|
Title
|
A Soft-Sensor Approach to Probability Density Function Estimation
|
|
Type
|
Book
|
|
Keywords
|
Fuzzy logic, Online estimation, Kernel density function, Gaussian mixture model, Soft sensor
|
|
Abstract
|
In this paper, based on soft-sensor idea, a fuzzy method is proposed to approximate the PDF of a system output online. To achieve this goal, Gaussian mixture model is generated by the fuzzy algorithm. The defuzzifier operator has been modified to make it suitable for this application. Means and variances of the model are adapted using observed data in each new sample. Then, rules weights are tuned by minimizing the expected L2 risk function of estimated and true PDFs. In contrast to the existing approaches, our approach does not require fine-tuning parameters for a specific application, we do not assume specific forms of the target distributions and temporal constraints are not assumed on the observed data. The algorithm is simple and easy to use. The CPU time of each iteration of the algorithm is lower than 0.005 second, which is suitable for most online real-world applications. Simulation results show capability of the proposed algorithm in online and accurate estimation of kernel density function.
|
|
Researchers
|
Javad Poshtan (Third Researcher), Yousef Alipouri (Second Researcher), Majid Ghaniee Zarch (First Researcher)
|