Sort:
Open Access Issue
Multi-Agent Modeling and Simulation in the AI Age
Tsinghua Science and Technology 2021, 26(5): 608-624
Published: 20 April 2021
Abstract PDF (438.8 KB) Collect
Downloads:233

With the rapid development of artificial intelligence (AI) technology and its successful application in various fields, modeling and simulation technology, especially multi-agent modeling and simulation (MAMS), of complex systems has rapidly advanced. In this study, we first describe the concept, technical advantages, research steps, and research status of MAMS. Then we review the development status of the hybrid modeling and simulation combining multi-agent and system dynamics, the modeling and simulation of multi-agent reinforcement learning, and the modeling and simulation of large-scale multi-agent. Lastly, we introduce existing MAMS platforms and their comparative studies. This work summarizes the current research situation of MAMS, thus helping scholars understand the systematic technology development of MAMS in the AI era. It also paves the way for further research on MAMS technology.

Open Access Issue
A Multilevel Splitting Algorithm for Quick Sampling
Tsinghua Science and Technology 2021, 26(4): 417-425
Published: 04 January 2021
Abstract PDF (1.2 MB) Collect
Downloads:36

To reduce intermediate levels of splitting process and enhance sampling accuracy, a multilevel splitting algorithm for quick sampling is proposed in this paper. Firstly, the selected area of the elite set is expanded to maintain the diversity of the samples. Secondly, the combined use of an adaptive difference evolution algorithm and a local searching algorithm is proposed for the splitting procedure. Finally, a suite of benchmark functions are used for performance testing. The results indicate that the convergence rate and stability of this algorithm are superior to those of the classical importance splitting algorithm and an adaptive multilevel splitting algorithm.

Open Access Issue
A New Filter Collaborative State Transition Algorithm for Two-Objective Dynamic Reactive Power Optimization
Tsinghua Science and Technology 2019, 24(1): 30-43
Published: 08 November 2018
Abstract PDF (2.4 MB) Collect
Downloads:45

Dynamic Reactive Power Optimization (DRPO) is a large-scale, multi-period, and strongly coupled nonlinear mixed-integer programming problem that is difficult to solve directly. First, to handle discrete variables and switching operation constraints, DRPO is formulated as a nonlinear constrained two-objective optimization problem in this paper. The first objective is to minimize the real power loss and the Total Voltage Deviations (TVDs), and the second objective is to minimize incremental system loss. Then a Filter Collaborative State Transition Algorithm (FCSTA) is presented for solving DRPO problems. Two populations corresponding to two different objectives are employed. Moreover, the filter technique is utilized to deal with constraints. Finally, the effectiveness of the proposed method is demonstrated through the results obtained for a 24-hour test on Ward & Hale 6 bus, IEEE 14 bus, and IEEE 30 bus test power systems. To substantiate the effectiveness of the proposed algorithms, the obtained results are compared with different approaches in the literature.

Open Access Issue
A Projection Pursuit Dynamic Cluster Model Based on a Memetic Algorithm
Tsinghua Science and Technology 2015, 20(6): 661-671
Published: 17 December 2015
Abstract PDF (3.3 MB) Collect
Downloads:13

A Projection Pursuit Dynamic Cluster (PPDC) model optimized by Memetic Algorithm (MA) was proposed to solve the practical problems of nonlinearity and high dimensions of sample data, which appear in the context of evaluation or prediction in complex systems. Projection pursuit theory was used to determine the optimal projection direction; then dynamic clusters and minimal total distance within clusters (min TDc) were used to build a PPDC model. 17 agronomic traits of 19 tomato varieties were evaluated by a PPDC model. The projection direction was optimized by Simulated Annealing (SA) algorithm, Particle Swarm Optimization (PSO), and MA. A PPDC model, based on an MA, avoids the problem of parameter calibration in Projection Pursuit Cluster (PPC) models. Its final results can be output directly, making the cluster results objective and definite. The calculation results show that a PPDC model based on an MA can solve the practical difficulties of nonlinearity and high dimensionality of sample data.

Open Access Issue
Unsupervised Dynamic Fuzzy Cognitive Map
Tsinghua Science and Technology 2015, 20(3): 285-292
Published: 19 June 2015
Abstract PDF (573.4 KB) Collect
Downloads:26

Fuzzy Cognitive Map (FCM) is an inference network, which uses cyclic digraphs for knowledge representation and reasoning. Along with the extensive applications of FCMs, there are some limitations that emerge due to the deficiencies associated with FCM itself. In order to eliminate these deficiencies, we propose an unsupervised dynamic fuzzy cognitive map using behaviors and nonlinear relationships. In this model, we introduce dynamic weights and trend-effects to make the model more reasonable. Data credibility is also considered while establishing a machine learning model. Subsequently, we develop an optimized Estimation of Distribution Algorithm (EDA) for weight learning. Experimental results show the practicability of the dynamic FCM model. In comparison to the other existing algorithms, the proposed algorithm has better performance in terms of convergence and stability.

Open Access Issue
Knowledge Map Mining of Financial Data
Tsinghua Science and Technology 2013, 18(1): 68-76
Published: 07 February 2013
Abstract PDF (1.2 MB) Collect
Downloads:19

The Knowledge Map (KM) concept, which was derived from the Fuzzy Cognitive Map (FCM), is used to describe and manage knowledge. KM provides insight into the interdependencies and uncertainties contained in the system. This paper uses a model-free method to mine KMs in historical data to analyze component stock corporations of the Shanghai Stock 50 index. The analyses use static and time-domain analyses. The results indicate that a knowledge map is useful for representing knowledge and for monitoring the health of companies. Furthermore, sudden changes of the key features of the KMs should be taken seriously by policymakers as an alarm of a crisis.

Total 6