README.md

In chengfeifan/monitor: Write Process monitoring method

monitor

Write the algorithm for process monitoring. The method is PCA, KPCA, DPCA and DKPCA.

PCA

Run PCA on a linear autocorrelated model

library(monitor)
data(normal_data)
data(fault_data)
normal_embed <- scale_pca(normal = normal_data, fault = normal_data)
fault_embed <- scale_pca(normal=normal_data,fault=fault_data)
model_pca <- PCA_model(data=normal_embed, kernel_num=2)
threshold_pca <- threshold(pca_object=model_pca)
fault_detection <- faultDetect(threshold_object = threshold_pca, observation =  fault_embed)

The figure shows the result as follows:

DPCA

Run DPCA on a linear autocorrelated model

# DPCA
library(monitor)
data(normal_data)
data(fault_data)
normal_embed <- embed(scale_pca(normal = normal_data, fault = normal_data),2)
fault_embed <- embed(scale_pca(normal=normal_data,fault=fault_data),2)
model_pca <- PCA_model(data=normal_embed, kernel_num=5)
threshold_pca <- threshold(pca_object=model_pca)
fault_detection <- faultDetect(threshold_object = threshold_pca, observation =  fault_embed)
plot(fault_detection)

The figure shows the result as follows:

KPCA

Run KPCA on a nonlinear model

data("nonlinear")
nonlinear_train <- nonlinear[1:400,]
nonlinear_test <- nonlinear[401:nrow(nonlinear),]
normal_embed <- scale_pca(normal = nonlinear_train, fault = nonlinear_train)
fault_embed <- scale_pca(normal=nonlinear_train,fault=nonlinear_test)

kernel = rbfdot(sigma=1/7200)
model_kpca <- KPCA_model(data = normal_embed, kernel = kernel, kernel_num = 2)
threshold_kpca <- threshold(model_kpca)
fault_detection <- faultDetect(threshold_object = threshold_kpca, observation =  fault_embed)
plot(fault_detection)

The figure shows the result as follows:

DKPCA

Run DKPCA on a nonlinear model

data("nonlinear")
nonlinear_train <- nonlinear[1:400,]
nonlinear_test <- nonlinear[401:nrow(nonlinear),]
normal_embed <- embed(scale_pca(normal = nonlinear_train, fault = nonlinear_train),2)
fault_embed <- embed(scale_pca(normal=nonlinear_train,fault=nonlinear_test),2)

kernel = rbfdot(sigma=1/7200)
model_kpca <- KPCA_model(data = normal_embed, kernel = kernel, kernel_num = 5)
threshold_kpca <- threshold(model_kpca)
fault_detection <- faultDetect(threshold_object = threshold_kpca, observation =  fault_embed)
plot(fault_detection)

The figure shows the result as follows:

chengfeifan/monitor documentation built on May 14, 2019, 2:29 p.m.