What This Category’s Data Looks Like
The data for this category primarily comes from wafer incoming material ledgers, workshop MES systems, operation logs of diffusion furnaces, coating equipment, and printers, EL and IV test reports, and laboratory process inspection data. Daily documents include process cards, diffusion and coating parameter tables, printing and sintering process records, EL inspection images, IV grading data, and batch yield reports. Data updates occur in sync with each production batch, with real-time updates for single-wafer testing, and daily aggregation of production and quality data. Fields cover process parameters, equipment operating status, single-wafer test results, batch traceability identifiers, and more.
What Constraints These Characteristics Impose on Workflow Orchestration
Dispersed data sources and varying update rhythms require workflows to support scheduled pulling and real-time synchronization of multi-source data. High-frequency updates of single-wafer-level test data require workflow nodes to support fine-grained parameter matching and result verification. Mixed calls of multiple document types require workflows to configure different nodes for processing structured parameter tables and unstructured inspection images. The batch traceability business requirement requires workflows to embed context transfer logic associated with variables and nodeId, ensuring each operation can be bound to the corresponding production batch. Additionally, the equipment inspection and maintenance scenario requires rapid response to anomalies, so workflows must preset trigger conditions for exception branches to adapt to exception judgment rules for different processes.
Configuration Settings
This document is written based on FastGPT v4.17.0.
| Configuration Item | Recommended Approach | Rationale |
|---|---|---|
| Global variable configuration | Bind production batch ID and equipment number as variables | Adapt to the batch traceability requirements of photovoltaic cell manufacturing, ensuring each round of Q&A is linked to corresponding production data |
| Workflow trigger cycle | 10–30 minutes | Match the update rhythm of equipment operation logs and test data, avoid invalid calls |
| Context window size | First 800–1200 characters | Adapt to the content length of single documents such as process cards and EL inspection images, retain key parameters and anomaly information |
| Exception branch trigger rules | Match EL anomaly markers and IV efficiency decline thresholds | Based on the consequences of errors in this category, quickly trigger the maintenance Q&A process |
| Historical record storage | Archive by production batch, retain for 7 days | Align with data update frequency, facilitate traceability of inspection and maintenance historical operations |
| Node jump configuration | Associate nodeId to bind process nodes | Ensure the workflow executes in the order of photovoltaic cell manufacturing processes, avoid process chaos |
Three Common Configuration Mistakes
- Unconfigured GPT-4o-mini call error logs appear during workflow runtime. The cause is failing to turn off the global default model call switch, which causes workflow nodes to automatically trigger unconfigured model requests.
- Global variable historical records in the workflow cannot be modified. The cause is failing to enable variable editing permissions for the workflow, or failing to bind editable historical record fields through the
variablesconfiguration item. - User-uploaded attachment-style inspection materials cannot be obtained through workflow nodes. The cause is failing to configure the receiving type and storage path of the file upload node, and failing to associate the corresponding data extraction rules.
How to Confirm Proper Workflow Configuration
- Manually trigger the workflow, check whether the trigger node starts per the preset cycle or event, and confirm no additional unconfigured model call logs are present.
- Access the global variable configuration page, verify that
variablesincluding production batch ID and equipment number are correctly bound. Variable values can be viewed via test nodes. - Simulate an exception scenario to trigger the preset exception branch, confirm whether the workflow jumps to the maintenance Q&A node.
- Upload test EL inspection images or process parameter documents, and verify that the context window correctly intercepts content meeting business requirements.
The values provided are common starting points and should be validated against samples relevant to the specific deployment.