Skip to main content
Back to challenges
Challenge

OPC UA Anomaly Detection Engine using Vercel AI SDK and OpenAI o3

Industrial manufacturing plants rely heavily on OPC UA telemetry streams from SCADA systems to monitor motor vibration, temperature, and power consumption. However, basic threshold alerts generate excessive false positives and fail to provide actionable root-cause explanations for complex multi-variable degradation patterns. In this challenge, developers will construct a real-time predictive maintenance pipeline that consumes raw industrial telemetry, detects subtle sensor anomalies using time-series analysis, and leverages the Vercel AI SDK integrated with OpenAI o3 to generate precise root-cause diagnostics. The system must process high-frequency SCADA streams, aggregate moving-window metrics, and trigger an agentic root-cause investigation whenever anomalous telemetry patterns are detected. Using Vercel AI SDK's structured outputs and tool calling, the AI agent will query historical maintenance logs and equipment blueprints to output verified repair recommendations, minimizing unscheduled plant downtime.

EngineeringHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Build an agentic SCADA telemetry anomaly detector using Vercel AI SDK and OpenAI o3 to analyze time-series metrics and issue root-cause diagnostics.

Delivery guide

How work is evaluated

Evaluation

Evaluates the precision of anomaly detection and accuracy of AI-generated root-cause work-orders against ground truth telemetry anomalies.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

Anomaly Detection Precision

Correctly flags synthetic anomalies without false alerts on baseline stream

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

Diagnostic Accuracy

Semantic overlap of root cause diagnosis with gold evaluation data • target: 0.9 • range: 0-1

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

What you should walk away with

  • Process high-frequency OPC UA telemetry streams for multi-variable motor metrics

  • Implement time-series anomaly detection with dynamic thresholding

  • Construct a Vercel AI SDK agent leveraging OpenAI o3 for root-cause analysis

  • Generate structured JSON maintenance work-orders with confidence scoring

Resources and assets

Reference links and supporting material

Dataset notes

Synthetic OPC UA SCADA time-series telemetry representing motor bearing fatigue, stator over-current, and alignment failures with labeled maintenance ground truth.

Challenge setup

Evaluates the precision of anomaly detection and accuracy of AI-generated root-cause work-orders against ground truth telemetry anomalies.

Challenge input

JSON containing 10-minute sliding window of OPC UA metrics (vibration, bearing temp, current draw)

Vercel AI SDK

Mandatory framework for TypeScript agent tooling

OpenAI o3

Assigned high-reasoning model for diagnostic inference

Evaluated output

JSON with fields: anomaly_detected (boolean), primary_fault (string), root_cause_explanation (string), target_component (string)

Evaluation checks
  • Correctly flags synthetic anomalies without false alerts on baseline stream
  • Semantic overlap of root cause diagnosis with gold evaluation data • target: 0.9 • range: 0-1
Configured evaluation evidence
  • Benchmark: Industrial Telemetry Diagnostics Benchmark
  • Diagnostic Accuracy target: 0.9
  • 1 public reference case
Configured runtime
  • JavaScript execution harness
  • node sandbox (unavailable on Versalist)

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

Action Space
  • OpenAI · Required
  • OpenAI o3 · Optional
Policy Serving
  • o3 · Optional

Evaluation contract

  • Anomaly Detection Precision · Weight 1
  • Diagnostic Accuracy · Weight 1

Metrics

  • Diagnostic Accuracy · Target 0.9

Benchmarks

  • Industrial Telemetry Diagnostics Benchmark
  • general

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON containing 10-minute sliding window of OPC UA metrics (vibration, bearing temp, current draw)

    Output: JSON with fields: anomaly_detected (boolean), primary_fault (string), root_cause_explanation (string), target_component (string)

    Execution harness: JavaScript

Evaluation tools

  • Vercel AI SDK · Orchestration framework
  • OpenAI o3 · Reasoning LLM

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about OPC UA Anomaly Detection Engine using Vercel AI SDK and OpenAI o3