What the data for this category looks like
The data sources for thermal intelligent due diligence reports are thermal operation management systems, smart meter backends, pipeline network operation and maintenance logs, and local regulatory submission reports. Update frequency follows the meter reading cycle: a single record is typically generated hourly or daily. Each individual document is a structured entry containing fields including pipeline network number, supply and return water temperature, pressure, cumulative flow, meter reading time, associated device ID, operation and maintenance batch number, and more. Supply and return water temperature is measured in ℃, pressure in MPa, and flow in m³/h. Each record is bound to a unique device identifier and meter reading timestamp.
What constraints these characteristics impose on the citation sources and traceability workflow
The fine-grained single records and multi-field characteristics of thermal data require that device identifiers and meter reading times be displayed during traceability to avoid ambiguous cited content. The hourly or daily update feature requires the traceability system to mark data batches to prevent repeated citation of historical records. The large number of structured fields and association with unique device IDs require precise field extraction during parsing to avoid missing traceability information. The large number of batch-uploaded thermal data files with short individual entries requires the recall configuration to balance quantity and accuracy to prevent redundant or insufficient context.
How to set the configurations
| Configuration Key | Recommended Value | Rationale | |||
|---|---|---|---|---|---|
recall_count | Top 8 entries | Thermal data entries are short, so 8 entries cover complete pipeline operating condition context | |||
similarity_threshold | 0.72–0.78 | Thermal data has many fields, and similarity judgment must balance device matching and time correlation. A threshold that is too low will introduce irrelevant records | |||
citation_content_template | `{{source_name}} | Device ID:{{device_id}} | Meter Reading Time:{{time}} | {{content}}` | Must display core identifiers of thermal data to quickly locate specific records during traceability |
incremental_update_mark_field | Meter Reading Batch Number | Thermal data is updated in operation and maintenance batches. Using batch numbers as marks prevents duplicate import of historical data | |||
PARSE_FILE_TIMEOUT_SECONDS | 300 seconds | Structured thermal data files are mostly in CSV or Excel format, with moderate parsing time. 300 seconds covers batch file processing | |||
deduplication_rule | By device ID + meter reading time | The unique identifier for thermal data is the combination of device and time. This rule completely prevents duplicate citations |
The parameter values provided on this page are common starting points for configuration. Actual values are affected by material form, data volume, and business rules. Specific issues require specific analysis. It is recommended to test on your own samples before finalizing settings.
Common mistakes to avoid
- After uploading batch thermal data files, duplicate traceability entries appear during knowledge base citation. The cause is that the deduplication rule by device ID + meter reading time is not configured, resulting in the same meter reading record being identified multiple times.
- After selecting the variable reference mode in the AI chat component, the temperature setting button disappears, and generation parameters cannot be adjusted. The cause is that the advanced parameter configuration switch for variable references is not enabled in system settings.
- Only the file name is displayed in the citation content pop-up window, and device ID and meter reading time are not shown. The cause is that the placeholders for
device_idandtimefields are not configured in the citation content template.
How to confirm the configuration is complete
- Upload a single structured thermal data file, view the parsed field list in the knowledge base, and confirm that custom fields including
device_idandmeter_reading_timeare correctly extracted. - Initiate a test query, view the citation source pop-up window, and confirm that the displayed content includes the configured template fields.
- Upload multiple historical data entries for the same device, check whether the knowledge base citation automatically deduplicates and only retains records that meet the update rules.
- Adjust the
similarity_thresholdparameter, initiate multiple rounds of test queries, and verify that the number of recall results meets business expectations.
Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-09-14.