Explain the maintenance decision with evidence

Return to the shift question and explain the inspection scope, the first stop, and what runs in parallel among four HIGH machines.

Question for this chapter

Which machine goes first, and what evidence supports that decision?

Return to the original problem

Four of the six anomalies are HIGH, and all four machines share a health score of 90. There is no evidence to release any of them as normal. The solution is to keep every inspection in scope while making the first action and the parallel scope transparent.

Complete the solution

Initial problem	Applied solution	Evidence left behind	What changed
Sensor values used unlike units.	Read values with `quality_flag`.	Four `HIGH_*` rows, two `OFFLINE` rows	Alert candidates separated from unusable rows.
Severity across six anomalies looked arbitrary.	Read z-score and condition signal separately.	CNC-01 at 4.00 `MEDIUM`, CNC-07 at 4.22 `HIGH`	A condition signal pins the grade.
Raising sigma seemed likely to remove alerts.	Reran at 3.0, 3.3, and 4.1.	Four `HIGH` rows survive 4.1	Statistical and condition rules are separate.
Four machines were tied.	Split health priority from the z-score tie-break.	Four scores of 90; CNC-03 at 6.89	Inspection scope and first stop are separate decisions.
An alert row lacked field context.	Connected the five-category ontology.	Technician, zone, and shift	Every field on the work order is filled.
All four seemed to share one queue.	Compared technician against zone.	`TECH-03` with 2; `TECH-04` and `TECH-05` with 1 each	Three lanes cut total handling time.

State the final decision

CNC-03, CNC-05, CNC-07, and CNC-09 all have HIGH alerts from explicit quality signals, so isolate and inspect all four. CNC-03 and CNC-09 share one queue under Park Ji-hoon (TECH-03) in LOC-A2, so the workshop tie-break starts with CNC-03's vibration sensor, which has the largest relative deviation, and CNC-09 follows. CNC-05 belongs to Choi Min-seo (TECH-04) and CNC-07 to Jeong Ye-eun (TECH-05), so both proceed at the same time. CNC-01 (MEDIUM, 96) and CNC-02 (LOW, 99) were caught by statistics alone and stay running under observation. All six events were handled in the day shift (SHIFT-D).

Completion check

Separated four HIGH_* rows from two OFFLINE rows among 216 sensor rows.
Can explain why normalization reduces 216 rows to 213.
Confirmed what split the grades of CNC-01 and CNC-07 at nearly equal z-scores.
Can explain why four HIGH rows survive sigma 4.1.
Confirmed that all four HIGH machine health scores are 90.
Confirmed that 6 entities and 6 relations cover all five top categories.
Read the technician, zone, and shift for all four alerts in one query.
Can separate the inspection set, the watch list, the first inspection, and the parallel scope.

Apply next

In real maintenance work, put safety class, asset criticality, production impact, and the maintenance SOP ahead of z-score in the tie-break rule. The important part is to make the rule visible so every crew can reproduce the same decision from the same evidence.