andrevier / converterfaultdetection Goto Github PK

The paper: "Fault Diagnostic in Power System Using Wavelet Transforms and Neural Networks".
the authors: F.Charfi, F.Sellami, and K.Al-Haddad
Open a branch Charfi2006 to deal with this issue.

1. Create a file charfi2006.py to implement the paper's algorithm.
2. Implement the algorithm with tensorflow.

Implement a method in the ConditionMonitoring class to process the input current signals and get wavelet transforms.
Class file:
source/tools/conditionMonitoring.py

The description of the fault phenomenon for the inverter and their effect on the control and the motor.

Implement the fault detection algorithm from the paper:
A Diagnostic Method of Simultaneous Open-Switch Faults in Inverter-Fed Linear Induction
Motor Drive for Reliability Enhancement
Authors:Jung-Hyun Choi,Sanghoon Kim, Dong Sang Yoo, and Kyeong-Hwa Kim.
The algorithm should work for single open-switch and double fault in the same inverter's leg. According to the paper, this corresponds to the fault modes 1 (single leg fault) and 2 (single switch).
The cases necessary to test this two modes are:
(T1,T4), (T2,T5), (T3, T6), T1, T2, T3, T4, T5, T6.

The key points to write:

1. Why the theme is important to study?
2. Who cares about this theme?
3. Who it will be studied during the dissertation.

The README.md file and all the code is written in portuguese and needs translation to english for a broad audience.

Structure of the Literature Review

1. Reliability Engineering and Power Electronics
2. Condition Monitoring
3. The IGBT's functions and Failure Mechanisms
   3.1. IGBT's Main Technologies
   3.2. Failure Mechanisms
4. Inverter Fault Modes
5. Review On the Fault Diagnostic Methods
   5.1. DC Offset Based Methods
   5.2. Voltage Based Methods
   5.3. Spectrum Analysis and Wavelet Based Methods
   5.4. Machine Learning
6. Chapter's Summary

The RMS function implemented in the folder 'tools' receive a signal as input and calculate the RMS of each period of the fundamental frequency inside the signal array. The RMS curve is updated at the end of each period of N elements. Hence, the final array has the same length of the signal but has step changes along the curve.
The moving RMS uses a moving avarage of elements. Suppose, a = [1,2,3,4,5,6] and the number of each step is 2. So, the first group of 2 is (1,2), the second is (2,3), and so on. The array b has all the possibilities:
b = [0,(1,2), (2,3), (3,4), (4,5), (5,6)]
The array b has the same length of the array a is filled with leading zeros.

Review of the literature concerning the recent fault detection methods for the semi-conductor switches in the converter.

The dissertation is about detection of faults in IGBTs.
The text need to explain why the work chooses this switch instead of others, such as IGCTs.

Describe the simulation model used in the simulink file.
The description must have:

1. The type of the model, once is a pre-installed model in the MATLAB;
2. How it work;
3. Normal behavior of the system and plots of the main variables.

Check list for the simulation:

1. Choose the motor driver and its variables;
2. Change the driver's converter for faults.
3. Write a script to set up the simulation's parameters.

The data of stator currents, line voltages, torque and speed has been collected in each simulation so far. But, some of the data collected in the same leg open-circuit faults, like T1 and T4, has more the 100MB which surpass the maximum allowed in Github. So, the data has to be moved to an external drive or a cloud.

The current Latex version of the text is in the overleaf platform. It's a good procedure to keep a copy in the project repository and to track the issues.