PROCESS R&D · ANALYSIS · OPTIMIZATION

From process datato evidence-based R&D decisions.

Open-source Process R&D and Optimization System. Organize recipe versions, experiment records, and run evidence to support quality analysis, process diagnosis, and recipe optimization.

INGOT / EVIDENCE LOOP

Every decision traces back to run evidence.

  1. 01
    Recipe version

    Objectives, settings, safety boundaries

  2. 02
    Run

    Actual settings, trajectories, context

  3. 03
    Quality

    Inspection outcomes and provenance

  4. 04
    Evidence

    Comparable runs, differences, candidate causes

  5. 05
    Correction

    Engineer decisions and the next validation

Next same-version run → inspection outcomes → reassess

Workflow illustration · Not the actual interface · No outcome data

Traceable evidenceTestable causesReviewable recommendationsReusable conclusions

FROM DATA TO DECISION

Bring research records, process analysis, and recipe decisions into one workflow.

Organize objectives, experiment records, and engineering judgment by recipe version. The system directly links actual settings, process context, and quality outcomes from completed runs without manual recipe reclassification. Compare differences and check evidence before reviewing recommendations and recipe revisions.

LINKED RECORDS01 / 04

Organize research records

Recipe version

Published settings · objectives · adjustable scope

Run record

Applied version · actual settings · process context

Experiment facts

Retained in run and quality records, without separate experiment entry

Workflow illustration · Not the actual interface · No outcome data

Current capabilities and limits

ENGINEER IN THE LOOP

The system organizes facts and recommendations; engineers make R&D decisions.

Recipe versions retain parameter settings; run and quality records retain actual process data and outcomes. Analysis explains differences and evidence boundaries, and recommendations preserve inputs and rationale. Engineers review adoption and corrections, then use later outcomes to reassess their decisions.

01

Define

question · variables · boundaries

02

Record

settings · trajectories · outcomes

03

Qualify

quality · provenance · completeness

04

Diagnose

compare · candidates · evidence

05

Recommend

objectives · constraints · uncertainty

06

Review

decisions · revisions · later outcomes

THE ENGINEERING TOOLBOX

Confirm that data are trustworthy before reviewing the next correction.

The recommender checks whether production runs are complete, comparable, and linked to quality outcomes before forming a candidate within declared variables, safety boundaries, and observed coverage. Engineers review recommendations; the system never changes production parameters automatically.

RUNSSTATISTICSMODELS
DATA

Confirm data usability

Check completeness, actual values, units, time, and provenance, together with version changes and applicability.

COMPARE

Process diagnosis

Use matching, robust statistics, stage trajectories, and context stratification to narrow candidates.

TEST

Preserve mechanism notes

Attach parameter effects, known boundaries, and engineering judgment to a specific recipe version with run, quality, and reviewed process-document references.

REVISE

Review the next correction

Small corrections stay on the current recipe version; the next same-version run links automatically. Significant changes use proposed settings to create a revision draft with provenance and rationale.

Data qualityReal runsVersion lineageFragment citationsReviewed knowledgeEngineering decisions

RECORDS · ANALYSIS · DECISIONS

Research records, analysis, and decisions have clear responsibilities.

The Web app, API, database, and optimizer form the runtime stack; connectors are optional data sources. Formal records retain provenance, versions, and review status, and analysis and recommendations reference those facts. Engineers confirm parameter changes and applicability.

01RESEARCH RECORDSrecipes · variables · boundaries

Organize objectives, settings, adjustable scope, and applicability

02RUNS AND QUALITYsettings · trajectories · inspections

Link run context and quality outcomes; check versions, completeness, and review status

03ANALYSIS AND RECOMMENDATIONScomparison · diagnosis · optimization

Organize candidate causes and recipe recommendations constrained by evidence and safety boundaries

04ENGINEERING DECISIONSreview · versions · knowledge

Retain adoption, revision, or rejection rationale and reference reviewed process material

STABLE CORE, EVOLVING METHODS

Process capabilities evolve. Evidence boundaries remain fixed.

A change in machine, product, or process requires new mappings, variables, boundaries, and context. Run identity, evidence principles, recipe version lineage, and engineering authority remain stable.

01

Stays stable

Real data supports engineering judgment, and every conclusion traces to sources and remains testable

02

Configured per scenario

Equipment mappings, variables, stages, quality boundaries, context, and mechanism notes

03

Continues evolving

Statistics, diagnostic strategies, page layouts, and language models

RUN IT YOURSELF

Open-source Process R&D and Optimization System.

Ingot is Apache-2.0 licensed and self-hostable inside the plant. Source code, deployment instructions, and method boundaries are available for inspection; scenario-specific evaluation belongs to the deployer's own data.

Current maturity

Each process run is an experiment. Recipe versions hold parameter settings, and run records link actual parameters, process data, and quality outcomes. Admitted run evidence feeds recipe recommendations. Actual runs, parameter readback, and inspection records freeze the final outcome once. Core software workflows have automated tests; real-factory benefit validation remains incomplete. Default Compose provides a single-host reference deployment; deployers still own site security, recovery, capacity, and operational acceptance.

QUICKSTART
git clone https://github.com/liuweichaox/Ingot.git
cd Ingot
cp .env.example .env
# Edit .env: replace every change-this placeholder
docker compose -f docker-compose.app.yml config --quiet
docker compose -f docker-compose.app.yml up -d --build

START WITH ONE REAL DATA LOOP

Begin with a process question.

Publish a recipe version with variables and boundaries, link run and quality records, or start with existing runs to compare differences and review the next correction.