Process Monitoring And Fault Diagnosis For Multi-stage?Uneven-length Batch Processes

The batch process that produces product with small quantity and high added value, has been applied extensively in the field which is closely related to our daily life, such as chemicals, injection, medicines, fuel, metallurgical and so on. However, batch process is often accompanied by the boiler rupture, explosion, toxic gas leaking or other dangers, so we need to ensure it's safe and reliable operation in order to ensure the safety of the workers and the quality of the product. So it has become an important research topic to monitor the batch process and conduct fault diagnosis. Nowadays the process monitoring method based on multivariate statistics has gain wide application in the batch process monitoring.Theoretically, one of the outstanding features for batch process is that the process is repeated according to the pre-designed procedure. However, due to quality difference of the raw materials, the influence of the weather and so on, the process can't be repeated completely and therefore, the length of the collected data is no longer the same and this problem is known as the uneven-length data in batch process. Due to this problem, it is difficult for the monitoring and modeling of the batch process based on multivariate statistics.Dynamic time warping (DTW) is used in the field of signal match at first. Optimum matching signal is taken by computing the matching distance between the measured signal and standard signal, so we can get the the same length of the measured signal with the standard one. This text has improved the algorithm, we also find out the issue of DTW used in batch process with uneven-length data and it's corresponding solution. The uneven-length data is finally processed by evening the length of measured signals to the standard data.In addtion, this text use an injection process as the background. At first we ues DTW to process it's uneven-length data. On this foundation we can get the transition infromation based on soft-partion, and establish a monitoring model at every sub-period for online monitoring and fault diagnosis. Finally we use its uneven-length data as simulation data, and the result showed that the algorithm proposed in this text was correct and effective.