What This Category of Data Looks Like
Data for this category comes from design BOMs, silicon steel sheet and insulating material ledgers, vacuum drying equipment logs, insulating oil test reports, and partial discharge and withstand voltage records from high-voltage test stations. Data is updated per single product or each batch of raw materials. Supporting documents include winding and core assembly records, vacuum drying records, insulating oil test reports, partial discharge and withstand voltage test reports, and nameplate parameter tables. Core fields include insulation layer thickness, water content (unit: ppm), partial discharge amplitude (unit: pC), and withstand voltage test voltage (unit: kV). A single product can be associated with up to ten or more document entries.
How These Characteristics Constrain the "Workflow Orchestration" Link
The requirement to associate a single product with multiple heterogeneous documents means workflows must bind the unique identifier of the corresponding product via nodeId to avoid cross-batch data confusion. For the update rhythm of per single product or batch, trigger nodes must be configured to automatically start workflows according to production batches or completion nodes. The high-precision requirements of core fields require embedding parameter verification links in workflows to validate value ranges for fields such as insulation water content and partial discharge amplitude. In scenarios with multiple associated documents, variables must be used to pass field values from different documents to avoid process lag caused by excessive context load.
How to Set Configurations
| Configuration Item | Recommended Setting | Rationale for This Setting |
|---|---|---|
| Trigger Mode | Automatically trigger upon production completion node | Matches the business rhythm of per-product data updates, avoids missing batches from manual triggering |
variables | Bind the nodeId of the single product and core test fields | Enables cross-node transfer of single product data, avoids data confusion |
| Node Timeout | 600 seconds | Adapts to the time required for high-voltage test report upload and partial discharge data analysis |
| Knowledge Base Association Scope | First 8 documents | Covers the core documents required for full traceability of a single product, avoids excessive context load |
| Error Retry Count | 2 times | Balances process stability and execution efficiency, adapts to scenarios where device logs occasionally fail to read |
| Single Product Data Filtering | Match by nameplate serial number | Accurately locates full traceability data for the corresponding product, avoids interference from cross-product data |
Three Common Mistakes
- Redundant AI replies after a tool call node runs. The symptom is unexpected natural language replies appearing in node logs. The cause is that the node's automatic reply configuration was not disabled, leading to an extra reply being generated after the tool call.
- Code execution node output cannot be passed to the specified reply node. The symptom is that no code execution return fields appear in the logs of the specified reply node. The cause is that the output variable of the code node was not bound as the input for subsequent nodes via the
variablesconfiguration item. - Environment variable configuration failure prevents test data stored in minio from being read. The symptom is a storage connection error when the node runs. The cause is that the environment variable naming rules were not adjusted per the update requirements of FastGPT v4.17.0, causing minio connection parameters to not be correctly recognized.
How to Confirm Proper Configuration
- Manually trigger a workflow for the corresponding production batch, check that only the
nodeIdand core test fields of the single product are passed in the node logs, with no redundant content. - After running the code execution node, verify that the variables configured via
variablescorrectly carry data returned by the code, such as partial discharge amplitude and withstand voltage values. - Execute the knowledge base retrieval node, check that returned results include documents such as the insulating oil report and vacuum drying records for the corresponding batch.
- Simulate an abnormal exit of the code execution node, check that the workflow performs automatic retries according to the configured retry count.
The values provided here are common starting points, and should be evaluated against the reader's own samples.