Intelligent Evolutionary Algorithm Design Of Indoor Spatial Layout

A large number of real-life application problems can be solved as optimization problems,and evolutionary algorithms have been one of the effective methods to solve such problems.According to whether the optimization objective can be explicitly expressed by mathematical functions,optimization problems can be classified as explicit optimization problems and implicit optimization problems.For explicit objective optimization problems,evolutionary algorithms show strong advantages and do not require the objective function to satisfy continuous,differentiable and other constraints.However,for implicit optimization problems,the traditional evolutionary algorithms are out of power,and the optimization objectives of such problems generally depend on a variety of factors such as users’ subjective preferences,artistic cultivation,and life experiences,which are difficult to be expressed in a clear functional form.In the field of residential design,the traditional Chinese interior space layout problem is a typical implicit optimization problem.As real estate is the most rapidly developing industry in recent decades,the corresponding demand of residential design in the market is rising year by year,and the individualized demand of interior space layout design is becoming more and more complex.The interior space layout design and interior scene model are not only applied in the field of architectural design,but also gradually applied to game development,virtual reality and other fields.Although the market demand has greatly increased,the traditional design process of interior space layout has not been improved accordingly.There are a lot of low-efficiency work in the design process,which cannot efficiently meet the needs of a large market group and individual needs.To address the difficulties of the implicit optimization problem,the "human-computer interaction model" is introduced in the evolutionary algorithm,which replaces the hard-to-quantify function design work with a manual evaluation strategy.down,and after a large number of iterations,a satisfactory target solution is finally generated.The research work in this paper is divided into the following aspects:(1)An application model of interactive genetic algorithm in interior layout is proposed.The classical evolutionary algorithm,genetic algorithm,is invoked as the model support to automatically generate the interior space layout by giving the boundary of the building and the user’s requirements as design constraints,so that ordinary users with no professional basis can participate in the design process and get rid of the defects of insufficient user participation in traditional design methods.The main idea of the algorithm is to represent the room units in the interior space layout with multiple rectangular polygons,and transform the constraints and rules in the layout design into constraints on the relevant room units,design the constraint functions and the corresponding models of spatial factors and genetic codes according to the relevant rules,and complete the whole operation process from inputting parameters to generating the final target solution through the user interaction interface.The design method can be used not only for interior space layout design of residential buildings,but also extended to the space layout of large-scale scenarios,such as office buildings,shopping centers,supermarkets,etc.(2)An interactive algorithm framework of differential evolution algorithm suitable for interior house design is proposed.The disadvantage of the "human-computer interaction" model is that users are prone to fatigue after participating in multiple iterations of the algorithm through interaction,and the evaluation results will be severely distorted and affect the final result.into a state of contradiction.In the interior layout design,users do not have a high degree of recognition of the drawings involved in the profession,and when using the interactive scoring method to evaluate,it further aggravates user fatigue.In response to this defect,the differential evolution algorithm is introduced to replace the genetic algorithm,the user evaluation method is changed,and the scoring operation is replaced by the pairwise selection operation method,thereby alleviating user fatigue,and the traditional heuristic algorithm is easy to fall into local optimum.,the "backtracking strategy" is introduced to help the algorithm return to the upper route through the layer-by-layer return strategy,thereby jumping out of the local optimal state.Finally,the simulation experiment shows that,compared with the genetic algorithm,the differential evolution algorithm has advantages in reducing user fatigue and accelerating algorithm convergence.(3)A two-stage interior layout design method is proposed by introducing a reverse learning strategy.In order to improve the usability of the interior layout design method for people with different needs,the design method is divided into two stages,that is,to locate the position of the functional space first,and then generate the wall constraints of the surrounding area.In addition,the principle of priority positioning for the living room is adopted.After the location of the living room is determined,the rest of the functional areas can be randomly generated through the adjacent relationship with the living room or between areas.This method can break through the limitations of the number and location of functional areas,and meet the diverse needs of users.At the same time,because the new method removes a large number of hard constraints and adds random factors,in addition,a reverse learning strategy is introduced in the algorithm.This method can expand the search space based on the original data range by looking for opposite points,and optimize the data search range.Find a better solution and improve the convergence speed.Experiments show that after adding the reverse learning strategy,expanding the search range of the solution space helps to achieve the goal of generating the final solution faster.