Stochastic Plant Model for Uncertain Monitoring Data Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In emergency or disaster situations, plant monitoring data often includes uncertainty, making it difficult to predict plant states reliably when deterministic models are used, which can lead to inappropriate plant operation.
Innovation Solution
A plant operation assistance system that inputs monitoring data with probability distributions into a stochastic model to predict abnormal conditions and estimate uncertainty, allowing for more reliable predictions and operation considerations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a deterministic model is used for plant state prediction, then the model structure is simple and easy to implement, but the prediction reliability deteriorates when monitoring data includes uncertainty
Solution Approach 1:
The patent transforms the deterministic model parameters into probabilistic parameters by introducing probability distributions. The monitoring data is represented with probability distributions rather than fixed values, and the deterministic model parameters are converted to follow probability distributions, enabling the model to handle uncertainty while maintaining the original model structure
Solution Approach 2:
The patent replaces the deterministic mechanical prediction system with a stochastic prediction system. Instead of using fixed deterministic equations, the system uses probability theory and statistical methods to model plant states, substituting the rigid deterministic framework with a flexible probabilistic framework that can accommodate data uncertainty
2Productivity
If monitoring data with uncertainty is input into a deterministic model, then the processing is simple and fast, but the prediction reliability is reduced
Solution Approach 1:
The patent changes the parameter representation from fixed deterministic values to probability distributions. By representing monitoring data parameters as probability distributions and model parameters as probabilistic quantities, the system maintains processing efficiency while significantly improving prediction reliability under uncertain conditions
3Reliability
If a stochastic model is used to handle uncertain monitoring data, then the prediction reliability is improved, but the computational complexity increases
Solution Approach 1:
The patent segments the complex stochastic prediction problem into manageable components: (1) representing monitoring data with probability distributions, (2) transforming deterministic model parameters to probabilistic parameters, (3) using probability theory to compute prediction results. This segmentation makes the computational process more systematic and manageable
Solution Approach 2:
The patent introduces probability distributions as an intermediary layer between the monitoring data and the deterministic model. This intermediary transformation allows the deterministic model structure to be preserved while incorporating uncertainty handling, avoiding the need for completely complex stochastic models
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A plant operation assistance system (100) includes: a data obtaining unit (101) configured to obtain monitoring data indicating state quantity of a plant, the state quantity being detected by a sensor (10); an identifying unit (102) configured to identify, based on the state quantity, a probability distribution of the monitoring data; a model generation unit (103) configured to generate, based on a plant parameter composed from a database including design information of the plant, a stochastic model of the plant; a data processing unit (104) configured to assign the probability distribution to the monitoring data obtained by the data obtaining unit; and a prediction unit (105) configured to input the monitoring data assigned with the probability distribution, into the stochastic model, and predicts a state of the plant.