APC Seed Model Building From Historical Data for Faster MPC Deployment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional Advanced Process Control (APC) methods in process engineering industries are hindered by high implementation costs, undesirable production interventions during testing, and a steep learning curve for engineers, particularly due to the complexity of building Multivariable Predictive Control (MPC) models.
Innovation Solution
A computer-implemented method and system utilizing artificial intelligence (AI) and machine learning (ML) techniques to build a 'seed-model' from historical plant data, automating the identification of process variables, data cleansing, and model configuration, thereby simplifying the MPC application process and reducing the need for traditional testing steps.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional APC methods are used with manual model building, then model accuracy can be achieved, but implementation costs and time consumption increase significantly
Solution Approach 1:
The system performs preliminary actions by automatically collecting and preprocessing historical plant data before model building is needed. Data cleansing, variable identification, and dataset preparation are completed in advance using automated scripts, eliminating the need for manual data collection during traditional testing phases.
Solution Approach 2:
The patent replaces manual mechanical processes with automated computational systems. Instead of engineers manually collecting data, cleaning datasets, and building models step-by-step, automated scripts and machine learning algorithms perform these tasks, significantly reducing implementation time while maintaining model accuracy.
2Reliability
If traditional plant testing is conducted for model identification, then reliable process models can be obtained, but production disruptions occur
Solution Approach 1:
The system performs model identification in advance using historical data from normal production operations, before any testing is needed. By preparing datasets and building models preliminarily using existing operational data, the system eliminates the need for disruptive plant testing while still achieving reliable process models.
Solution Approach 2:
Instead of testing on the actual production system, the patent creates a virtual copy of the production environment using historical data. This digital replica allows for model identification and validation without affecting real production operations, maintaining reliability while avoiding disruptions.
3Reliability
If comprehensive APC implementation steps are followed, then control performance can be optimized, but the complexity and difficulty for engineers increase
Solution Approach 1:
The patent merges multiple separate APC implementation steps into a single automated workflow. Data collection, cleansing, variable identification, model building, and validation are combined into one integrated process that executes automatically, reducing the perceived complexity for engineers while maintaining comprehensive control performance optimization.
Solution Approach 2:
The system performs self-service by automatically executing the entire APC implementation workflow without requiring extensive engineer intervention. Automated scripts handle data processing, model identification, and validation tasks that traditionally required manual engineering effort, simplifying the process while achieving optimized control performance.
4Productivity
If automated data processing is implemented, then implementation costs are reduced, but data quality and model accuracy may suffer
Solution Approach 1:
The patent replaces manual data processing with automated computational systems that apply consistent, reproducible algorithms. Machine learning scripts automatically cleanse data, handle missing values, and validate quality metrics, ensuring high data quality standards are maintained while dramatically improving implementation efficiency and reducing costs.
Data Source
AI summary
Systems and methods provide a new paradigm of Advanced Process Control that includes building and deploying APC seed models. Embodiments provide automated data cleansing and selection in model identification and adaption in multivariable process control (MPC) techniques. Rather than plant pre-testing onsite for building APC seed models, the embodiments help APC engineers to build APC seed models from existing plant historical data with self-learning automation and pattern recognition, AI techniques. Embodiments further provide “growing” and “calibrating” the APC seed models online with non-invasive closed loop step testing techniques. PID loops and associated SP, PV, and OPs are searched and identified. Only “informative moves” data is screened, identified, and selected among a long history of process variables for seed model development and MPC application. The seed models are efficiently developed while skipping the costly traditional pre-testing steps and minimizing the interferences to the subject production process.