In　the　case　of　multicollinearity,　biased　estimators　are　always　introduced　to　correct　the　least　squares　estimator.　In　this　paper,　we　propose　a　new　biased　estimator　for　the　restricted　linear　model.　The　properties　of　the　new　estimator　and　its　superiority　over　the　restricted　least　squares　estimator　in　terms　of　the　mean　square　error　and　Pitman　closeness　criterion　are　theoretically　analysed.　Furthermore,　we　optimize　and　verify　the　feasibility　of　the　new　estimator　using　a　numerical　simulation.