A new model of high speed Grid Intrusion Detection System is presented which deploys the IDS in the grid environment and uses signature matching based detection technique.In order to realize the load balance between the grid resources,an improved genetic algorithm is used to perform task allocation.In addition,data fusion and frequent pattern mining technique are used to perform alert aggregation and correlation which can help to realize distributed attack detection.This new model not only utilizes the grid resource to detect attack but also fully exploits grid resource and realizes load balance.