Fan Condition Monitoring with MPLAB Machine Learning Development Suite

Deployed fan condition monitor

Repository Overview

This repository is a companion to the Fan Condition Monitoring with MPLABML tutorial on the Microchip Developer website. It contains the firmware to classify the operational state of a Honeywell HT-900 fan including whether the fan is on, which speed setting it's set at, and whether the fan is experiencing one of two fault conditions (tapping or shaking). The demo project runs on the SAMD21 Machine Learning Evaluation Kit with the Bosch BMI160 IMU (Mikroe IMU2 click board). The TDK variant (Mikroe IMU14 click board) of the evaluation kit cannot be used directly with this firmware because the sensor does not natively support the sample rate that was used when developing the fan classifier model.

The fan states that the model can recognize (shown in the video above) are summarized below:

• Unknown
• Fan Off
• Tapping
• Fan Speed 1
• Fan Speed 2
• Fan Speed 3
• Shaking

Hardware Used

• SAMD21 Machine Learning Evaluation Kit with Bosch BMI160 IMU (EV45Y33A)

Software Used

• MPLAB® X IDE (https://microchip.com/mplab/mplab-x-ide)
• MPLAB® XC32 compiler (https://microchip.com/mplab/compilers)
• MPLAB® Harmony 3 (https://www.microchip.com/harmony)
• MPLAB® Machine learning developement suite (https://www.microchip.com/en-us/tools-resources/develop/mplab-machine-learning-development-suite?s_kwcid=AL!18852!3!!!!x!!&utm_source=google_mchp&utm_medium=cpc&utm_term=&utm_campaign=&utm_content=x++c&gclid=CjwKCAiAqKbMBhBmEiwAZ3UboNOy-1r1CMfokCStar1Si7OWkH5M-T4t4Q8Nqw-kVxIYJcwicv7A5RoCWokQAvD_BwE&gad_source=1&gad_campaignid=23306459457&gbraid=0AAAAAD-ta3C1smiP3QE7o5Y8kvHfL92ZC&gclid=CjwKCAiAqKbMBhBmEiwAZ3UboNOy-1r1CMfokCStar1Si7OWkH5M-T4t4Q8Nqw-kVxIYJcwicv7A5RoCWokQAvD_BwE)

Related Documentation

• ATSAMD21G18 Product Family Page
• SAM-IoT Development Board Product Details

Data Collection

The dataset used for the development of this application consists of 6-axis IMU recordings taken with the SAMD21 BMI160 evaluation board mounted to the housing of a Honeywell HT-900. For further description of the application setup, see the article for this demo on the Microchip Developer website.

Data Collection Firmware

A binary build of the data logging firmware used in the data collection for this project can be downloaded from the releases page; to build data logging firmware with different sensor configurations, visit the ml-samd21-iot-imu-data-logger repository.

Sensor Configuration

The sensor configuration used in this demo is summarized in the table below.

IMU Sensor	Axes	Sampling Rate	Accelerometer Range	Gyrometer Range
Bosch BMI160	Ax, Ay, Az, Gx, Gy, Gz	400Hz (4x oversampling mode)	2G	125DPS

Note that for this project, the firmware (specifically, bmi160_sensor.c) was slightly modified to enable oversampling functionality. The change can be summed up in the following diff:

--- b/firmware/src/app_config/bmi160/bmi160_sensor.c
+++ a/firmware/src/app_config/bmi160/bmi160_sensor.c
@@ -162,7 +162,7 @@ int bmi160_sensor_set_config(struct sensor_device_t *sensor) {
     /* Select the Output data rate, range of accelerometer sensor */
     sensor->device.accel_cfg.odr = _GET_IMU_SAMPLE_RATE_MACRO(ACCEL); //BMI160_ACCEL_ODR_100HZ;
     sensor->device.accel_cfg.range = _GET_IMU_ACCEL_RANGE_MACRO();//BMI160_ACCEL_RANGE_2G;
-    sensor->device.accel_cfg.bw = BMI160_ACCEL_BW_NORMAL_AVG4;
+    sensor->device.accel_cfg.bw = BMI160_ACCEL_BW_OSR4_AVG1; //BMI160_ACCEL_BW_NORMAL_AVG4;

     /* Select the power mode of accelerometer sensor */
     sensor->device.accel_cfg.power = BMI160_ACCEL_NORMAL_MODE;
@@ -170,7 +170,7 @@ int bmi160_sensor_set_config(struct sensor_device_t *sensor) {
     /* Select the Output data rate, range of Gyroscope sensor */
     sensor->device.gyro_cfg.odr = _GET_IMU_SAMPLE_RATE_MACRO(GYRO); //BMI160_GYRO_ODR_100HZ;
     sensor->device.gyro_cfg.range = _GET_IMU_GYRO_RANGE_MACRO(); //BMI160_GYRO_RANGE_2000_DPS;
-    sensor->device.gyro_cfg.bw = BMI160_GYRO_BW_NORMAL_MODE;
+    sensor->device.gyro_cfg.bw = BMI160_GYRO_BW_OSR4_MODE; //BMI160_GYRO_BW_NORMAL_MODE;

Firwmare Operation

The firmware will reflect the state of operation of the demo using the onboard LEDs; this behavior is summarized in the table below.

State	LED Behavior	Description
Firmware error	Red LED lit	Fatal error. (Do you have the correct sensor plugged in?).
Data buffer overflow	Yellow+Red LED lit for 5 seconds	Processing is not able to keep up with real-time; data buffer has been reset.
Fan Off	No LED lit	Fan is off.
Fan Speed 1	Blue LED lit	Fan is on at speed setting 1.
Fan Speed 2	Green LED lit	Fan is on at speed setting 2.
Fan Speed 3	Yellow LED lit	Fan is on at speed setting 3.
Tapping	Yellow LED flashing	Detected tapping on the fan housing.
Shaking	Red LED flashing	Detected mount shaking.
Unknown	Red+Yellow LED flashing	Detected anomalous behavior.

In addition, the firmware also prints the classification output for each inference over the UART port. To read the UART port output, use a terminal emulator of your choice (e.g., PuTTY) with the following settings:

• Baudrate 115200
• Data bits 8
• Stop bits 1
• Parity None

A sample of the terminal output is shown in the figure below.

UART Terminal Output

Note that the output classification is given as an integer number which is the class ID; the class ID to class label mapping is described in the table below.

Class ID	Class Name
0	Unknown
1	Fan Off
2	Tapping
3	Fan Speed 1
4	Fan Speed 2
5	Fan Speed 3
6	Shaking

Usage with SensiML Open Gateway

The firmware UART output can be visualized with the SensiML Open Gateway application:

1. Open a terminal and change to the directory where you've checked out this repository.
2. Clone the open-gateway repository and install the dependencies:

   git clone https://github.com/sensiml/open-gateway
pip install -r open-gateway/requirements.txt

3. Change the baudrate (BAUD_RATE variable) in open-gateway/config.py to 115200
4. Change to the open-gateway directory and run the open-gateway application, passing in the fan demo's model.json description file:

   cd open-gateway
python app.py -m firmware/knowledgepack/libsensiml/model.json

5. Connect to the SAMD21 board in the gateway application:
   • Select the Recognition device mode.
   • Select Serial connection type, and enter the UART address (e.g. COM4) in the Device ID field.
   • Click Connect To Device.
6. Switch to the Test Mode tab and click Start Stream.

Firmware Benchmark

Measured with the BMI160 sensor configuration, -O2 level compiler optimizations, and 48MHz clock

• 63.2kB Flash
• 5.8kB RAM
• 12ms Inference time

Classifier Performance

Below is the confusion matrix result for the classifier evaluated on the entire ht-900 fan condition dataset.