PLC & AI Integration

Connect AI systems with Programmable Logic Controllers, SCADA, and industrial protocols to create intelligent, adaptive process control systems.

Understanding PLCs

A Programmable Logic Controller (PLC) is a ruggedized computer designed for industrial control. PLCs execute deterministic, real-time logic to control machinery, processes, and production lines. They are the backbone of industrial automation, running 24/7 in harsh factory environments.

Integration Architecture

Reading PLC Data with Python

from opcua import Client

# Connect to PLC via OPC UA
client = Client("opc.tcp://192.168.1.100:4840")
client.connect()

# Read sensor values
temperature = client.get_node("ns=2;s=Sensor.Temperature")
vibration = client.get_node("ns=2;s=Sensor.Vibration")
pressure = client.get_node("ns=2;s=Sensor.Pressure")

temp_value = temperature.get_value()
vib_value = vibration.get_value()
pres_value = pressure.get_value()

print(f"Temperature: {temp_value}°C")
print(f"Vibration: {vib_value} mm/s")
print(f"Pressure: {pres_value} bar")

# AI makes a decision based on sensor data
if predict_anomaly(temp_value, vib_value, pres_value):
    # Write recommendation back to PLC
    speed_setpoint = client.get_node("ns=2;s=Motor.SpeedSetpoint")
    speed_setpoint.set_value(reduced_speed)

client.disconnect()

Industrial Data Collection with MQTT

import paho.mqtt.client as mqtt
import json

def on_message(client, userdata, msg):
    """Process incoming sensor data."""
    data = json.loads(msg.payload)
    timestamp = data['timestamp']
    sensor_id = data['sensor_id']
    values = data['values']

    # Store for model training
    store_to_database(timestamp, sensor_id, values)

    # Run real-time inference
    prediction = model.predict([values])
    if prediction == 'anomaly':
        publish_alert(sensor_id, values)

client = mqtt.Client()
client.on_message = on_message
client.connect("mqtt-broker.factory.local", 1883)
client.subscribe("factory/line1/sensors/#")
client.loop_forever()

Industrial Protocol Comparison

Safety Considerations

• AI should advise, not override: AI recommendations should go through safety-validated PLC logic before acting
• Fail-safe defaults: If the AI system fails, PLCs should revert to safe operating parameters
• Rate limiting: Limit how often AI can change setpoints to prevent oscillation
• Human approval: Critical parameter changes should require operator confirmation