through

the pipe itself.

Alright.

Now, please close this valve.

Gradually, over here,

we'll hear some sounds—

water—

due to water pressure—

slowly

seeping out

from this gap.

At this moment, we can also observe

our AI model can accurately

and reliably identify

that it's now in a

leaking

state.

Once it stabilizes,

it's definitely in this

leaking

state.

Ah, at this point,

we can then trigger

the necessary alarms as needed.

And we've discovered that, um,

Although this pipeline isn't very long—just three or four meters—

it falls within

the theoretical resolution range

of this DAS system.

Nevertheless, we can clearly distinguish

whether water is flowing

or leaking.

At this point,

we'll restore it to its natural flow state.

Please open this valve.

Alright.

Then, in this manner,

the water will gradually

After the pressure release earlier,

the water can now flow naturally.

We can wait a moment

until it reaches a steady state.

Now,

as the leakage gradually disappears,

it has indeed switched back to

a normal flow state.

At this point, we observe

that since this crack is oriented slightly upward,

the leakage has ceased.

Then, using the opposite pipeline

to distinguish the status.

Currently, it detects a

normal flow

state.

Mm-hmm.

Alright.

It has stabilized in this correct

monitoring state.

Now we'll turn off the water.

Please shut off the water.

Alright. We may need to

wait a moment

for the water in the well to drain completely.

Mm-hmm.

Alright.

Once it enters a steady state,

it will be in this waterless

condition.

At this point, it means there is no water left.

Overall, we use this distributed fiber optic

intelligent sensing system

to demonstrate its ability to distinguish

between different pipeline states.

Yes.

So the structure of this AI model

remains applicable in such scenarios.

Alright, thank you all.