HOME · INDUSTRIAL AI · ARYAVAKAV

From raw plant data
to the right decision

Vakav is the AI department of AryaVakav, a knowledge-based company that builds industrial control and instrumentation systems. On the same control-room data, we build models that predict quality, recommend optimal settings and optimize the process. Everything runs inside your organization, with no internet dependency.

01 · THE PROBLEM

The data is there.
Decisions arrive late.

Your control system records thousands of values every second, and most of them stay in the archive. The challenges we see again and again in industry:

  • Predicting product quality before lab results arrive
  • Controlling the process in special conditions without trial and error
  • High energy consumption and emissions
  • Control-system data that goes unused
  • No coherent approach to data analysis and engineering
DCS · HISTORIAN · LAST 24HNot analysed
WTH15 · FEED204.6 t/h
PROCESS GAS948 Nm³/t
BUSTLE GAS377 °C
02 · HOW VAKAV THINKS

Three steps to a decision

Every Vakav solution is built on the same three levels. Data enters at the first step and becomes a sharper decision at each one.

FORECAST · 8H
LEVEL 1 · PREDICTION

Prediction

Product quality and process behaviour, online, before lab results arrive.

SETPOINT · RANGE
LEVEL 2 · RECOMMENDATION

Recommendation

The optimal range for every parameter, on the control-room display, next to the operator.

COST · MINIMUM
LEVEL 3 · OPTIMIZATION

Optimization

Connecting modules together to bring the whole process to its optimum.

03 · PLATFORM

VAI, Vakav's industrial AI platform

VAI is installed on a server inside your plant and connects directly to the DCS, databases, files and IoT devices. Data is analysed where it lives and results appear on the control-room display. If the operator chooses, results are written back to the DCS.

OPC UAMODBUSMQTTHISTORIANAIR-GAPPED
BUILD · RUN

Every model, a live workflow

Data arrives from the DCS and Historian, is cleaned and synchronised, reaches the prediction model, and returns to the control room as recommended settings and quality alarms. Scripts, workflows, schedules and apps, all in one environment.

PYTHONFLOWSCHEDULEAPP BUILDER
VAI · DRI-QUALITY-PIPELINESCRIPTFLOWAPPRunning
UA
DCSOPC UA · 248 TAGS
H
Historian24-hour archive
fx
PreprocessClean · sync
ML
MD% prediction91.55
L2
Recommended settingsTo the control-room display
!
Quality alarmIF MD < 91.0
6 NODES · 5 LINKSLAST RUN 0.82 S
04 · SOLUTION · STEEL

DRI Expert
DRI quality, before the lab

A module built on VAI for the Midrex direct-reduction process. It predicts product metallization and carbon online and recommends the optimal range for process parameters to the operator.

  • Online data from the DCS, or Historian exports and archives
  • Cleaning, synchronisation and feature selection
  • MD and C prediction with recommended settings
DRI EXPERT · MIDREXOnline
MD %91.62 / 91.55Actual / predicted
C %2.16 / 2.18Actual / predicted
FORECASTNOWMEASURED
WTH15 · feed rate193 – 216 t/h
Process gas flow917 – 981 Nm³/t
Bustle gas temperature362 – 390 °C
SAMPLE VALUES · DRI EXPERT DASHBOARD
05 · SOLUTION · VISION

Machine vision
for engineering drawings

Thousands of instrument tags in PDF drawings are detected with OCR and automatically matched against Excel lists. Processing runs on a GPU server inside your organization, and no drawing ever leaves it.

  • Smart tag detection in PDF files with OCR
  • Automatic matching with Excel data and a mismatch report
  • Annotated PDF and Excel report as output

The product's final name and hardware specifications are to be confirmed.

P&ID-104.PDF · OCR · GPU4 tags · 1 mismatch
INSTRUMENT-INDEX.XLSX
TT-2031Matched
PT-1102Matched
FT-3310Matched
LT-1250Mismatch with drawing
06 · KNOWLEDGE AI

Vakav Autonomous
a fully offline knowledge assistant

A ChatGPT-like experience that runs on your organization's own servers. It reads procedures, process diagrams and technical documents, and answers with exact source citations.

RAGPDF · WORD · EXCEL · AUDIOCITATIONS
OFFLINE BY DESIGN

Your data never leaves

Knowledge flows from the documents in the same workspace to the answer, and nothing crosses the organization's network boundary. Separate workspaces for each unit and access levels for each user.

ORGANIZATION NETWORK · NO DATA LEAVES
VAKAV AUTONOMOUSLOCAL LLM · OFFLINE
What changed in the new safety procedure for the DR unit?
Three main changes: revised allowable gas temperature ranges, an approval step before restart, and an updated list of personal protective equipment.
SAFETY-MANUAL-1403.PDF · P.12SAFETY-MANUAL-1403.PDF · P.31
Ask a question…ON-PREM
07 · ENGAGEMENT

From assessment to operation

We start with an industrial pilot on a real problem in your own environment and move step by step to full operation.

DATA AUDIT
STEP 1

Needs assessment

Understanding the process, the available data and its gaps, and a measurable goal.

SOLUTION DESIGN
STEP 2

Design & development

Designing modules for each unit and integrating data end to end.

ON-PREM SERVER
STEP 3

Installation & testing

Deployment on the plant server and testing under real conditions.

CONTINUOUS GAIN
STEP 4

Training & support

Enabling your team and continuously improving the models.

08 · WHO WE ARE

AI, built by the people who build control systems

Since 2019, AryaVakav has built distributed control systems, safety systems, advanced process controllers and instrumentation equipment, and is part of Control Pouyan Group. Vakav brings this control-room know-how to AI.

EAGLE DCS · CONTROL LOOP TIC-101AUTO
VAKAV AIOptimal setpoint
EAGLE ESDSafety interlock
SP385 °C
EAGLE DCSPID controller
FV-101Control valve
TT-101Temperature transmitter
PROCESS · PVProcess
2019AryaVakav founded
Eagle DCSNative distributed control system
Eagle ESDSafety control system
Control PouyanParent industrial group
09 · START

Let's start with a pilot in your plant

Pick one concrete problem in your process. We'll show, in your own environment, what value industrial AI creates.

info@aryavakav.com+98 21 2637 4300 · +98 21 2665 2854Unit 2, No. 340, Shahid Kolahdouz St., between Varasteh and Fakourian, Tehran, Iran