Automated Waste Collection Control for Energy-Aware Emptying Cycles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automated waste collection plants face inefficiencies in energy and economic performance due to inadequate adaptation to plant occupation levels and energy costs, lacking precise modeling and control mechanisms.
Innovation Solution
A computer-implemented method using deep learning and reinforcement learning neural networks to predict energy consumption and optimize waste collection by adjusting vacuum-based systems, incorporating a virtual agent for real-time decision-making based on inlet loading states.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If automated waste collection plants operate continuously to maintain service quality, then service quality is improved, but energy consumption increases
Solution Approach 1:
The system dynamically adjusts the operation of vacuum generators and collection vehicles based on real-time inlet occupation levels. Instead of continuous operation, the system activates vacuum generators only when waste accumulation reaches threshold levels, creating a dynamic on-demand operation mode that adapts to varying waste generation rates across different locations and time periods.
Solution Approach 2:
The system uses sensor data from inlets to automatically determine when collection is needed, eliminating the need for continuous monitoring and manual dispatch decisions. The occupation level sensors enable the system to self-regulate its operation based on actual waste accumulation, triggering collection only when necessary rather than operating on fixed schedules.
2Reliability
If more collection resources are deployed to handle higher occupation levels, then service quality is improved, but operational costs increase
Solution Approach 1:
The system changes operational parameters such as vacuum generator power levels, collection vehicle routing, and dispatch timing based on measured occupation levels. When occupation is low, the system reduces vacuum power or delays collection; when occupation is high, it intensifies collection activity, creating a parameter-based response that matches resource deployment to actual need rather than using fixed resource allocation.
Solution Approach 2:
The system applies different collection strategies to different inlets based on their individual occupation characteristics. High-occupation inlets receive more frequent or intensive collection, while low-occupation inlets are collected less frequently or with lower vacuum power, creating a localized, differentiated approach that optimizes resource use according to spatial variations in waste generation.
3Productivity
If energy-intensive collection operations are performed frequently, then waste removal efficiency is improved, but energy consumption increases
Solution Approach 1:
The system applies partial action by using reduced vacuum power levels for inlets with low occupation levels, rather than always applying full vacuum power. This partial action approach maintains adequate waste removal for low-occupation inlets while significantly reducing energy consumption compared to full-power operation, trading some removal speed for energy efficiency when full capacity is not needed.
Solution Approach 2:
The system implements periodic collection cycles based on occupation thresholds rather than continuous or fixed-schedule operation. Collection is triggered periodically when sensors detect that waste accumulation has reached a predetermined level, creating an event-driven periodic action pattern that matches collection intensity to actual waste generation patterns rather than operating on rigid time intervals.
Data Source
Figure 1
Figure 2~3
AI summary
The present invention relates to a method for the intelligent control of waste collection in an automated waste collection plant. Said method comprises storing operating information of automated waste collection plants and an energy value corresponding to the energy used in the emptying cycles; calculating an energy usage model by means of a first algorithm; and training a virtual agent by means of a second algorithm that evaluates the energy consumption of a series of actions of a plant in response to several waste occupation states in the inlets. The virtual agent receives, in real time, information relating to the loading state of the inlets of a real plant and establishes decision-making policies of said real plant based on the received information and on actions considered optimal during the training. The real plant is then operated according to the established decision-making policies.