Scenario Analysis and Configuration Diagram Analysis
What most data analysis tools solve is "what to do once you have the data". But in real projects, the hardest questions often appear one step earlier:
- Should this production line be modeled at all?
- Which variables should be collected? Which are X, and which is Y?
- How can the point list in the heads of the veteran engineers on the shop floor be turned into a table quickly?
Scenario Analysis and Configuration Diagram Analysis are designed exactly for this——they serve the stage before modeling and help you think clearly about "is it worth doing, and what data should be collected".
🚪 Entry
The green "Scenario" button in the top toolbar (list icon) → expands a three-item menu:
| Menu Item | Function |
|---|---|
| Scenario Analysis | Describe your business in text and generate a scoring table and a data template |
| Configuration Diagram Analysis | Upload an image (configuration diagram / DCS screenshot) and identify the parameters in it |
| Analysis Report History | View, export, and delete historical reports |
💡 These two features do not depend on any existing model, and do not require importing data first; they are available at any time.
📋 Scenario Analysis
What Problem Does It Solve
Facing a production line that has never been modeled, the debate engineering teams most often get stuck in is "which scenario should we do first". Scenario Analysis uses a structured scoring logic to turn that debate into comparable numbers.
Steps
Click Scenario Analysis
Choose the document generation language (Chinese / English)
Enter a scenario description in the text box (required):
Please enter a detailed description of your industry, scenario, equipment, process, etc...
Click Start Analysis
⚠️ The scenario description is required——when it is not filled in, the "Start Analysis" button is not clickable. The more specific the description (industry, equipment, process, target product quality indicators), the more closely the output matches reality.
Analysis Process
- The report is generated section by section in a streaming manner and automatically scrolls to the bottom; there is no need to wait for it all to finish
- The official recommendation is to allow 30~60 seconds
- If it fails midway, a clear error message is given
Output
When the analysis is complete, three things are produced at the same time:
| Output | Content |
|---|---|
| Main analysis report | A structured interpretation of the scenario and recommendations for moving forward |
| Scenario priority scoring table | Scored on the two dimensions of "business value × data availability", directly telling you what to do first |
| Data collection template | Clearly lists which variables to collect, which are X, and which is Y; can be handed directly to the shop floor |
Export
Two export buttons are provided at the bottom of the report:
- Export Markdown —— Suitable for archiving and pasting into a knowledge base
- Download Excel —— Exports the scoring table and data template as a multi-worksheet Excel file that can be sent directly to the shop floor for data entry
💡 This is the most valuable part of Scenario Analysis: what you get is not an "AI-generated article", but a blank form you can take straight out to collect data.
🖼️ Configuration Diagram Analysis
What Problem Does It Solve
In a factory, the variable list often exists in the form of a configuration diagram / DCS screen rather than an Excel file. The traditional approach is to have an engineer copy them from the screen one by one by hand——slow, error-prone, and dependent on individual experience.
Configuration Diagram Analysis uses a multimodal large language model to "read the image and recognize the parameters" directly.
Steps
Click Configuration Diagram Analysis
Choose the select language / Language option (Chinese / English)
Upload images:
- Drag and drop or click to upload is supported
- Up to 3 images
- Applicable to: DCS screens, configuration diagrams, business flow diagrams
(Optional) Fill in a scenario description related to the image
Please enter a scenario description related to the image (optional)...
Click Start Analysis
💡 Uploading clear images helps the AI identify parameters and flows accurately. When taking a screenshot, try to include complete variable names and units.
Output
| Output | Content |
|---|---|
| Main analysis report | A structured interpretation of the flows and parameters in the image |
| In-image parameter table | The list of points/parameters extracted from the image |
| Data template | A fillable data table generated from the recognition results |
Export
Export Markdown and Download Excel are supported as well (the parameter table and data template are multi-worksheet).
⚠️ A multimodal model must be used: a text-only model cannot read images. Please choose a model with vision capability in Settings → LLM Configuration.
📚 Analysis Report History
The reports produced by both features are automatically retained in Scenario → Analysis Report History.
The report list supports:
- View report content: Includes the body, the recorded business scenario description, the underlying large language model used, and the generation time
- Export Markdown
- Download Excel: Exports the tabular data in the report as a multi-worksheet Excel file
- Delete: Clean up reports you no longer need
💡 The report records the model used at the time and the scenario description, making it easy to review later "under what premises this conclusion was reached".
🎯 Typical Workflow
Evaluating a new production line before project approval
Scenario Analysis (list candidate scenarios + score them)
↓
Compare the scoring table → choose the highest-priority scenario
↓
Configuration Diagram Analysis (take DCS screenshots → get the point parameter table)
↓
Collect historical data according to the data template
↓
AI Auto Modeling / Manual modeling
↓
AI Cleaning → produce a diagnosis reportThis is the complete implementation path recommended by the platform: first use AI to think clearly about "what to do and what to collect", then use AI to accomplish "how to model and how to clean".
⚠️ Usage Notes
- Both analyses require the large language model to be configured first (Settings → LLM Configuration)
- token consumption is incurred: the reports are fairly long, and Configuration Diagram Analysis has to process images, so the cost is higher than text-only analysis
- AI output needs human review: parameter names and units recognized from the image may be misread; please check them against the original image before use
- Images are not persisted: uploaded images are used only for this analysis request and are not saved as files in the instance
🔗 Related Reading
- AI Capabilities Overview —— All AI features and LLM configuration instructions
- Factory Scenario Data Modeling Guide —— The complete methodology for scenario identification, variable classification, and modeling
- Step 1: Data Preparation —— How to import and configure data once you have it