Data Profile for This Category
This category covers scenarios related to the production of injection molded parts, automotive interior components, home appliance plastics, and medical plastics. Daily documents include injection molding process cards, raw material drying records, mold history logs, first-piece inspection reports, patrol inspection records, and material change and machine cleaning ledgers. Data sources include resin batch property sheets, injection molding machine operating parameters, mold temperature controller data, Manufacturing Execution Systems (MES), and dimensional and appearance inspection results. Updates occur once per production batch, each mold change or material change, and once per shift. Core fields include filling, packing, and cooling window parameters, raw material drying and regrind ratios, defect analysis records, and more. Standard production units such as temperature, pressure, duration, and weight are used.
Constraints for Source Citation and Traceability
High data update frequency and scattered trigger nodes require the traceability system to support quick location of the latest documents by production batch and mold change event. This prevents use of outdated process parameters or records. Diverse document types and clearly subdivided fields require precise matching of the field scope of corresponding documents. This stops invalid citations across document types from interfering with process judgment. Data sources involve multiple devices and systems, so traceability information must link to original data source identifiers. Defect analysis is strongly correlated with production parameters, so specific units and batch information of parameters must be retained during citation. This prevents process adjustment errors caused by missing information. Incorrect citations may lead to batch dimensional and appearance defects, mold damage, or material mixing safety issues. The traceability link must strictly bind data generation time and production events. This ensures the timeliness and accuracy of citations.
Configuration Guidelines
This configuration is developed for FastGPT v4.17.0.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
quoteMaxToken | 800–1200 token | Matches the core content length of single documents such as injection molding process cards and mold history logs, to avoid truncation of critical parameters |
responseData | Only documents related to the current production batch | Aligns with the category’s characteristic of updating data per production batch and mold change event, to narrow the scope of invalid traceability |
detail | Display full document fields and source identifiers | Ensures core production information such as resin batch and injection molding machine parameters can be verified, to guarantee traceability accuracy |
| Single recall limit | Top 6 results | Adapts to scenarios where multiple types of documents are retrieved in parallel, to avoid redundant information interfering with process judgment |
| Matching accuracy threshold | Calibrated via actual testing | As this category’s data has high requirements for parameter accuracy, matching rules must be adjusted based on actual production scenarios |
Three Common Configuration Errors
- Phenomenon: Writing a knowledge base ID or reference variable directly in a prompt word produces expected output, but an error occurs when the variable is applied to the knowledge base configuration. Cause: The system does not complete binding verification between the knowledge base and application variables, leading to variable parsing failure.
- Phenomenon: Retrieving multiple types of documents together results in invalid citations of non-injection molding process documents. Cause: Filter rules based on production batch and document type are not configured, leading to recall of cross-scenario data.
- Phenomenon: Citation results omit specific duration parameters from raw material drying records. Cause: The configuration item for full field display is not enabled, leading to core parameters being truncated or excluded from the citation scope.
How to Verify Successful Configuration
- Initiate a query with specific injection molding parameters, and check if the returned results include the document identifier for the current production batch.
- Review the citation details module to confirm full display of document fields and source information.
- Adjust the single recall limit configuration, and verify that the number of returned results matches the set range.
- Simulate a mold change or material change event, and confirm that the system automatically switches to the document scope corresponding to the target batch.
The values provided here are common starting points, and should be evaluated against samples from the target production environment.