Introduction to the AI Services module

Overview

The AI Services module is a comprehensive artificial intelligence and machine learning platform for industrial applications. It provides intelligent analytics capabilities enabling automated anomaly detection, pattern recognition, and predictive analytics for industrial processes.

Architecture

The AI Services module consists of three main components:

AI Core (Java)

A Spring Boot-based application that serves as the central orchestration layer:

Provides REST APIs for model deployment and inference management
Manages AI model lifecycle (deployment, execution, monitoring)
Handles security, authorization, and multi-tenancy
Coordinates communication between frontend and Python-based ML services
Integrates with external systems via event-driven architecture

AI Python Services

Python-based microservices implementing the machine learning algorithms using a Nameko microservice architecture:

AI Training Service: Trains machine learning models based on historical process data
AI Scoring Service: Executes trained models to score new data in real-time
Implements domain-specific ML modules (CAS and SDM)

AI Frontend

An Angular/NX-based user interface:

Provides visualization and configuration capabilities
Enables users to manage AI models and view analysis results
Offers intuitive dashboards for monitoring model performance

Key Capabilities

Curve Analytics Service (CAS)

The CAS module provides self-sustained unsupervised anomaly detection for process curves:

Automatically detects abnormal patterns in time-series process data
Requires no manual labeling or supervised training
Continuously learns from process behavior
Identifies outliers and anomalies in real-time

Sequence Detection Module (SDM)

The SDM module identifies repeating patterns of events that lead to machine faults:

Analyzes sequences of events preceding failures
Discovers hidden correlations between events and outcomes
Presents actionable insights to users
Enables proactive maintenance strategies

Technology Stack

Backend: Java (Spring Boot), Python (Nameko microservices)
Frontend: Angular, TypeScript, NX workspace
Infrastructure: Docker containers, Kubernetes deployment
Integration: RabbitMQ (message broker), REST APIs, Event-driven architecture
ML/Data Processing: scikit-learn, NumPy, SciPy, pandas

Deployment

The AI Services module is deployed as three Docker containers:

ai/ai-core - Java Spring Boot application
ai/ai-training - Python-based training service
ai/ai-scoring - Python-based scoring service

All components are designed for cloud-native deployment on Kubernetes with support for horizontal scaling and high availability.