A Research On Auto-routing Methods And Applications Of Ship Piping

Modern ship system consists of a large number of pipes with various functions,which are distributed in the hull structure and extends to every corner of a ship.The transportation of liquid or gas between different areas and equipment,and the layout of power and communication cables all rely on ship pipes.Therefore,piping design is very important in ship design.The quality of ship piping design can directly affect the operation of various ship machinery and electrical equipment;also,it can provide the basis for optimal layout of the whole ship.ISO divides the ship piping design into five successive phases:preliminary design,functional design,detailed design,production engineering,and system support information.Ship pipe route design(SPRD),as the initial activity of pipe detailed design,is particularly important because many other activities of detailed design depend on it.The number of ship pipes is huge;the constraints of pipe routing are complicated;the design results are prone to be affected by the change of hull structure and equipment location;moreover,every ship has a unique specification,which needs rework on piping for each vessel.All of the factors above make SPRD extremely difficult and time-consuming.At present,there is no practical commercial automatic pipe routing system in use.With the improvement of the shipbuilding industry,the traditional pipe routing method based on manual work becomes one of the bottlenecks in modern ship production.To shorten the design time and improve the quality of piping,it is necessary to study the new methods and software architectures for ship pipe routing.Since the 1970s,pipe routing has become a topic that attracts lots of attention.The design environment extends from two-dimensional space to three-dimensional space;the routing objectives and constraints become more and more complex,during which many effective methods have been put forward.However,due to the particularity and high complexity of the problem,so far there are still many problems to be studied further.Firstly,the objectives and constraints of SPRD are classified from the perspective of automatic pipe routing algorithm,and the methods for handling constraints are designed;then,according to the analysis results of constituent elements,mathematical representation,and typical routing cases in ship pipe optimization,a new software architecture and several application workflows for automatic ship piping system are put forward,which can provide direction to the study on pipe routing methods.To solve the four typical routing cases,several kinds of novel ship pipe routing methods are proposed and implemented.Meanwhile,actual constraints and algorithm performance are considered.The main contributions of this dissertation to ship pipe routing methods include:(1)To solve SPRD efficiently under complex constraints,the routing methods and optimization strategies in grid space based on shortest path faster algorithm(SPFA)and improved A*algorithm are proposed.SPFA is introduced into three-dimensional space to solve SPRD,in which the grid exploring strategy,the path backtracking mechanism and the sub-optimal path generating method are designed.In the improved A*algorithm,the data structure and updating strategy of open set,the algorithm parallelization,the evaluation functions and the optimization of algorithm workflow are studied.The experimental results show that the proposed deterministic pipe routing algorithms have good performance,and can generate high-quality routing paths that satisfy the objectives and constraints.(2)When routing multiple pipes or branch pipe,the routing result depends on the routing sequence of every single pipe or branch.To get an optimal result in this kind of cases,a combined optimization method based on Lee algorithm and tabu search is proposed.Lee algorithm is selected as path generating algorithm,and tabu search is selected as routing sequence optimization algorithm.Path generating algorithm builds routing result according to the routing sequence,and the sequence optimization algorithm adjusts routing sequence according to the routing result.The two algorithms execute the searching process based on mutual feedbacks,and gradually get an entire optimal solution.The experimental results indicate that the combined algorithm owns both the high efficiency of deterministic routing algorithm and the searching ability of non-deterministic artificial intelligent algorithm.After dividing the large-scale piping problem into different pipe groups according to the size of pipe diameters,the algorithm is suitable for routing multiple pipes or branch pipe in each group.(3)To get an overall optimization result in some collaborative routing cases,the pipes(branches)of multiple pipes(branch pipe)need to be routed cooperatively.Aiming at this issue,a unified algorithm framework based on cooperative co-evolution algorithm is proposed,genetic algorithm and particle swarm optimization algorithm are adopted respectively to execute the cooperative evolutionary process on the pipes represented by different populations.The co-evolution algorithm and intelligence optimization algorithm are combined to avoid the combination explosion caused by routing multiple pipes under complex constraints.A fixed-length path encoding scheme is put forward,which generates the sub-segment grid paths between adjacent connection points by A*algorithm or routing style method.Thus,the transformation and maintenance cost of the path individuals during evolution process can be reduced.To improve the low-performance issue caused by a large amount of calculation in cooperative co-evolution algorithm,a parallel pipe routing algorithm based on allocation strategy of computing task is presented,and a design of cooperative co-evolution genetic algorithm with A*algorithm to generate path individuals is given,which is implemented by OpenMP technology.Experimental results demonstrate the feasibility and advancement of the proposed algorithms,which can generate solutions with cooperative routing effects and have good performance on efficiency and convergence.In addition,due to the searching mechanism of non-deterministic algorithms,more candidate solutions can be obtained by setting different parameters or running at different times.