Skip to Content

SiePortal

The integrated platform for your product selection, buying and support workflow - bringing together Industry Mall and Online Support.

SiePortalDeutsch

Search

Home
Support Support
{{(apiData.Support | filter:{'rank':1}:true)[0].title}} Support
- Menu temporary not available
- Menu temporary not available
- Menue temporary not available
- {{item.title}}

Products & Services
- Menu temporary not available
- Menue temporary not available
- {{item.title}}
Support Support
{{(apiData.Support | filter:{'rank':1}:true)[0].title}} Support
- Menu temporary not available
- Menue temporary not available
- {{item.title}}
MySiePortal MySiePortal
{{(apiData.MySiePortal | filter:{'rank':1}:true)[0].title}} MySiePortal
- Menu temporary not available
- Menue temporary not available
- {{item.title}}
Cart ({{apiData.CartCount}})

Navigation

Work on analog data to detect a hardware problem (process)

Created by: laboutvid at: 6/15/2023 12:01 PM (5 Replies)

Rating (0)

Thanks 0

6 Entries

6/15/2023 12:01 PM	Rate (0)
laboutvid Joined: 5/31/2023 Last visit: 1/10/2025 Posts: 3 Rating: (0)	Hello everyone First of all, sorry if I am not clear in my words, I am French and my english is not very good ! I turn to you because I have a need but I don't know exactly how to go about it... I explain : I have a hundred proportional valves on our machine (Glass bottle manufacturing), each of them have 2 analog signals 4-20mA, one is the setpoint (opening of a proportional valve), the other is the feedback position of the proportional valve . The proportional valves operate in cycles of approximately 10s. For each cycle, I must be able to compare feedback and setpoint in order to be able to detect a drift in the operation of the valve (it no longer opens enough caused by mecanical problem for example) If a malfunction of the valve is detected, our supervision system (Osisoft) must be able to display the trend (only the cycles where a large difference between setpoint and feedback is detected. I will use an S7-1500 and the analog inputs will be on Profinet UR20 remote modules from Weidmuller Thanks in advance

Suggestion To thank Quote Answer

6/15/2023 12:54 PM	Rate (0)
0_Einstein Diamond Expert Joined: 7/7/2010 Last visit: 9/4/2025 Posts: 16058 Rating: (2513)	One approach might be to build logic and HMI / supervision setup to allow "learning" what a good valve control looks like. Unfortunately, you may need 100 such learnings - which can be done using Siemens HMI with faceplate(s) and associated FB(s). Essentially put the system into "learn mode" for a specific valve, then trigger 0-100% change over 10s and capture the response. Save that as "normal". Then periodically compare actual position vs. setpoint against "normal" for all the valves, reporting any out of spec instances with a count. When the count reaches a certain threshold (valve specific and experimentally determined over time), report it is now in "malfunctioning". Noise and normal system disturbances will generate deviations from the saved "normal". It's a matter of how long it is out of band relative to normal and at what point in the open/close. If it's 50% open and 10% out, it's probably already "malfunctioning", whereas at 5% and 1% out, it's probably just fine - depending on valve specs and process requirements. JeromeSGL Hello everyone First of all, sorry if I am not clear in my words, I am French and my english is not very good ! I turn to you because I have a need but I don't know exactly how to go about it... I explain : I have a hundred proportional valves on our machine (Glass bottle manufacturing), each of them have 2 analog signals 4-20mA, one is the setpoint (opening of a proportional valve), the other is the feedback position of the proportional valve . The proportional valves operate in cycles of approximately 10s. For each cycle, I must be able to compare feedback and setpoint in order to be able to detect a drift in the operation of the valve (it no longer opens enough caused by mecanical problem for example) If a malfunction of the valve is detected, our supervision system (Osisoft) must be able to display the trend (only the cycles where a large difference between setpoint and feedback is detected. I will use an S7-1500 and the analog inputs will be on Profinet UR20 remote modules from Weidmuller Thanks in advance
	science guy
Suggestion To thank Quote Answer

6/18/2023 1:00 PM	Rate (0)
laboutvid Joined: 5/31/2023 Last visit: 1/10/2025 Posts: 3 Rating: (0)	Hello and thank you for your precise answer. My main need at the moment is not to find the maximum value between the setpoint and the feedback. My need is on the implementation of the program allowing me to compare these values. My first idea was : - copy the analog input values of setpoint and feedback and record each in a table every 100ms for example and this so as to have the data on the last 5 sliding seconds. - calculate the average of each table at each cycle. - compare the 2 averages of setpoint and feedback table, if the difference between the 2 averages is greater than x, I copy the table to another table and ask OsiPi to retrieve this table. You should also know that each value reading must be timestamped so that Osipi can display them in the form of a graph. My problem is that I don't know how to do this on Tia Portal...
	Last edited by: JeromeSGL at: 06/18/2023 14:00:44 Last edited by: Jen_Moderator at: 06/19/2023 11:36:13 Text formatting optimized.
Suggestion To thank Quote Answer

6/18/2023 1:37 PM	Rate (0)
Gustavo Antonio B... Member Joined: 6/20/2017 Last visit: 8/31/2025 Posts: 486 Rating: (41)	In my opinion there is a lot of data to handle, but if you have a big enought CPU I think that can be solved. In similar cases I will trace the analog input of each position sensors, maybe after normalize the value to an engineering scale, recording a buffer of values at constant time, for example every 100 ms, (in 10 s for each sensor the buffer must be able to store 100 items), and at the end of the cycle you can analyze the data with your own function, using some math: you can calculate the maximum aplitude, the area of the curve, the flexes and all what can be youseful for describe process that you want you to take under control. More simple: you can store a buffer of a correct curve, next, define the tolerance around this one and for each sensors cycle analyze if the feedback curve is in tolerance or not. It's not difficult at all. With some rows of SCL and the correct math to describe the model of th process and some test and defining the limits for the warnings, it can be done.
	Gustavo Antonio Banchetti Brucatori.
Suggestion To thank Quote Answer

6/18/2023 2:40 PM	Rate (0)
laboutvid Joined: 5/31/2023 Last visit: 1/10/2025 Posts: 3 Rating: (0)	Sorry for this question but how do you make the curves of the values you are talking about?
	Last edited by: Jen_Moderator at: 06/19/2023 11:36:45 Text formatting optimized.
Suggestion To thank Quote Answer

6/18/2023 4:04 PM	Rate (0)
Gustavo Antonio B... Member Joined: 6/20/2017 Last visit: 8/31/2025 Posts: 486 Rating: (41)	For example, with the code in the attached Tia 18 program. (Rename it from zip, to zap18). Create 1 DB for each, copying from the 01, and call the function that you can see in the main, one time for each sensor to track, changing the inpute parameters... You will have 200 records: 20 seconds of data, that can be analized as wish. Attachment Progetto1.zip (393 Downloads)
	Gustavo Antonio Banchetti Brucatori.
Suggestion To thank Quote Answer

|

Share this page:

Share this page on...

Follow us on