Research On Algorithm Of Minimum Expected Weights For Uncertain Planning

Intelligent planning is an important research direction of artificial intelligence.It can model the realistic problem,to solve problems in the real life through the planning model.In recent years,uncertain planning,as its branch,gradually get focus in the field of scholars and engineering experts.In the analysis of real-life statistics,the mathematical expectations are used to accurately estimate the future development trend of something.After the uncertain model of real life is planned,the factors that determine the effect of the expected value and the control of the variables in the planning model are studied,so that the problem can be approximated.Therefore,this paper introduces the minimum expected weight to solve the problem of the minimum expected weight from the specific state to the target state.After solving the planning problem in the field of uncertain planning,we can get three solutions: strong planning solution,strong cyclic programming solution and weak programming solution.Because of the strong planning solution and the strong circular planning solution is more practical significance,this paper mainly studies the planning problems in these two different situations.Point at the problem of strong planning in the field of uncertain planning,the traditional method of solving the plan do not involve the uncertainty of the planning action and the cost of its consumption,which makes the reliability of the planning solution not good.In view of this problem,this paper proposes the concept of expectation value of the sum of action weights,and proposes the minimum expected weight solving algorithm for strong planning solution based on this concept.The algorithm starts from the target state set,the reverse search can reach the state of the target state set and the desired desired weight is minimized,then the state is added to the searched state set and the remaining state is updated to the expected weight of the searched state set.The The above method is iterated until the searched state set no longer changes.By analyzing the time complexity of the algorithm and the experimental simulation,the minimum expected weight algorithm proposed in this paper can quickly obtain the minimum expected weight strong programming solution.For solving the problem of strong cyclic planning,it is necessary to consider not only the uncertainty of planning action and the cost of consumption,but also how to avoid invalid search when the algorithm is running.Based on the concept of expected weights,this paper proposes a minimum expected weight solving algorithm under strong cyclic programming.The algorithm uses the state stratification method to preprocess the actions and states that can not form the strong cyclic programming solution,therefore,to avoid the search of invalid actions.Solve the strong cyclic programming solution through the depth-first traversal,and transform the solutions into relativistic equations,and then through the LU decomposition method to solve the linear equations to find the minimum expected weight strong cycle planning solution.Finally,we use an example of the algorithm to verify the correctness of the proposed method.The experimental results show that this algorithm can quickly solve the strong periodic programming solution of the minimum expected weight in the uncertain planning field.