What data for this category looks like
Data mainly comes from silicon wafer incoming material records, production MES systems, real-time operation data from diffusion furnaces, coating equipment, and printing presses, test results from EL and IV testing equipment, and laboratory test records. Update cycles cover each production batch, each individual product inspection, and daily summaries. Daily documents include process cards, diffusion and coating parameter ledgers, printing and sintering process records, EL defect detection images, IV performance grading data, and yield statistics documents. Fields include process window parameters, defect coordinates, electrical performance values, batch traceability codes, and more, with no unified fixed units as they vary by inspection item.
What constraints these characteristics place on workflow orchestration
Data sources for this category are scattered, and update cycles are frequent. This requires workflows to support multiple data source connections and near-real-time triggering. Diverse document types include structured process parameters and unstructured EL detection images. This requires workflows to adapt to different format input and parsing logic. Strong batch traceability requirements mean workflows must bind batch codes as core variables parameters to achieve full-process data association. Process windows are sensitive, and errors can impact product performance. This requires workflows to embed parameter threshold verification and exception identification nodes to timely intercept process data that deviates from normal ranges.
How to configure the workflow
| Configuration Item | Recommended Approach | Rationale |
|---|---|---|
variables | Bind batch codes, process parameter IDs, and test result IDs | Match the strong batch traceability requirements of photovoltaic cell manufacturing to achieve full-process data association |
| Data source trigger frequency | Trigger upon completion of each batch or each piece inspection | Match the update cycle of each batch and each piece for this category, avoiding invalid triggers |
| Input format adaptation | Enable structured parameter parsing and EL image preprocessing nodes | Adapt to diverse document types such as process parameter ledgers and EL detection images |
| Abnormal verification thresholds | Calibrate based on actual measurements | Match the sensitive characteristics of process windows such as texturing and diffusion, avoiding misjudgment or missed judgment |
| Node timeout setting | 600 seconds | Reserve sufficient time for data pulling and processing when connecting multiple device data sources |
| Single processing data limit | Top 1000 entries | Adapt to the detection data volume of a single batch, avoiding node overload |
nodeId | Assign a unique identifier to each process node | Facilitate workflow debugging and abnormal node location, in compliance with FastGPT v4.17.0 node management rules |
Three common configuration mistakes
- Phenomenon: Console.log logs from code running nodes cannot be viewed in the inspection console. Cause: The workflow's log persistence switch is not enabled, so runtime logs are not saved.
- Phenomenon: A parameter missing error occurs during workflow runtime. Cause: Core fields such as batch codes are not bound in the
variablesconfiguration, so nodes cannot obtain necessary business data. - Phenomenon: No response after the workflow is triggered. Cause: The data source trigger frequency is set to fixed daily triggering, which does not match the batch-by-batch and piece-by-piece update cycle of this category, so the trigger condition is not met.
How to confirm the configuration is complete
- Review the workflow's data source configuration to confirm the trigger rules match the completion nodes of production batches or individual test pieces.
- Check the
variablesconfiguration items to confirm core fields such as batch codes and process parameters required for this category are bound. - Run a test workflow to verify that logs from the code running node are properly output to the specified location.
- Simulate input of abnormal process parameters to confirm the abnormal verification node can trigger the corresponding alarm process.
The values provided are common starting points and should be measured against the reader's own samples.