Train Uncoupling Robot Visual Design And Control

In recent years,with the rapid modernization development of the railway,the comprehensive automation systems have applied in Xin Feng town,north railway station of Chengdu,north rail-way station of Zhengzhou and other large Marshalling stations,which has improved the efficiency of vehicles.But the work of train hook is done by hand all the time,which is easy to pick the wrong hook and has a serious impact on the efficiency,and even takes personal accident.Designing of the control system of the train hook robot which can automatically finish the work of train hook at a specified time that is an important topic.The paper mainly study on the steps of the train hook by applying robot,confirming the recognition of the train hook handle position and the planning of trajectory,designing the Servo control system.And for each part of the research were summarized and analyzed at home and abroad.Because of the train hook work indiscriminately the day and night or weather,so the apperance of the object or change dramatically surrounding environment,but most algorithms often use the way through extracting the target of the same characteristics(such as color,texture,outline and so on.)and only in a short ti me the suitable environment can track the object.So adopting the incremental learning PCA based on SKL in the process of tracking target could effectively study and update the target subspace of low dimensional s ubspace.The train hook is very strict to the requirement of time,but the traditional incremental learning PC A algorithm based on SKL adopted free sample particle filter algorithm can not reach the the requirement i n the real-time,so the particle swarm algorithm is combined with free sample particle filter algorithm to red uce the number of particles of the need to meet the requirements of real-time.Although the traditional robot trajectory planning often uses five times polynomial trajectory planning,which can ensure the robot arm's movement smoothly,and can accurately reach the target location and successfully caught up goals.In the process of movement Train Uncoupling Robot needs to ensure to obtain the optimal trajectory under the constraints of the robot's maximum speed,maximum acceleration and maximum fluctuation velocity.Using simulated annealing algorithm is proposed to combine with penalty function method for the optimal time of trajectory planning of robot.The traditional PID control has achieved good results in the linear control,however,there exists nonlinear problem in the work progress of train unhooking robots.The traditional PID algorithem is no longer applicable.Implicit generalized predictive control has advantages of updating model online rea l-time and achieving rolling optimization,only in view of the weak nonlinear problem.Considering realizing nonlinear estimation after increasing nonlinear terms,the paper proposes expansion implicit generalized predictive based on IPSO.