IGCC Plant Variable Estimation Using Extended Kalman Filter
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current IGCC power generation plants face challenges in accurately estimating plant variables due to limited online sensors and unaccounted modeling and sensing uncertainties, leading to potential constraint violations and inaccurate assessments of operating conditions.
Innovation Solution
A system and method that combines sensor measurements with an extended Kalman filter (EKF) for estimating plant variables, incorporating a preemptive-constraining processor to prevent constraint violations and a measurement-correction processor to update the dynamic model, allowing for accurate estimation of variables not directly sensed by the sensor suite.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If model-based estimation is used to estimate plant variables, then estimation accuracy is improved, but modeling and sensing uncertainties are not appropriately accounted for
Solution Approach 1:
The patent implements a feedback mechanism where the estimator uses measured plant variables to continuously update and correct state estimates. The measurement-correction processor compares estimated variables with actual sensor measurements and adjusts the state estimates accordingly, creating a closed-loop system that accounts for sensing uncertainties through continuous feedback correction.
Solution Approach 2:
The patent applies preemptive constraining to the state estimates before they are used for control decisions. By预先 constraining the estimates to satisfy known plant constraints (such as physical bounds and operational limits), the system proactively prevents unreliable estimates from causing problems, rather than reacting after uncertainties cause issues.
2Measurement precision
If extensive processing is applied to account for uncertainties and constraints, then estimation accuracy is improved, but computational burden increases substantially
Solution Approach 1:
The patent extracts and handles constraints separately from the main estimation algorithm. Instead of incorporating complex constraint satisfaction into the core estimator, it applies preemptive constraining as a separate, simpler post-processing step that removes constraint violations from the estimates without requiring complex computational procedures.
Solution Approach 2:
The patent changes the parameters being estimated by focusing on a reduced set of critical state variables rather than attempting to estimate all plant variables. The estimator concentrates computational resources on estimating key states that most impact plant performance, while other variables are derived or measured directly, reducing overall computational burden.
3Device complexity
If limited online sensors are used in the gasification section, then device complexity is reduced, but available information for monitoring and control is insufficient
Solution Approach 1:
The patent introduces an estimator as an intermediary that bridges the gap between limited sensor measurements and the full plant state. The estimator acts as a mediator that infers unmeasured variables from available measurements and model relationships, providing comprehensive monitoring information without requiring comprehensive sensor coverage.
Solution Approach 2:
The patent creates virtual copies of physical sensors through software-based estimation. Instead of installing physical sensors throughout the gasification section, the system creates virtual sensor readings through the estimator that replicate the information that would be obtained from additional physical sensors, at a fraction of the cost and complexity.
Data Source
AI summary
System and method to estimate variables in an integrated gasification combined cycle (IGCC) plant are provided. The system includes a sensor suite to measure respective plant input and output variables. An extended Kalman filter (EKF) receives sensed plant input variables and includes a dynamic model to generate a plurality of plant state estimates and a covariance matrix for the state estimates. A preemptive-constraining processor is configured to preemptively constrain the state estimates and covariance matrix to be free of constraint violations. A measurement-correction processor may be configured to correct constrained state estimates and a constrained covariance matrix based on processing of sensed plant output variables. The measurement-correction processor is coupled to update the dynamic model with corrected state estimates and a corrected covariance matrix. The updated dynamic model may be configured to estimate values for at least one plant variable not originally sensed by the sensor suite.