Abstract: Network and system security is essential to modern digital communication. There will be multiple successful attempts by hackers and other intruders to obtain illegal access to networks and internet services. The development of secure systems is continuously accompanied by new dangers and the corresponding responses. Using intrusion detection systems (IDS) is one choice. An intrusion detection system. Main objective is to safeguard its resources from possible dangers. These acts are categorized as either potential or conventional attacks since it is capable of analysing and forecasting user behaviour. To find network intrusions, we typically employ Support Vector Machines (SVM) and Rough Pure Mathematics (RST). RST is used to preprocess and decrease the amount of data when packets are retrieved from the network. The RST-selected options provided to the SVM model enable it to provide many views and an explanation. A good application of the approach is to reduce the density of knowledge in a certain field. By comparing the data with Principal element analysis, the experiments show how RST and SVM schema may lower the false positive rate and increase accuracy (PCA)