INTRODUCTION The linear programming algorithm has been studied for almost 60 years. In many linear programming problems, most constraints will be redundant and will not constrain optimally. The active set method is a method in which the solution is achieved by adding one constraint (single cut) or multiple constraints (multiple cut) at a time to iteratively solve a small subset of problems. The active set method was improved by Goh Saito (2012) using the bounding technique. An active set method used for optimal constraint selection rule starts with a relaxation of P. In the proposed covering method, constraints are added until they constrain all variables. In addition to the active set method, efficient constraint selection rules can increase the performance of the LP solver algorithm. Some of the constraint selection criteria are as follows: SubCOS (Prior) VIOL (Posterior) RAD (Prior) VRAD (Posterior) Constraint selection criteria can be classified as either prior or posterior methods. In the previous method, all constraints are sorted only once before the solver starts, but to use the posterior method, in each iteration a new calculation must be performed. Unlike previous methods, Posterior methods depend on x_r^* and require additional processing for each iteration. Different constraint selection rules give different priorities to each constraint and consequently have different performance. Therefore there is interest in developing an efficient constraint selection rule to measure the probability that constraining rows are in optimal conditions. Meanwhile, based on different criteria, constraints can be sorted both preliminary and postliminary based on a certain probability of being binding in case of optimal solution.. The goal of this research is to improve......half of document ..... The addition parameter .ti-cut determines the maximum number of violated cuts allowed per iteration. Controls the amount of cuts that can be added in each iteration. Generally, a large number of constraints are violated in the first iterations, so a control mechanism is needed to prevent system overload. Then use the bounding technique to add constraints from S_j to the bounded region in each direction. So to find the solution we use the primordial simplex. Violation checking will be performed only within S_j and violated cuts will be added relative to the initial multi-cut addition parameters. The added constraints are called operational constraints. After optimizing the problem on S_j , the violation check is performed on {S-{S_j }} . The process restarts from the beginning if a violation is detected by {S-{S_j }}. The Pesudo Pareto grouping code for NNLP is as follows: