Discover What's New
Get more out of MATLAB and Simulink by downloading the latest release.
R2021b Release Highlights
New Products
- RF PCB Toolbox- Perform electromagnetic analysis of printed circuit boards
- Signal Integrity Toolbox- Simulate and analyze high-speed serial and parallel links
Major Updates
- 激光雷达的工具箱- Use Lidar Viewer app to visualize, analyze, and preprocess lidar point clouds interactively
- Simulink Code Inspector- Use Code Inspector contextual tab to check compatibility, inspect code and view results directly in the model
- Simulink Control Design- Design Model Reference Adaptive Controllers
- Symbolic Math Toolbox- Get guidance for symbolic workflows with next-step suggestions in MATLAB Live Editor
- Wavelet Toolbox- Use wavelet analysis to process and extract features for signals and images for AI workflows
Release Details by Product
MATLAB Product Family
- Editor: Explore new capabilities for code refactoring and block editing, as well as improved code suggestions, code completion, and debugger
- Python Interface: Run Python commands and scripts from MATLAB
- BackgroundPool: Run MATLAB functions on a background thread
- Live Editor Tasks: Interactively summarize, transform, and filter groups of data (compute by group) and center and scale data (normalize data)
trenddecomp
Function: Find trends in data- HDF5: Support for HDF5 version 1.10, including Single-Writer/Multiple-Reader (SWMR), Virtual Dataset (VDS), and Metadata Cache Fine-Tuning
- Hardware: Use new apps for communicating with devices via serial (serial explorer) and TCP/IP (TCP/IP explorer)
ode78 and ode89
Functions: Use high-order Runge-Kutta solvers for ordinary differential equations- Projects: Organize, manage, and share your work using projects in MATLAB Online
- Lidar Viewer App: Visualize, analyze, and preprocess lidar point clouds interactively using the Lidar Viewer app
- Aerial Lidar Processing: Apply deep learning and SLAM algorithms on aerial lidar data
- Machine Learning and Deep Learning Feature Extraction: Perform signal and image analysis, preprocessing, and feature extraction with wavelet techniques and interactive apps for AI models
- Acceleration and Deployment: Speed up wavelet algorithms using multicore processors and GPUs; generate code for desktop prototyping and embedded deployment
Simulink Product Family
- Multiple Simulations Panel: Run multiple simulations for different scenarios from Simulink Editor
- Custom Tabs: Create custom Simulink toolstrip tabs
- Mini Map: Determine context when viewing part of a block diagram
- Entry and Exit Junctions:Create entry and exit connections across hierarchy boundaries
- Physical Interfaces with Simscape: Create physical interfaces, ports, and connections on components
- Software Architectures: Create software architectures from existing components
- Test Harness Support: Create test harnesses for System Composer components
- Runtime Parameter Tuning:Tune parameters while a deployed simulation is running
AI, Data Science, and Statistics
Statistics and Machine Learning Toolbox
- 浅神经网络: Use in Simulink as native blocks; automate hyperparameter tuning
- Machine Learning: Explore k-means clustering in a live task and use isolation forest for anomaly detection
- Deployment: Export models from the Classification and Regression Learner app to MATLAB Production Server
- 1D Convolutional Networks: Create and train networks for sequence and time-series data
- Deep Network Designer: Export trained networks to Simulink
- Neural ODEs: Compute deep learning solution of nonstiff ordinary differential equations using
dlode45
Control Systems
- Model Reference Adaptive Control Block: Design and simulate Model Reference Adaptive Controllers
Reinforcement Learning Toolbox
- Rewards Generation: Automatically generate reward functions from controller specifications
Model Predictive Control Toolbox
- Nonlinear MPC: Implement multistage nonlinear MPC controllers with the Embotech FORCESPRO solver
- Nonlinear ARX Models: Create models that use regression functions based on machine learning algorithms
Predictive Maintenance Toolbox
- Diagnostic Feature Designer: Rank unlabeled features, and generate spectral features for characteristic fault frequency bands in rotating machinery
RF and Mixed-Signal
- New Product: Perform electromagnetic analysis of printed circuit boards
- New Product: Simulate and analyze high-speed serial and parallel links
Math and Optimization
- Next-Step Suggestions: Get guidance for symbolic workflows with next-step suggestions in MATLAB Live Editor
- Problem-Based Workflow: Solve nonsmooth or global optimization problems using the problem-based optimization workflow