事例Case study

LogiFlex

AI service for logistics load optimization

Duration6 months

AILogisticsManufacturing

Logistics

+7%

goods turnover growth

+7%

goods turnover growth

概要About the project01

We developed an AI framework based on generative learning that optimizes how goods are loaded into trucks. The system takes an order from 1C and outputs a ready-to-use PDF loading plan. It accounts for each item’s stacking requirements, truck axle loads, and other physical parameters.

課題Key challenge02

A multi-parameter optimization problem with dozens of simultaneous constraints. Classical approaches took hours to produce a result. LogiFlex computes an optimal arrangement (94% of the best physically possible one) in about 10 seconds.

要点The solution

The solution is built on a genetic algorithm that treats each sequence of goods as an individual in the population. Within seconds, a surrogate mathematical model compares more than 10,000 possible arrangements, scoring them on center-of-gravity distribution and the physical compatibility of goods (for example, whether light items can go on top of heavy ones and vice versa). The output is a set of “crossed-over” top candidates. After several rounds of “evolution” and more than 20 analysis and processing techniques, LogiFlex produces the best arrangement, one that satisfies most of the parameters professional logisticians rely on.

工程Process03

0101
Logistics analysis
Studying warehouse structure, product categories, and scenarios.
0202
Constraint model
Defining loading parameters: dimensions, weight, product compatibility, and more.
0303
AI algorithm development
Building the genetic algorithm and the mathematical optimization system.
0404
ERP integration
Connecting the system to the company’s internal data.
0505
Simulation & testing
Modeling different loading scenarios and how the system behaves under load.
0606
Launch & analytics
Deploying the AI service to production and analyzing its impact on goods turnover and logistics.