Research And Implementation Of X Company Production Dispatching System Based On Data-Dirven

At present,the manufacturing industry is undergoing a critical transition period from rapid development to high-quality development.Customers are increasingly demanding comprehensive indicators such as product individuation,price,quality,time,and service.In order to solve the above problems,a new generation of intelligent manufacturing model characterized by digitization,networking and intelligence has become a research hotspot as a product of this era.Its application will greatly promote the agility and standardization of the manufacturing process.Among them,the new model has achieved remarkable results in optimizing the job shop production scheduling problem(JSSP),and has been successfully applied to some leading manufacturing enterprises.However,due to the weak foundation of small and medium-sized enterprises,if they forcefully copy the application model of benchmarking enterprises,it will lead to a series of disastrous consequences.This paper takes X company JSSP as the research background,and faces the current situation of job shop with multiple varieties,small batch production methods and frequent orders that cause scheduling confusion.Based on intelligent algorithms such as data processing and knowledge mining,it helps small and medium-sized enterprises to enable digital and networked enterprises,then cause realize the guiding ideology for the successful application of the new model.To avoid the use of traditional mathematical modeling and algorithm complexity of high and low degree of universal application,complete the simulation method of computational costs and higher shortcomings,using data driven method,combining with computer simulation technology,machine learning technology to machine for processing the products with minimum average completion time and minimum average delay time as the goal,balance the completion time of the products in each order in the production system,and research and develop a dispatching system that assigns dispatching rules(DR)in real time on a machine basis to achieve optimization of X company Production dispatching,realizing the real-time assignment of DR to the machine,and improving the production efficiency of the job shop.First of all,on the basis of collating a large number of relevant documents and in-depth investigation of X company,in order to ensure the sample data volume of the model construction and experimental analysis process,according to the production characteristics of the X company's job shop and the approval of the relevant management personnel of X company,the simulation The processing time and preparation time of products of the same category and process flow in the processing workshop on the machine passed through are used as the original static attribute data of this type of product.According to the original static attribute data,the maximum remaining processing time,relaxation time,etc.The original dynamic attribute data of the category product,and the original static attribute data to form the original attribute data;then,the DR library is established,and the computer simulation technology is used to select the products of this category from the DR library for products of the same category and the same process by machine.The assigned optimal DR,where the DR is used as the label of the original attribute data sample,and the obtained attribute data of the original labeled data is processed to obtain the original available labeled data suitable for the machine learning classification algorithm,and then the original available labeled data Data preprocessing is used to obtain high-quality data;secondly,a random forest(RF)multi-classification model is constructed to predict the optimal DR based on input attributes,and a support vector machine(SVM)two-classification model is constructed to monitor whether the DR output by the RF multi-classification model is available.Re-simulation of the sample attribute data corresponding to the DR judged as unavailable.The optimal DR is selected from the DR library as the available DR under the sample,and the RF multi-classification model,the SVM twoclassification model and the simulation optimal DR are finally formed DR optimization model;Finally,according to the business process of the entire production dispatching process of X company's processing workshop,a production dispatching system is developed and the proposed model is embedded in it,so as to realize the application of the model proposed in this article.The research results on the research and implementation methods of production dispatching in the processing workshop of X company provide reference significance for small and medium discrete manufacturing enterprises to leapfrog transformation of intelligent manufacturing mode.The DR optimization model proposed by it is simple in development process,low in calculation cost,and solves the problem.The advantages of frequent order disturbances,such as a high degree of universality,can be widely used in solving small and medium-sized discrete manufacturing enterprises JSSP.