Simulation-oriented model reuse in cyber-physical systems: A method based on constrained directed graph

Modeling and Simulation of Cyber-Physical Systems(MSCPS)is demanding in terms of immediate response to dynamic and complex changes of CPS.Simulation-oriented model reuse can be used to build a whole CPS model by reusing developed models in a new sim-ulation application,which avoid repeated modeling and thus reduce the redevelopment of submodels.Model composition,one of the important methods,enables model reuse by selecting and adopting diversified integration solutions of simulation components to meet the requirements of simulation application systems.In this paper,a real-time model integration approach for global CPS modeling is proposed,which reuses devel-oped submodels by compositing submodel nodes.Specifically,a constrained directed graph of submodels for the whole system which can meet the simulation requirements is constructed by reverse matching.Submodel properties,including co-simulation distance between submodel nodes,reuse benefit and simulation performance of model nodes,are quantified.Based on the properties,the model-integrated solution for the whole CPS simulation is retrieved throughout the model constrained digraph by the Genetic Algo-rithm(GA).In the experiment,the proposed method is applied to a typical model integrated computing scenario containing multiple model-integration solutions,among which the Pareto optimal solutions are retrieved.Results show that the effectiveness of the model integration method proposed in this paper is verified.

Object-oriented Modular Model Library for Distillation

For modeling and simulation of distillation process, there are lots of special purpose simulators along with their model libraries, such as Aspen Plus and HYSYS. However, the models in these tools lack of flexibility and are not open to the end-user. Models developed in one tool can not be conveniently used in others because of the barriers among these simulators. In order to solve those problems, a flexible and extensible distillation system model library is constructed in this study, based on the Modelica and Modelica-supported platform MWorks, by the object-oriented technology and level progressive modeling strategy. It supports the reuse of knowledge on different granularities： physical phenomenon, unit model and system model. It is also an interface-friendly, accurate, fast PC-based and easily reusable simulation tool, which enables end-user to customize and extend the framework to add new functionality or adapt the simulation behavior as required. It also allows new models to be composed programmatically or graphically to form more complex models by invoking the existing components. A conventional air distillation column model is built and calculated using the library, and the results agree well with that simulated in Anen Plus.

Efficient Model Store and Reuse in an OLML Database System

Deep learning has shown significant improvements on various machine learning tasks by introducing a wide spectrum of neural network models.Yet,for these neural network models,it is necessary to label a tremendous amount of training data,which is prohibitively expensive in reality.In this paper,we propose OnLine Machine Learning(OLML)database which stores trained models and reuses these models in a new training task to achieve a better training effect with a small amount of training data.An efficient model reuse algorithm AdaReuse is developed in the OLML database.Specifically,AdaReuse firstly estimates the reuse potential of trained models from domain relatedness and model quality,through which a group of trained models with high reuse potential for the training task could be selected efficiently.Then,multi selected models will be trained iteratively to encourage diverse models,with which a better training effect could be achieved by ensemble.We evaluate AdaReuse on two types of natural language processing(NLP)tasks,and the results show AdaReuse could improve the training effect significantly compared with models training from scratch when the training data is limited.Based on AdaReuse,we implement an OLML database prototype system which could accept a training task as an SQL-like query and automatically generate a training plan by selecting and reusing trained models.Usability studies are conducted to illustrate the OLML database could properly store the trained models,and reuse the trained models efficiently in new training tasks.

Federated selective aggregation for on-device knowledge amalgamation

In the current work,we explored a new knowledge amalgama-tion problem,termed Federated Selective Aggregation for on-device knowledge amalgamation(FedSA).FedSA aims to train an on-device student model for a new task with the help of several decentralized teachers whose pre-training tasks and data are different and agnos-tic.The motivation to investigate such a problem setup stems from a recent dilemma of model sharing.Due to privacy,security or in-tellectual property issues,the pre-trained models are,however,not able to be shared,and the resources of devices are usually limited.The proposed FedSA offers a solution to this dilemma and makes it one step further,again,the method can be employed on low-power and resource-limited devices.To this end,a dedicated strategy was proposed to handle the knowledge amalgamation.Specifically,the student-training process in the current work was driven by a novel saliency-based approach which adaptively selects teachers as the par-ticipants and integrated their representative capabilities into the stu-dent.To evaluate the effectiveness of FedSA,experiments on both single-task and multi-task settings were conducted.The experimental results demonstrate that FedSA could effectively amalgamate knowl-edge from decentralized models and achieve competitive performance to centralized baselines.