{
  "id": 12849441,
  "title": "The Machine Data You're Collecting but Not Using",
  "url": "https://urgent.news/2026/10/08/the-machine-data-youre-collecting-but-not-using",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-10-08T10:56:59.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/emmanuelldev/the-machine-data-youre-collecting-but-not-using-19ob"
  },
  "original_language": "en",
  "account": "Many manufacturing plants generate vast amounts of process data every second, storing it in historians and historians for years. However, most of this data remains unused, as there is no connection between the collected information and the decisions made on the shop floor. Engineers only notice issues when equipment trips or produces observable symptoms, rather than detecting early warning signs in the data itself.\n\nThe key issue seems to be the lack of a pathway from the data to actionable decisions. While plants possess ten years of operational data, maintenance teams only learn about failures when equipment stops working.\n\nThree AI use cases are particularly well-suited for plants: predictive maintenance, process parameter optimization, and quality correlation analysis. Predictive maintenance aims to detect patterns in sensor data that precede equipment failure, enabling proactive repairs. Process parameter optimization seeks to identify the best settings for various process variables based on historical data and quality outcomes. Quality correlation analysis looks for upstream process parameters that correlate with downstream quality issues.\n\nWhile predictive maintenance receives much attention, process optimization and quality correlation analysis can often be implemented more quickly and with less effort. Predictive maintenance works best when applied to well-instrumented rotating equipment like pumps, motors, compressors, and fans, where failure modes are well understood and sensor data provides early warning signs. However, it struggles with infrequent failures, lack of sensor coverage, and failure modes that occur too quickly for early detection to be helpful. Implementation challenges largely revolve around data quality, including accurate and complete failure event history, properly aligned sensor timestamps, and consistent sensor tagging across the plant.\n\nData quality issues are often overlooked in favor of focusing on the AI model. Sensor drift, missing data, and inconsistent tagging can all compromise the accuracy of the model. A realistic AI project plan for plant data should account for data assessment, cleaning, and gap-filling before model development. Skipping these steps can lead to disappointing results.",
  "summary": "Most modern plants log process parameters every second, store years of it in historians, and act on almost none of it. The gap is not data collection — it is the connection between what is stored and a decision someone can act on. Most modern manufacturing plants are already data-rich. PLCs log process parameters every few seconds. SCADA systems aggregate them. Historians — OSIsoft PI,…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}