Warehouse Storage Position Recommendation Using Demand Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current warehousing methods lack an efficient way to determine optimal storage positions for products based on their attributes and inventory dynamics, leading to suboptimal shelving costs and logistics in warehouses.
Innovation Solution
A method and apparatus that utilize a pre-trained current-period outbound quantity prediction model to generate a characteristic vector from product attributes, calculate shelving costs, and determine a central storage position, ultimately recommending storage positions that minimize shelving costs by considering inventory levels and path lengths within the warehouse.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If classified storage is used to place articles of the same category in the same warehouse, then product location can be located conveniently and storage environment can be arranged according to product attributes, but storage position optimization based on demand prediction and inventory dynamics is not achieved
Solution Approach 1:
The patent transforms the static classified storage approach into a dynamic optimization system by introducing characteristic vectors that capture product attributes, inventory levels, and demand predictions. The system continuously updates storage recommendations by changing parameters such as predicted outbound quantity, inventory turnover rate, and shelving cost coefficients, allowing the storage layout to adapt dynamically while maintaining categorical organization benefits
Solution Approach 2:
The system implements feedback mechanisms by using predicted outbound quantities and actual inventory data to continuously refine storage position recommendations. The optimization model receives feedback from inventory dynamics and demand predictions, adjusting storage assignments to minimize shelving costs while maintaining the organizational structure that provides location convenience
2Productivity
If random storage is used to directly place newly-arrived articles to the nearest or random available shelf, then storage space utilization is improved, but shelving costs are not optimized considering product attributes and demand
Solution Approach 1:
The patent moves beyond simple random or nearest-shelf placement by introducing multiple dynamic parameters including predicted outbound quantity, product volume, current inventory levels, and shelving cost coefficients. These parameters transform the storage assignment from a geometric problem to a multi-dimensional optimization problem that balances space utilization with cost efficiency
Solution Approach 2:
The system performs preliminary actions by pre-calculating characteristic vectors for products and pre-training prediction models for outbound quantity forecasting. These preliminary computations enable the system to make informed storage decisions that optimize both space utilization and shelving costs, rather than making simple reactive placement decisions
3Productivity
If associative storage layout is used to place products that are often delivered simultaneously in close positions, then delivery efficiency is improved, but optimal storage positions cannot be determined based on predicted demand and inventory levels
Solution Approach 1:
The patent enhances associative storage by incorporating dynamic parameters such as predicted outbound quantity, inventory turnover rate, and shelving cost coefficients into the storage position determination. Products that are frequently delivered together are associated not only by historical delivery patterns but also by real-time demand predictions and inventory levels, optimizing both delivery efficiency and resource utilization
Solution Approach 2:
The system uses feedback from demand prediction models and inventory monitoring to continuously refine associative storage assignments. The optimization model receives feedback on actual versus predicted outbound quantities, adjusting storage positions to maintain delivery efficiency while adapting to changing demand patterns and inventory dynamics
4Device complexity
If traditional storage methods are used without demand prediction, then system complexity is reduced, but shelving costs cannot be minimized based on predicted outbound quantity and inventory dynamics
Solution Approach 1:
The patent introduces key parameters including predicted outbound quantity, inventory turnover rate, and shelving cost coefficients to transform traditional storage methods into an optimization system. These parameter changes enable the system to calculate optimal storage positions that minimize shelving costs by considering both current inventory levels and forecasted demand, achieving cost efficiency through structured complexity
Solution Approach 2:
The system replaces traditional mechanical or manual storage assignment methods with data-driven optimization algorithms. Instead of relying on physical inspection or manual decision-making, the system uses prediction models and optimization calculations to automatically determine optimal storage positions, reducing operational complexity while improving cost efficiency
Data Source
AI summary
In the method according to embodiments, a characteristic vector of a preset dimension is generated based on attribute information of the to-be-shelved product in each attribute of a preset attribute set, and the generated characteristic vector is imported into a pre-trained current-period outbound quantity prediction model to obtain a predicted current-period outbound quantity of the to-be-shelved product; a shelving cost of the shelving task is determined based on a current inventory of the to-be-shelved product in the target warehouse, the to-be-shelved number, a volume of the to-be-shelved product, and the predicted current-period outbound quantity of the to-be-shelved product; a central storage position of the shelving task is determined based on the shelving cost of the shelving task, a predetermined product shelving cost table and a predetermined storage position shelving path length table; and a recommended storage position information set corresponding to the shelving task is generated and outputted.