Python development services to enable insight and automation for Industry 4.0.
Python is the preferred industry language for scientific computing, data science and machine learning.
This high-level script programming language is a good alternative language for MATLAB, but is free, and has a large community that allows developers to learn, collaborate and knowledge share. This surging popularity comes from its readability, simplicity and superb array of open source scientific packaged programming tools.
It is the first language of choice for many data scientists; typical applications include scientific research, data processing, signal acquisition, websites and IoT.
Adding value to your projects
Python is fast to program, which means development timescales are shorter than using many other software languages such as Java or C++. Its seed of programming is on par with MATLAB or LabVIEW. It runs on multiple operating systems, from Windows, OS X to Linux plus a wide range of hardware setups and offers fast data processing speeds.
For anything involving data, Python is an obvious choice thanks to its ecosystem of tools and libraries including the highly optimised numpy/sciphy libraries for scientific work making data processing fast. There is also the Scikit-Learn library which contains a huge array of off-the-shelf machine learning tools and interactive jupyter notebooks. All of these tools are free to use so there are no ongoing costs or license fees. We recommend the Anaconda distribution released by Continuum Analytics, and we use this distribution in-house.
CONDITION MONITORING CASE STUDY: OIL & GAS COMPANY
A £multi-billion global leader in the Oil & Gas industry was already collecting significant amounts of operational data and required a partner to help analyse it and improve their operations.
The first phase of the project was to convert the existing finite state machine model to Python code. Using Python to analyse the existing model our team were able to improve the performance of the model by tuning the parameters. We also trained a new machine learning model on the historic data which outperformed the existing model by 20% (tested on data sets provided by the client) which will both reduce operating costs and time to production.
CONDITION MONITORING CASE STUDY: FMCG COMPANY
A £multi-billion global leader in the FMCG industry wanted to optimise the performance of their production line and more specifically a group of identical, highly customised equipment.
The first phase of the project was to install additional sensors on one set of equipment selected by the client. We proposed to analyse data from the additional sensors to identify faulty parts or components (for instance, faults in the existing sensors) in the system. In addition, we worked on captured data to develop a model to detect any faults in the process. With the model, we could provide more accurate and faster information to the controller to improve the performance of the closed-loop system.
The second phase will extend the methodologies to further equipment.
DATA SCIENCE CASE STUDY: RENEWABLE ENERGY ORGANISATION
A renewable energy engineering research group were curious to explore how satellite images can be utilised to develop a machine learning prediction model for solar irradiance
Based on an understanding of solar irradiance fundamentals the team reviewed the available meteorological data sets to select the most relevant data to the prediction requirements. The team then formulated a prediction problem with specific forecast duration and resolution.
The second phase of the project involved processing satellite images (The size of a typical image in this project is ~40MB), extracting features, defining performance metrics and training an artificial neural network. The accuracy of the resulting solar irradiance prediction model was higher than our client expected.
DATA SCIENCE CASE STUDY: DEFENCE SECTOR
An organisation in the Defence sector wished to understand if it was possible to determine specific household activity based on energy consumption.
The first phase of the project was the design of a customised data logger to capture high speed voltage and current data.
The second phase was to detect and disaggregate a specific electrical device profile from the lumped energy data. We trained a detection machine learning model on data from one brand of the device, the accuracy of detecting it when the device is turned on was >95%. The accuracy was above satisfactory when other brands were examined.
The third stage of the project involved the creation of auto-generated reporting.
News & tips
Sign up to our email newsletter to receive regular LabVIEW tips and related updates from the Austin Consultants team.
Get in touch
Talk to us about your project today.