Learn about the technology behind our solutions

Overview

 

VAYU Sense brings an innovative platform to pharma and food Tech production. With the use of cloud services and AI, VAYU Sense’s technology improves processes performance.

VAYU Sense’s main solution, Athena closed-loop controller (I.C), is an autonomous cloud service that controls the process in real-time. Athena is a process custom AI agent, that is trained offline to navigate the process through the best working points, based only on the process archived data.

The fact that the algorithm is designed to be generic, without taking any pre-assumption, Athena can serve all types of processes like: High cell density, API, Mammalian, etc.

VAYU Sense’s platform offers two additional complimentary services:

Cassandra – an AI-based surveillance service that alerts in cases anomaly is detected.

Aletheia – a monitoring service that enables tracking the process from every platform all the time.

Get an insight from our experts

How it works

From state to action

Athena closed-loop controller is synchronized with the process and starts working together with the batch initialization. Athena continuously reads the process monitored parameters (for example: pH, DO, CO2, etc.), analyzes them, and as a reaction changes the values of the controlled parameters (for example: agitation, feeding etc.).

Definition: A combination of monitored values called ‘State’, and a combination of values of controlled parameters called ‘Action’. Both sets of monitored and controlled parameters are defined by the customer.

Figure 1 – Athena converting State into Action

The customer’s goals for process improvement define Athena’s target for improvement. Athena offline learns how to convert State into Action to reach the target for improvement. The outcome of the learning phase is a controller that changes in real-time the values of the controlled parameters according to the values of the monitored parameters.

Athena solution structure

The Athena controller is implemented as a standard client/server application.

The Client – a local service installed in the customer site; integrated to the control system. It reads the process’s State from the control system, sends a request to Athena’s cloud service, and writes back the Action received from Athena.

The Server – a service running in the cloud environment. It receives requests containing the process state from the client, calculates the Action, and sends it as a response to the client.

Figure 2 – Athena Client/Server Architecture

Solution Customization

The Athena controller algorithm is based only on data collected from a specific process; no pre-assumptions are used. The data that is used can come from two sources:

  1. Historical data of the process archive.
  2. Data collected from designated experiments performed for the learning purpose.

Figure 3 – Athena controller training flow

Figure 4 – Example for improvement using Athena training outcome

Iterative development using VAYU Sense Experiment Generator

Since an experiment is an expensive resource, the number of experiments executed to create the data set to be used for training should be minimal. To minimize the number of experiments to execute, a smart DOE (design of experiments) is required.

VAYU Sense offers to use its smart AI-based tool, the Experiment Generator, to be used during the data collection phase, to plan the next optimal experiment to be executed in terms of process performance improvement.

By using the Experiment Generator after each experiment, the training of Athena can extract the most from the next experiment so that its data will be added to the collection.

Figure 5 – Athena iterative development using the Experiment Generator

VAYU Meter

The VAYU Meter is an optional hardware feature that measures non-invasively CO2 and O2 in MidIR spectroscopy. It is designed to work in harmony with Athena.

  • Dynamic range: 0-100%
  • Relative measurement accuracy ±3%
  • Connected to the exhaust pipe of your stainless steel or disposable fermenter.
  • Guaranteed 24/7 functionality

    IT Security

    Isolation of Athena from production network

    Athena offers a solution for isolating the production network from the internet. The isolation is achieved by splitting the local agent (local service) to two separate components:

    Backend – connected to the cloud service via internet.
    Frontend – connected to the customer’s control system on the local production network (usually via OPC).

    The Backend and Frontend communicates via encrypted protocol over serial-port connection.

    Figure 6 – Ahtena designed to meet IT security requirements

    Data encryption

    The communication between the backend and the cloud service is implemented with REST protocol over HTTPS, and with authentication based on URL + API_KEY.

    Security standards

    • Data traffic TLS encrypted
    • Services are not exposed to the outer world, access through Reverse Proxy
    • No remote access to databases
    • Private IPs never disclosed to outside world
    • HTTPS communication

    Cyber security

    • Design according to Domain Driven methodology
    • CDC (Change Data Capture)

    High availability

    • Availability with Healthy check (Liveness, Readness)
    • Scalability by K8S cluster (Node Pool)
    • External load balancer

    Monitoring and alerting

    • Real-time monitoring with ELK
    • Services monitoring logs via Stack Driver
    • Livness – Spring Boot Admin

    Enterprise data

    • VPC peering for data access
    • Entire source code maintained on private Git which is certified according to ISO/IEC 2700

    From state to action

    Athena closed-loop controller is synchronized with the process and starts working together with the batch initialization. Athena continuously reads the process monitored parameters (for example: pH, DO, CO2, etc.), analyzes them, and as a reaction changes the values of the controlled parameters (for example: agitation, feeding etc.). Definition: A combination of monitored values called ‘State’, and a combination of values of controlled parameters called ‘Action’. Both sets of monitored and controlled parameters are defined by the customer.

    Figure 1 – Athena converting State into Action

    The customer’s goals for process improvement define Athena’s target for improvement. Athena offline learns how to convert State into Action to reach the target for improvement. The outcome of the learning phase is a controller that changes in Real-Time the values of the controlled parameters according to the values of the monitored parameters.

    Athena solution structure

    The Athena controller is implemented as a standard client/server application. The Client – a local service installed in the customer site; integrated to the control system. It reads the process’s State from the control system, sends a request to Athena’s cloud service, and writes back the Action received from Athena. The Server – a service running in the cloud environment. It receives requests containing the process state from the client, calculates the Action, and sends it as a response to the client.

    Figure 2 – Athena Client/Server Architecture

    Solution Customization

    The Athena controller algorithm is based only on data collected from a specific process; no pre-assumptions are used. The data that is used can come from two sources:

    1. Historical data of the process archive.
    2. Data collected from designated experiments performed for the learning purpose.

    Figure 3 – Athena controller training flow

    Figure 4 – Examples for improvement using Athena training outcome

    Iterative development using VAYU Sense Experiment Generator

    Since an experiment is an expensive resource, the number of experiments executed to create the data set to be used for training should be minimal. To minimize the number of experiments to execute, a smart DOE (design of experiments) is required.

    VAYU Sense offers to use its smart AI-based tool, the Experiment Generator, to be used during the data collection phase, to plan the next optimal experiment to be executed in terms of process performance improvement.

    By using the Experiment Generator after each experiment, the training of Athena can extract the most from the next experiment so that its data will be added to the collection.

    Figure 5 – Athena iterative development using the Experiment Generator

    VAYU Meter

    The VAYU Meter is an optional hardware feature that measures non-invasively CO2 and O2 in MidIR spectroscopy. It is designed to work in harmony with Athena.

    • Dynamic range: 0-100%
    • Relative measurement accuracy ±3%
    • Connected to the exhaust pipe of your stainless steel or disposable fermenter.
    • Guaranteed 24/7 functionality

      IT Security

      Isolation of Athena from production network

      Athena offers a solution for isolating the production network from the internet. The isolation is achieved by splitting the local agent (local service) to two separate components:

      Backend – connected to the cloud service via internet.
      Frontend – connected to the customer’s control system on the local production network (usually via OPC).

      The Backend and Frontend communicates via encrypted protocol over serial-port connection.

      Figure 6 – Ahtena designed to meet IT security requirements

      Data encryption

      The communication between the backend and the cloud service is implemented with REST protocol over HTTPS, and with authentication based on URL + API_KEY.

      Security standards

      • Data traffic TLS encrypted
      • Services are not exposed to the outer world, access through Reverse Proxy
      • No remote access to databases
      • Private IPs never disclosed to outside world
      • HTTPS communication

      Cyber security

      • Design according to Domain Driven methodology
      • CDC (Change Data Capture)

      High availability

      • Availability with Healthy check (Liveness, Readness)
      • Scalability by K8S cluster (Node Pool)
      • External load balancer

      Monitoring and alerting

      • Real-time monitoring with ELK
      • Services monitoring logs via Stack Driver
      • Livness – Spring Boot Admin

      Enterprise data

      • VPC peering for data access
      • Entire source code maintained on private Git which is certified according to ISO/IEC 2700

      Artificial intelligence needs real people

      VAYU Sense is more than just technology: intelligence, experience and wholehearted commitment from our team bring bits and bytes to life and ensures that our customers can benefit from artificial intelligence.

      KI-Bundesverband Logo

      The KI quality seal of the KI Bundesverband e.V. pursues the goal of a people-centered and human-beneficial use of Artificial Intelligence. By defining and adhering to an overarching understanding of values and processes, the seal ensures ethical service and product de- velopment. In the center stand quality criteria ethics, impartiality, transparency, as well as security and data protection.

       

       

       

      Artificial intelligence needs real people

      VAYU Sense is more than just technology: intelligence, experience and wholehearted commitment from our team bring bits and bytes to life and ensures that our customers can benefit from artificial intelligence.

       

       

       

      The KI Bundesverband is Germany’s largest network of companies operating in the field of AI. The members of the KI Bundesverband are committed to ensuring that this technology is applied in the spirit of European and democratic values and that Europe achieves digital sovereignty. The association connects “Europe’s brightest minds” with the aim of strengthening innovative power, encouraging openness for AI innovations among European companies and strengthening companies for global competition.