Design And Implementation Of Measurement Type Recommendation Algorithm And System On NX Platform

With the development of industry 4.0,information technology is gradually applied to the industry,making the industry gradually develop to the direction of intelligent manufacturing,which also requires the development of industrial software to the direction of intelligent and cloud service.The NX platform is also moving in that direction.However,because the NX platform is not open source and lacks data sets,intelligence is still in the exploratory stage.NX Measurement is the Measurement system of NX platform,which is one of the core systems of NX platform.If we can improve its intelligence degree,we can improve the convenience and user experience of NX platform.Based on the above background,this thesis will provide a recommendation system to predict the type of measurement expected by users and provide a one-click configuration of measurement parameters on this basis,thus enhancing the intelligence of NX Measurement.The main tasks are as follows.（1）The data structure of Single Measure and Measure Scalar was analyzed by Auto Test and NX Open provided by NX,and a data extraction system for this data type was designed and implemented.（2）Based on the background that NX is offline and the data set is small,the data input type of the recommendation algorithm is determined through item CF algorithm,and the item CF,user CF,Slope One and KMeans algorithms are tested and analyzed,so as to design and implement an improved hybrid recommendation algorithm suitable for the measurement type of NX platform.The algorithm combines the weighted combination of item CF and user CF,and introduces the time series of input data and user preference factors.The algorithm makes up for the defect that item CF ignores the association between users,and the coverage is more than 5% higher than the single collaborative filtering algorithm,and it has a higher F1 value.（3）A non-intrusive cold-start strategy for NX platform is designed.The policy is implemented by user-defined configuration and popular data,combining personalized cold start policy and non-personalized cold start policy.（4）A recommendation system for measurement types has been designed and implemented based on the NX framework without changing the NX Measurement process.The recommendation system provides users with a reliable list of recommended measurement types,with one-click configuration of measurement parameters,hotkeys,and other functions.In addition,the system allows the user to configure data sources and personal data,ensuring a high degree of personalization.In summary,this thesis implements the extraction function of NX platform measurement data,and based on this design,a hybrid recommendation algorithm and recommendation system applicable to NX platform measurement types are implemented.The recommendation system was successfully applied to the measurement system of industrial software,and the intelligence of NX Measurement was improved.