Download presentation
Presentation is loading. Please wait.
1
Machine SCADA Data Analysis Using MATLAB
ASPEC Team Meeting "Design Tools and Templates" Friday, 3 Aug 2018, Sydney Tatsuya Uno
2
Project Examples Reclaimer EOL fatigue study (W020)
SCADA data Actual loads and cycles Strain gauge data FE model verification/calibration Reclaimer slew drive alarms investigation (W043k) SCADA data Identify pattern of alarms Establish cause Engineering Inspections and Plant Integrity Reviews (WR___) SCADA data Check operating within design assumptions Report on utilisation and productivity Reclaimer boom dynamics fatigue assessment (W301) 10 Hz SCADA data Identify patterns Quantify loads and cycles Compare against design
3
MATLAB Introduction and Demonstration
What is MATLAB? Numerical computing software (compare symbolic) Programming language Strengths Large amounts of data Numerous mathematical functions Visualisation tools Repetitive tasks Support Updated, advancements made Well documented (official and community) Weaknesses Learning curve Cost
4
Demonstration Layout Show example Windows Import data - drag and drop
Directory, command, history, workspace Show example Import data - drag and drop Find example with both numerical and categorical data Table, timetable - structure, metadata summary(), head(), tail() Tab completion F1 Help Open Plot plot(), histogram() zoom, pan, data label Plot controls – title, labels, legend Generate script Publish to *.doc
5
SCADA Data RFI
6
SCADA Data RFI (contd.)
7
SCADA Data RFI (contd.)
8
Importing Data into MATLAB - Datastore
Repository for a collection of data files (w/ same structure and formatting) Designed for data too large to fit in memory (see tall arrays) How to use Specify which files form the datastore (raw data) Specify how a single file from the datastore is to be read Process one, multiple, or all files at a time How we use (data not too large to fit in memory) Load all files from the datastore Clean up data Save as a single file (*.mat)
9
Importing Data into MATLAB - Datastore (contd.)
ReadFcn example (fileDatastore 'ReadFcn')
10
Importing Data into MATLAB - Datastore (contd.)
Problem: Datastore file order based on filename Need to sort data chronologically Method 1 Read datastore (raw data) once and sort files Update datastore with chronological order of files Read from datastore (raw data) again in correct order Method 2 Import the data in any order then sort all the rows in memory ¿Porque no los dos?
11
Importing Data into MATLAB - Import Functions
CSV format reads much faster than Excel format Recommend first converting *.xls? files to *.csv outside of MATLAB Import CSV with readtable(), with detectImportOptions() Not recommended readtable() without detectImportOptions() Too slow unless 'TreatAsEmpty' known csvread() Numeric data only xlsread() Too slow with non-numeric data textscan() Too low level Cell arrays use lots of memory
12
Importing Data into MATLAB (contd.)
Achtung! Excel formatting of CSV files Saving CSV Numbers and dates saved with current display format Lost decimal places, seconds, minutes, hours, etc Opening CSV Date formatting applied (may differ from original) Recommendations Format dates with required precision or as "General" (serial date) Use CSVed (freeware), alternatively use Notepad
13
Summarise Data - Example
14
Importing Data into MATLAB (contd.) - Issues with Data
Summarise data, look for common issues Inconsistent file structure, format Large gaps between files, files overlapping in time Low sampling frequency Batch plot - More issues Wrong data Too much missing data Erroneous data Interpolated data (e.g. BW current, position) Linear Zero-order hold Resolution of position data Zero error (weightometer, cylinder pressures, BW torque arm) Sign convention Drive power/torque Speed %
15
Issues with Data - Examples
16
Issues with Data - Examples
17
Issues with Data - Examples
18
Issues with Data - Examples
19
Issues with Data - Examples
20
Processing Data with MATLAB - Scripts
Run in order. Results to publish. Run in any order. Add new ones. Read raw data *.csv Read single *.mat Read single *.mat
21
MATLAB Scripts and Visualisation Examples
Conveyor strength (F), and fatigue (CF.F) Utilisation Boom tip position Luffing cylinder force Digging load strength (U), and fatigue (CU.U) Performance Points 2D vs 3D plots Linear vs log scales
22
MATLAB Scripts and Visualisation Examples (contd.)
FS FT
23
MATLAB - Recommendations
General F1 contextual help Tab completion Publish scripts to *.doc without code Use and write custom functions for repetitive tasks - neater scripts Consider result of NaNs in data for all operations performed SCADA analysis Use datastore for reading raw data Use readtable() with detectImportOptions() for reading *.CSV files Use timetable (preferred) or table data types retime() synchronize() isregular() Store metadata via TT.Properties Standardise naming of timetable variable names Reference timetable variables by name (not number)
24
Bonus slides from Feb 2016
25
Visualisation Linkaxes
26
Licencing and Resources
ASPEC MATLAB licences 1×Standalone (Perth) 1×Network license Signal Processing Toolbox Optimization Toolbox MATLAB Documentation MATLAB Courses Fundamentals online course notes (screenshots) are available MATLAB Central (forum) File Exchange (user written functions and apps/tools)
Similar presentations
© 2025 SlidePlayer.com Inc.
All rights reserved.