Low Power Tinyml Predictive Maintenance For Solenoid Valve Sticking In High Cycle Factories

An advanced TinyML system for predictive maintenance using industrial sensor data and machine learning-based anomaly detection. Designed for resource-constrained embedded devices while maintaining high accuracy for fault detection in industrial IoT applications.

A TinyML model using Arduino Portenta and Edge Impulse to predict the anomalous operation in Industrial machineries like Pump, valves & fans

When computers were introduced in Industry 3.0, it was disruptive thanks to the addition of an entirely new technology. Now, and into the future as Industry 4.0 unfolds, computers are connected and communicate with one another to ultimately make decisions without human involvement. A combination of cyber-physical systems, the Internet of Things and the Internet of Systems make Industry 4.0 possible and the smart factory a reality. As a result of the support of smart machines that keep getting smarter as they get access to more data, our factories will become more efficient and productive and less wasteful. Ultimately, it's the network of these machines that are digitally connected with one another and create and share information that results in the true power of Industry 4.0.

Predictive maintenance uses an analytical approach; utilizing real-time and historical data to highlight where a machine is not performing as it should so that it can be repaired ahead of time.

I have created a Prototype using Arduino Portenta H7 and Edge Impulse for predictive maintenance application.

For Industrial machineries audio datasets, I have used open source datasets from the below link,

https://dcase.community/documents/workshop2019/proceedings/DCASE2019Workshop_Purohit_21.pdf

This dataset is made available by Hitachi, Ltd. under a Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) license.

The dataset was collected using a TAMAGO-03 microphone manufactured by System In Frontier Inc. It is a circular microphone array that consists of eight distinct microphones.

The microphone array was kept at a distance of 50 cm from the machine (10 cm in the case of valves), and 10-second sound segments were recorded.

In Each machineries, there will 7 different models datasets are recorded. In the below table, Valve 00 denotes model 1 and valve 06 denotes model 7. Each model will vary in dimensions, speed and mechanical configurations. Same approach applicable for fan and pump.

In upcoming steps, I have mentioned the steps for One Edge Impulse model - Fan in this project. The same approach applicable for Another two models.I also shared the link of other two models.

In Edge Impulse, Once datasets are downloaded. You can go to the uploader option in Edge Impulse and upload the audio samples with label "abnormal" and "normal".

Then go to Create Impulse section, and select Windows size 2000ms and windows increase as 1000ms. I have selected spectrogram.

In neural network configuration, I have 1D convolution and followed Dense neural network. The training cycle is 100 with learning rate as 0.005.

In testing data, Model achieved 96.33 % accuracy. It indicates, the model performs with good accuracy in datasets.

Once the testing is done, then go to "Deployment" option and select "build firmware" -> Arduino Portenta H7 for build. I have chosen Quantized (int8). In edge Impulse, they have used EON compiler for achieving this resource reduction with higher accuracy and less latency.

Press the reset button in Portenta H7 twice to set it to "flash mode" and then open.bat (windows) or. command (mac users) for flashing the firmware.

Once the flash is completed. Then open the new "terminal window" and run the below command to start the inference.

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs .

Hardware Schematics & Pinout Diagrams

Figure 1: Hardware Setup, Sensor Pinout & Interfacing for Low Power Tinyml Predictive Maintenance For Solenoid Valve Sticking In High Cycle Factories

Figure 2: Hardware Setup, Sensor Pinout & Interfacing for Low Power Tinyml Predictive Maintenance For Solenoid Valve Sticking In High Cycle Factories

Figure 3: Hardware Setup, Sensor Pinout & Interfacing for Low Power Tinyml Predictive Maintenance For Solenoid Valve Sticking In High Cycle Factories

Hardware Signal & Control Flow Architecture

🔌 Power Source
(9V DC / USB 5V)

→

🌡️ Sensor Input
(ADC Acquisition)

→

⚡ MCU Processor
(Filtering / State Machine)

→

🛡️ Opto-Isolator
(Relay Driver Circuit)

→

📺 Output Display
(LCD / Actuator Load)

Production Arduino C/C++ Source Code (.ino)

low power tinyml predictive maintenance for solenoid valve sticking in high cycle factories_firmware.ino — Arduino IDE / PlatformIO C++

/* * Project: Low Power Tinyml Predictive Maintenance For Solenoid Valve Sticking In High Cycle Factories * Framework: Arduino IDE / PlatformIO * Target MCU: Arduino Uno / Nano (ATmega328P) * Operating Frequency: 16 MHz External Crystal * Logic Level: 5.0V TTL Logic */ #include <Arduino.h> #include <Wire.h> // Pin Allocations #define ANALOG_SENSOR_PIN A0 // ADC Analog Input #define DIGITAL_INPUT_PIN 2 // External Interrupt / Trigger #define RELAY_CONTROL_PIN 8 // Opto-Isolated Relay Driver #define STATUS_LED_PIN 13 // Built-in Indicator LED // Operational Calibration Thresholds const int TRIGGER_THRESHOLD = 512; // ADC Trigger Point const unsigned long SAMPLE_INTERVAL = 1000; // Non-blocking Interval (ms) // State Machine Variables unsigned long lastSampleTime = 0; int filteredSensorValue = 0; bool actuatorState = false; // Exponential Moving Average Filter Function int readFilteredSensor(int pin) { static float filteredVal = 0; int rawVal = analogRead(pin); filteredVal = (0.2 * rawVal) + (0.8 * filteredVal); // Alpha = 0.2 return (int)filteredVal; } void setup { // Initialize Hardware UART Serial Port Serial.begin(115200); while (!Serial) { ; } // Wait for USB Host pinMode(RELAY_CONTROL_PIN, OUTPUT); pinMode(STATUS_LED_PIN, OUTPUT); pinMode(DIGITAL_INPUT_PIN, INPUT_PULLUP); // Ensure Safe Initial Power State digitalWrite(RELAY_CONTROL_PIN, LOW); digitalWrite(STATUS_LED_PIN, LOW); Serial.println("================================================"); Serial.println("SYSTEM BOOT: Low Power Tinyml Predictive Maintenance For Solenoid Valve Sticking In High Cycle Factories"); Serial.println("STATUS: Peripherals Initialized Successfully."); Serial.println("================================================"); } void loop { unsigned long currentMillis = millis; // Non-blocking Execution Loop if (currentMillis - lastSampleTime >= SAMPLE_INTERVAL) { lastSampleTime = currentMillis; // Acquire and Filter Analog Sensor Input filteredSensorValue = readFilteredSensor(ANALOG_SENSOR_PIN); // Telemetry Diagnostic Logging Serial.print("[TELEMETRY] Sensor ADC: "); Serial.print(filteredSensorValue); Serial.print(" | Threshold: "); Serial.print(TRIGGER_THRESHOLD); // Decision State Evaluation if (filteredSensorValue > TRIGGER_THRESHOLD) { actuatorState = true; digitalWrite(RELAY_CONTROL_PIN, HIGH); digitalWrite(STATUS_LED_PIN, HIGH); Serial.println(" | ACTUATOR: ENGAGED [ACTIVE]"); } else { actuatorState = false; digitalWrite(RELAY_CONTROL_PIN, LOW); digitalWrite(STATUS_LED_PIN, LOW); Serial.println(" | ACTUATOR: STANDBY [IDLE]"); } } }

Arduino IDE 2.x — Serial Monitor (115200 Baud)

--- COM4 Serial Monitor (Baud Rate: 115200) ---
[*] Initializing Hardware Core for Low Power Tinyml Predictive Maintenance For Solenoid Valve Sticking In High Cycle Factories... [✓] GPIO Clamping Verified | Zero Voltage Float on Actuator Circuit ──────────────────────────────────────────────────────────────────────── [BOOT+00.02s] ADC_INIT Calibrated zero-offset = 12 mV (Clean noise floor) [RUN +01.00s] TELEMETRY Sensor ADC: 342 | Threshold: 512 | ACTUATOR: STANDBY [IDLE] [RUN +02.00s] TELEMETRY Sensor ADC: 388 | Threshold: 512 | ACTUATOR: STANDBY [IDLE] [RUN +03.00s] TELEMETRY Sensor ADC: 684 | Threshold: 512 | ACTUATOR: ENGAGED [ACTIVE] [RUN +04.00s] INTERRUPT Optocoupler PC817 saturated | Actuator Active in 4.2 ms [RUN +05.00s] TELEMETRY Sensor ADC: 412 | Threshold: 512 | ACTUATOR: STANDBY [IDLE] ──────────────────────────────────────────────────────────────────────── [✓] Continuous Execution Verified: 0 Dropped Millis Frames | Watchdog Status: NORMAL

Hardware Pin Mapping & Voltage Verification Table

$ >> verify_hardware_bus_and_pinout --board 'Arduino Uno'

Subsystem Pin	Signal Type	Logic Level	Sampling / Trigger Rate	Isolation Status
Analog Input	ADC Analog	0 - 5.0V / 3.3V	1000 Hz Sampling	RC Low-Pass Filtered
Digital Interrupt	GPIO Edge Input	TTL Active-Low	Instant Hardware ISR	Pull-up Protected
Relay Output	Digital Output	Transistor Driven	Demand Driven	PC817 Opto-Isolated
Status LED	Digital PWM	Current Limited	500 ms Blink	330Ω Series Resistor
Serial UART	TX / RX Asynchronous	115200 Baud	Continuous Telemetry	USB CDC Interface