Hybrid Differential-Algebraic Process Model Mode Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Hybrid differential algebraic process models (DAPs) face challenges in real-time calculability due to their structural variability and complexity, leading to significant computational and memory requirements, especially when dealing with multiple possible modes, which complicates the optimization and execution of these models on process computers.
Innovation Solution
A method is introduced where the probable future modes of the hybrid DAP are predicted during runtime by evaluating state variables, allowing for the extraction of mode-specific DAPs, which are then optimized and executed using a group of executable mode-specific process models, with the prediction process running parallel to the calculation process on different processors to reduce memory load and enhance real-time performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If all possible mode-specific process models are extracted and stored for a hybrid DAP, then the calculation speed for each mode is improved, but the memory requirements and device complexity increase significantly
Solution Approach 1:
The patent extracts and stores mode-specific process models in advance before runtime, organizing them in a lookup table structure. This preliminary preparation allows the runtime system to quickly retrieve the appropriate mode-specific model without performing complex evaluations during execution, thus improving calculation speed while managing memory through structured organization
Solution Approach 2:
The patent segments the complete hybrid DAP into multiple mode-specific process models, each representing a distinct operational mode. By dividing the overall system into separate, manageable segments (mode-specific models), the system can store only the necessary subsets in memory rather than the entire hybrid model, reducing memory requirements while maintaining fast calculation capability
2Adaptability or versatility
If the complete hybrid DAP is used for calculation, then all possible modes are covered, but the calculation effort and time increase significantly
Solution Approach 1:
The patent performs preliminary extraction of mode-specific process models from the hybrid DAP and stores them in a lookup table before runtime. This advance preparation eliminates the need for complex runtime evaluations to determine appropriate modes, allowing instant retrieval of mode-specific models and significantly reducing calculation time while maintaining complete mode coverage
Solution Approach 2:
The patent creates simplified copies of the hybrid DAP in the form of mode-specific process models. Each mode-specific model is a streamlined copy tailored to a particular operational mode, removing unnecessary complexity and conditions. These copies enable fast calculation by eliminating the need to evaluate the complete hybrid model structure during runtime
3Productivity
If mode-specific process models are extracted and stored, then real-time calculability is improved, but the initial processing and memory management complexity increase
Solution Approach 1:
The patent performs the complex extraction and organization of mode-specific process models in advance during a compilation or setup phase. This preliminary action transfers the management complexity from runtime to an offline preparation stage, allowing the runtime system to simply retrieve pre-organized models from a lookup table, thereby improving real-time calculability without burdening the runtime system with complex model management
Solution Approach 2:
The patent introduces a lookup table as an intermediary data structure between the hybrid DAP and the runtime calculation system. This intermediary organizes mode-specific models in a standardized, easily accessible format, simplifying the interface between model extraction and runtime execution. The lookup table acts as a buffer that manages the complexity of storing and retrieving multiple mode-specific models efficiently
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention describes and illustrates a method for calculating the state variables (x) of a hybrid differential-algebraic process model (hybrid DAP, hDAP) in successive time steps on a process computer (1) having a process interface (2), wherein the process computer (1) is set up in such a manner that at least one process variable (xp) of a physical process (3) can be recorded by the process computer (1) and/or an output variable for influencing the physical process (3) can be output by the process computer (1) via the process interface (2), and a current mode (mt+1) of the hybrid DAP (hDAP) is determined in a calculation process by evaluating the state variables (x), and, if the current mode (mt+1) of the hybrid DAP (hDAP) differs from the previously applicable mode (mt) of the hybrid DAP (hDAP), an executable mode-specific process model (mDAPt+1, exe) corresponding to the current mode (mt+1) is selected from a group (4) of executable mode-specific process models (mDAPexe) and is used as a basis for further calculation. It is possible to quickly calculate the different modes of the hybrid DAP, with a reduced memory loading of the process computer, by determining at least one likely future mode (mi, mj, mk) of the hybrid DAP (hDAP) in a prediction process at the calculation run time by evaluating the state variables (x), extracting the mode-specific DAP (mDAPi, mDAPj, mDAPk) corresponding to the determined likely future mode (mi, mj, mk) from the hybrid DAP (hDAP), generating an executable mode-specific process model (mDAPi,exe, mDAPj,exe, mDAPk,exe) from the extracted mode-specific DAP (mDAPi, mDAPj, mDAPk) and supplying said process model to the group (4).