Predictive Maintenance in Motor Control Applications

Deployed Predictive Maintenance System

Repository Overview

This repository is a companion to the Predictive Maintenance with MPLAB ML tutorial. It contains the firmware to classify the operational state of a motor - whether the motor is in normal condition or experiencing any anomalies such as unbalanced load by monitoring Iq current of the motor on dsPIC33CK LVMC Development Board with the 24V 3-Phase Brushless DC Motor

The supported Classifications (shown in the video above) are:

• Normal operation
• Unbalanced load

In addition, there is also an 'unknown' class for the operation those don't fall under these categories.

Hardware Used

• dsPIC33CK LVMC Development Board (DM330031)
• 24V 3-Phase Brushless DC Motor (AC300020)

Software Used

• MPLAB® X IDE
• MPLAB® XC16 Compiler
• MPLAB® Code Configurator
• MPLAB® ML Model Builder

Related Documentation

• dsPIC33CH512MP508 Product Family Page
• motorBench® Development Suite Software Plugin Details

Data Collection

The dataset used for the development of this application consists of Motor IQ current and Motor RPM measurements taken from a 24V 3-Phase Brushless DC Motor connected to a dsPIC33CK LVMC Development Board, running on an FOC motor control algorithm. No sensors were used in this application. For further description of the application setup, see the Predictive Maintenance with MPLAB ML tutorial.

Data Collection Firmware

A binary build of the data logging firmware used in the data collection for this project can be found in the binaries folder of this repo; to build data logging firmware for different configurations, visit the ml-33ck265mp508-lvmc-datalogger repository.

Firmware Operation

The behavior of the firmware can be summarized as operating in one of the distinct states described in the table below.

State	UART Terminal output	Description
Motor OFF	Motor is in idle state!	Motor is in off state
Normal	Normal_Operation	Motor is running normally(no load or balanced load)
Unbalanced load	Unbalanced_load	Unbalanced load is detected
Buffer Overflow	Overrun!	Processing is not able to keep up with real-time; data buffer has been reset.
ERROR	ERROR: Got a bad motor status	The motor controller has recognized a fault condition, under which it is unable to proceed safely with normal control activities.

Note the firmware class ID mapping is as below:

• Unknown - 0 (input outside of modeled behavior)
• Normal_Operation - 1
• Unbalanced load - 2

Running the application Demo

• Start MPLAB X IDE and open the project anomaly-detection-ml-33ck256mp508-lvmc.X with device selection dsPIC33CK256MP508.

• Set the project anomaly-detection-ml-33ck256mp508-lvmc.X as the main project.

• Open the app_config.h file located under Header Files.

• Ensure that the macro DATA_STREAMER_FORMAT is set as DATA_STREAMER_FORMAT_NONE

​

• Open project properties and ensure that the selected MPLAB® XC16 Compiler and Device Pack support the device configured in the firmware.

• Build the project and program the device.

• Open the terminal window on Data Visualizer, select the COM port of the LVMC board, and set the below configurations

  • Baud rate 115200
  • Data bits 8
  • Stop bits 1
  • Parity None

The below message will be printed right after the device reset. Indicating Knowledge pack number and the motor status.

• Press switch SW1 to start the motor. If the motor runs normally, the below output will be shown.

If an unbalanced load is detected:

Firmware Benchmark

With -O2 level compiler optimizations, and 100MHz clock (Fcy)

• 25.9kB Flash
• 9.3kB RAM
• 25ms Inference time(average)

Classifier Performance

Below is the confusion matrix for the test dataset. Note that the classes are highly imbalanced so accuracy is not a good indicator of overall performance.