Power Quality Classification

Multi Layer Perceptron

Dataset 1:

STM32CUBEMX Results

Dataset 2:

STM32CUBEMX Results

Test Results

Dataset 1:

Test accuracy - 100% (using accuracy as a loss function)

Train accuracy - 99.97%

Validation accuracy - 99.92%

Dataset 2:

Test accuracy - 99.24% (using accuracy as a loss function)

Train accuracy - 100%

Validation accuracy - 99.94%

CNN

Dataset 1:

STM32CUBEMX Results

Dataset 2:

STM32CUBEMX Results

Test Results

Dataset 1:

Test accuracy - 100% (using accuracy as a loss function)

Train accuracy - 100%

Validation accuracy - 100%

Dataset 2:

Test accuracy - 99.80% (using accuracy as a loss function)

Train accuracy - 99.74%

Validation accuracy - 100%

Accomplishments

• Python Keras model as output.
• The module developed shall read through and parse the csv input files.
• Model should be built and evaluated within your python code.
• STM32MX project to be created to evaluate the memory size of the model
• Model should be adjustable dynamically to changed sampling rate or number of samples
• Ability to detect a combination of conditions

To know more details check this out.