Batch and semi-batch processes have played an important role in the chemical industry, because of their capabilities to produce high-value products such as fine chemicals. Monitoring batch processes is very important to ensure their safe operation and to assure the production of consistent high-quality products. Multivariate statistical analysis method using multiway principal component analysis(MPCA) provides efficient monitoring scheme by removing constraints in the batch process such as the presence of significant nonlinearity, the absence of steady state. But this method has too many parameters, delays computing speed, makes it inaccurate to predict future observations. In this study we suggested more rapid, accurate monitoring and diagnosis methods by constructing local models based on time intervals. In special we could select faulty sensors by checking the real-time contribution of variables on the abnormal events. The suggested monitoring and diagnosis schemes were tested to PVC polymerization reactor, capable of monitoring faults and diagnosing the assignable causes in the good performance.