Ansicht
Dokumentation

APO-PPS-OPT: Multi-Agent Optimization ( RELNAPO_30A_SP4_PPS_OP3 )

APO-PPS-OPT: Multi-Agent Optimization ( RELNAPO_30A_SP4_PPS_OP3 )

TXBHW - Original Tax Base Amount in Local Currency   General Data in Customer Master  
This documentation is copyright by SAP AG.
SAP E-Book

Short text

APO-PPS-OPT: Multi-Agent Optimization

Use

As of Release 3.0A (First Customer Shipment and Support Package 4), you can use multi-agent optimization for parallel optimization on an optimization server with several processors. In multi-agent optimization, the system uses all processors in parallel for optimization purposes; you can perform optimization with another (user-defined) objective function on each processor. Each "agent" optimizes the same specified planning problem with "his" objective function. However, the agents do not perform optimization independently of one another, but inform each other about solutions already achieved in each optimization step, and exchange good partial solutions. (These synergy effects mean that you cannot replace multi-agent optimization by performing the optimization in parallel with the corresponding weighting different simulation versions).

The use of multi-agent optimization offers the following advantages:

  • You achieve a valuable solution quality within a short time. The increase in speed roughly corresponds to the number of processors.
  • Your decision margin when selecting a solution increases or you can compensate for uncertainty when defining the objective function by using different objective functions. In this way, you receive - for the same planning problem - several alternative solutions of comparable overall quality, but different valuations for the interim objectives (for example, delays or setup times).
Example: When optimizing several objectives in Production Planning and Detailed Scheduling, you must weight the individual interim objectives in the overall objective function. The weighting is generally not based on business costs, but instead reflects preferences known as steering costs). This causes uncertainty in the weighting. For example, if you weight delays to setup times in a ratio of 1:2, you can compensate a longer delay of 20 minutes by reducing the setup times to 10 minutes. If the system finds the solution alternatives listed in the following table during the optimization process, it issues only the best solution for the optimization. However, the slighly less suitable alternatives B or C may also be of interest for you.
Solution Delay Setup time Objective function value
A 4000 0 4000  
B 2001 1000 4001  
C 0 Min. 2002 4004  

Delays and setup times are specified in minutes.

You perform the settings for parallel optimization in Customizing for Basis Settings under Optimization -> Basic Functions. You configure the optimization profiles for the agents in multi-agent optimization in the optimization profile or when calling the optimization function on the tab Multi-agent optimization.

Effects on Existing Data

Effects on Data Transfer

Effects on System Administration

Effects on Customizing

Further Information






General Data in Customer Master   Addresses (Business Address Services)  
This documentation is copyright by SAP AG.

Length: 4170 Date: 20240329 Time: 121101     sap01-206 ( 79 ms )