Creating the system for choosing the solution from a series of alternatives

A major vertically-integrated company

Deadlines

2021 – today

Team

~20 persons

Technologies

• Java: Spring
• Vert.x
• Python: Jango
• Pandas
• SkiPy
• Apache NiFi
• Dgraph
• Postgres
• JS React

Problem:

• The managers need to choose an optimal solution from a series of alternatives when planning major projects*. The optimal option is the one leading to a stable growth of the Company’s profits. To assess the options based on many inter-connected parameters and uncertainties, a range of values for a series of parameters is used. Without automation the number of options for review is limited, which deteriorates the accuracy and quality of decision-making

Goal

• Create the AI-based platform for decision-making — a system allowing for quick, cheap and reasonable assessment of a broad variety of strategic alternatives at the very early stage of a major project

Results

• The decision-making support system was developed; it includes developing the math simulation scenarios, forming the analysis nodes for the dependency tree, building the calculation logic of low-code Python nodes, Y-trees, building a decision-making map using the uncertainties and cartographic data
• The system’s modules include:
  • Multi-task operation. Defining the options for the efficient plan of the Company operation through different scenarios
  • Ontology-based graph database. Automatically and by way of using AI helps to form the calculation graphs from uncertainty nodes, formulae and target functions, decreasing the time for engineering by 70%
  • Calculation module. A set of adjusted math algorithms and a tool for defining the users’ algorithms used in the dependency tree nodes.
  • Visualization. Reflecting the results of Trade-Off calculations, Tornado diagram, benchmarking in the format of graphs, matrices and decision-making maps

• by 30%

  The time for decision-making decreased. Quality of forecasts improved due to creating digital twins of projects

• by 60%

  Time for preparation and approval of IP** program and optimal development concept decreased

• 5

  Customer’s subsidiaries are using the system functionality

• by 30%

  The efficiency of the asset development strategy selected at the early stage improved

* Major project is a targeted program, which is socially meaningful, requires high level of effort and long time for implementation

** IP means industrial pilots