Back propagation network can provide dynamic maps and deal with historical data by increasing self-loop. By using the improved circle back propagation neural network, Multi-information fusion algorithm are realized and used to environment control system of textile workshop. Hardware circuit of the system is designed with the embedded microprocessor and appropriate detection devices. The training and algorithmic design of self-loop back propagation network can be achieved by embedded systems programming based on the portable operating system. Application shows that the multi-information fusion control system of textile workshop Environment, which based on local self-loop back-propagation network, has high control precision.