To effectively use the raw materials in tank in workshops of chemical engineering, embedded remote monitoring system is designed in this paper aiming at liquid ammonia. With the core ARM-Linux, the design uses modern sensor technology to monitor the pressure, liquid level, concentration of ammonia gas and temperature in tank. In traditional monitoring, non-linear structure can be damaged and monitoring efficiency is low in the process of processing data and modelling. A process monitoring method based on negative selection algorithm is put forward. The method firstly uses maximum variance unfolding method to get low dimensional manifold from original data. LTAS is then used to model low dimensional manifold and get “super global group” model. Process can be monitored in this way. Simulation result indicates the method is better than other method for monitoring process. In conclusion, the monitoring method based on ARM is effective for chemical raw materials in tank.