Wastewater treatment plants (WWTPs) remain notorious for poor data quality and sensor reliability problems due to the hostile environment, missing data problems and more. Many sensors in WWTP are prone to malfunctions in harsh environments. If a WWTP contains any redundancy between sensors, monitoring methods with sensor reconstruction such as the proposed one can yield a better monitoring efficiency than without a reconstruction scheme. An enhanced robust process monitoring method combined with a sensor reconstruction scheme to tackle the sensor failure problems is proposed for biological wastewater treatment systems. The proposed method is applied to a single reactor for high activity ammonia removal over nitrite (SHARON) process. It shows robust monitoring performance in the presence of sensor faults and produces few false alarms. Moreover, it enables us to keep the monitoring system running in the case of sensor failures. This guaranteed continuity of the monitoring scheme is a necessary development in view of real-time applications in full-scale WWTPs.Process operators obtain information on the current process conditions from a range of sensor types. Hence, the accuracy of sensors is crucial to successful process control and monitoring and the ability to detect sensor faults is very useful, especially when processes are monitored and controlled based on process information from many sensors. Sensors may exhibit partial failures such as bias, drift or precision degradation as displayed in Figure 1. It causes the accuracy and reliability of the measurement to decrease, which may result in an erroneous control action and false perception of the performance of the monitored system. Faulty sensors that are either completely or partially failing (hard fault or soft fault) provide incorrect information for monitoring and control. This can be detrimental to various data-driven decision schemes. Moreover, data may not be available due to sensor malfunction or communication problems within the data collection system. These data problems make it difficult to extract and interpret information from data. Monitoring or control using the measurements is then problematic.Conventional engineering methods to find and to correct for sensor faults make use of procedures that check and recalibrate the sensors periodically. Often, this does not satisfy the requirements of the hostile environment in environmental processes, such as water, waste and air pollution. Therefore, prompt detection of the occurrence and correct identification of the location of sensor faults and reliable reconstruction (or recovery) of faulty sensors is of primary importance for efficient operation. In contrast to fault detection and isolation, sensor fault detection and validation is quite a new research area, which is required for use in wastewater treatment, but has few application results (Qin and Li, 1998; Qin, 2003;Volcke et al., 2005). This paper concentrates on formulating a process monitoring system to the problem of ''faulty sensor'' characteris...