Writing player drivers

本文介绍如何为Player平台创建驱动程序插件。包括静态和插件驱动的区别、构建过程、关键方法如Setup()、Shutdown()和Main()的实现细节,以及如何通过配置文件进行设备特定设置。

Citer from:

http://cgi.cse.unsw.edu.au/~cs4411/wiki/index.php?title=Writing_player_drivers

 

Every driver plugin should define the following functions:

In which you read in general configuration data from the player config file. Interface specifies which interface you are supporting, such as localise or position.

Driver::Setup()

In which you do any device-specific setup, such as opening file buffers or configuring serial ports, and then start the driver thread.

Driver::Shutdown()

In which you do any device-specific cleanup, such as closing file buffers.

Driver::Main()

An infinite loop, which repeatedly updates data and sends for whichever interface you are supporting.

Driver::SendData()

In which we actually put the data in a form that Player can read and send it off.

 

Super Detailed Guide

(I have no idea where we found this, so can't reference it properly. But it is awesome!)

plugin_driver HOWTO: Creating a Plugin Driver

The are two distinct type of drivers in Player:

- Static drivers have their code in the main Player distribution, and are statically linked into the server. Generally speaking, such drivers will be added by the lead developers.

- Plugin drivers are shared objects that are loaded at runtime (like loadable modules in the Linux kernel). They are the recommended method for all new, experimental or third party drivers.

 

Plugin drivers have serveral advantanges over their static counterparts:

- They are easier to build (no mucking about with autotools, or digging about in the server internals).

- They permit rapid development and a much faster code/compile/test cycle (the server does not need to be recompiled/relinked when the driver changes).

- Code can be maintained in a separate source repository. This is particularly useful for users with funky one-off drivers that belong with the rest of their code, data, papers, etc (and not in the Player source tree).

This document describes the process for creating new plugin drivers.

It assumes you are familiar with C++, class inheritance, and thread programming.

 

1. A plugin driver example

 

Sample code for a very basic plugin driver is provided in the examples directory; for a default install, this will be:

/usr/local/src/player/examples/plugins/exampledriver/

Copy the files from this directory and rename Makefile.example to Makefile. Try building the example:

$ make

This produces a plugin driver named exampledriver.so. You can test the model using the included configuration file:

$ player ./example.cfg

The driver block in the config file has an additional field @c plugin specifying the path to the plugin, i.e.,:

 

2. Writing a simple driver

The first step in creating a new driver is to decide which interface it will support.  The existing interfaces are described in the reference section, and their various message structures and constants are defined in player.h.  Although you can create a new interface, you should try to fit your driver to an existing interface, of which there are many.  By deciding to support an existing interface, you'll have less work to do in the server, and will likely have instant client support for your driver in several languages.
Standard API, part of which the new driver must implement (other parts To create a new driver, you should create a new class for the driver,  which should inherit from the Driver class. 
This base class defines a it may choose to override).  We now describe the salient aspects of the Driver class.

2.1 The constructor
Simple drivers will use the following Driver constructor:

This constructor will establish the buffers and queues that allow the Player server to interface with the driver. 
The preamble for the example driver, for example, looks like this:

The preamble indicates that this driver:
 - Supports the position interface (indicated by @c   PLAYER_POSITION_CODE);
 - Allows read/write access from clients (@c PLAYER_ALL_MODE);
 - Has a data buffer large enough to accomodate a  standard position interface data packet (@c sizeof(player_position_data_t));
 - Has a command buffer large enough to accomodate a  standard position interface command packet (@c sizeof(player_position_cmd_t));
 - Can queue up to 10 incoming configuration requests and 10 outgoing configuration replies.
The @c cf and @c section parameters are passed in by the server; these values to access driver-specific options stored in the Player configuration file.  Thus, for example, the driver constructor can read the value of some setting "foo":
 
This might be the serial port from which to read data, for example.
See the ConfigFile documentation for the various kinds of options that can be read from the configuration file.

2.2 Setup()
When the first client subscribes to a driver, the driver's Setup() method is called; every driver @b must implement this method. 
The Setup() method generally does two things:
- Do some device-specific initialization (e.g., open a serial port).

- Start the driver thread using Driver::StartThread().
Player is a multi-threaded application, with most drivers running in  their own thread; this makes it particularly easy to write drivers that read/write data from serial ports, network sockets, disk files and so on.
After initialization, Setup() should return either zero to indicate that the device was successfully setup, or non-zero to indicate that setup failed (the latter will cause Player to terminate).

2.3 Shutdown()

When the last client unsubscribes from a device, the driver's Shutdown() method is called; every driver @b must implement this method. 
The Shutdown() method generally does two things:
- Stop the driver thread using Driver::StopThread().
- Do some device-specific finalization (e.g., close a serial port).
 Note that the ordering is important here: we must shut down the driver thread @b before we release the resources it is using; to this end, Driver::StopThread() will tell the driver thread to quit, and wait until it exits before returning.
Shutdown() should return either zero to indicate that device was successfully shutdown, or non-zero to indicate that shutdown failed.

2.4 Main()
The Driver::Startup() method will result in the creation of a new thread of control, which will immediately call Driver::Main().  Every driver must this overload this method.  The overloaded Main() method is generally responsible for translating between some device-specific  API and the standard Player interfaces.  The basic steps are as follows:
- For incoming data:
- Read data from some external device (over a serial port, for example).
- Massage the data into one of the Player interface data structures.
- Write the data to the Player server using Driver::PutData().
- For outgoing data (commands):
- Read commands from the Player server using Driver::GetCommand().
- Massage the command into a device specific format.
- Write the command to the external device.
The Driver::PutData() and Driver::GetCommand() methods manage all of the buffering and locking required to synchronize the driver and server threads.
The overloaded Main() method is also responsible for handling any configuration request sent to the driver; the basic steps are as follows:
- Check for a new request using Driver::GetConfig().
- Do whatever needs to be done for this request.
- Send a response using Driver::PutReply().
Note that your driver must respond, one way or another, to each and every request.  Failing to respond to requests will cause client  programs to "hang" when talking to the driver.
The overloaded Main() must also be capable of graceful termination  (i.e., the function should exit cleanly when the user kills the server).  This can be achieved by calling @c pthread_testcancel() from within the main loop.  This function checks to see if the driver  thread should terminate, and if so, immediately exits Main().  If additional cleanup is required, drivers can also overload Driver::MainQuit(), which is guaranteed to be called on thread termination.

2.5 Instantiation
In order to instantiate a driver, the Player server needs to know two things:
- The driver's name (as it will appear in the configuration file).  
- The driver's @b factory function (used to create a new instance of the driver).  The factory function will generally look something like this:
 
Note that this function may be called multiple times: once for each occurance of the driver name in the configuration file.
Each driver must register itself with the server using the DriverTable::AddDriver(), providing the driver name and factory function.  Drivers must therefore define a registration function as follows:

This function must be called exactly once, on program startup.  For plugin drivers, this can be achieved by defining an initialization function for the shared object:
 
This function will be called when the shared object is loaded, before any drivers are instantiated.  The @c extern @c "C" block is used to prevent C++ name mangling of the intialization function.

2.6 Building the shared library
The example driver includes a Makefile for building shared objects.
To manually build a shared object, try:
$ g++ -Wall -g3 -c exampledriver.cc
$ g++ -shared -nostartfiles -o exampledriver.so exampledriver.o
While the above method will probably work, it is recommended that you use pkg-config to get the compile flags; i.e.,
$ g++ -Wall -g3 `pkg-config --cflags player` -c exampledriver.cc
$ g++ -shared -nostartfiles -o exampledriver.so exampledriver.o

@todo
- Add example of building plugin using libtool.
- Add "theory" from Architecture Chapter of the old manual.

 

3. References:
[1] http://playerstage.sourceforge.net/doc/Player-cvs/player/group__tutorial__plugins.html

 

 

 

 

 

内容概要:本文提出了一种基于非合作博弈理论的居民负荷分层调度模型,并结合双层鲸鱼优化算法(Two-level Whale Optimization Algorithm)进行高效求解,模型与算法均通过Matlab代码实现。研究针对电力系统中居民侧用电负荷的复杂调度问题,引入非合作博弈机制刻画各用户之间的利益竞争关系,实现负荷的分层优化分配;同时设计双层优化架构,上层优化资源配置,下层模拟用户自主决策行为,提升了模型的实用性与合理性。通过智能优化算法求解多层级、非凸非线性的博弈模型,有效提高了调度方案的收敛性与全局寻优能力,适用于现代智能电网中的需求侧管理与能源优化场景。; 适合人群:具备电力系统基础理论知识和Matlab编程能力,从事智能电网、能源优化调度、需求侧管理、博弈论应用等方向的科研人员、高校研究生及工程技术人员。; 使用场景及目标:①应用于居民区电力负荷的分层优化调度系统设计与仿真分析;②为非合作博弈在多主体能源系统建模中的应用提供方法论支持;③利用双层鲸鱼算法解决具有嵌套结构的复杂双层优化问题,提升求解效率与调度方案的可行性。; 阅读建议:建议读者结合提供的Matlab代码深入理解模型构建逻辑与算法实现流程,重点关注博弈模型的效用函数设计、纳什均衡求解思路以及双层优化结构的迭代机制,宜配合实际用电数据开展复现实验以验证模型有效性与鲁棒性。
内容概要:本文围绕基于自适应神经模糊推理系统(ANFIS)智能控制器的可再生能源微电网功率管理系统展开研究,结合Simulink仿真实现,深入探讨了微电网中功率的智能调控与经济机组组合调度问题。通过引入ANFIS控制器,有效应对风能、光伏等可再生能源出力的波动性与不确定性,提升系统运行的稳定性与电能质量。研究内容涵盖微电网多源协调控制策略、功率平衡管理、优化调度模型构建及仿真验证,实现了对分布式电源、储能系统和负荷的协同优化,兼顾经济性与可靠性目标,并通过仿真平台验证了所提方法的有效性与优越性。; 适合人群:具备电力系统、自动化或新能源相关专业背景,熟悉Matlab/Simulink仿真环境,从事微电网能量管理、智能控制、能源优化等领域研究的研究生、科研人员及工程技术人员。; 使用场景及目标:①用于高比例可再生能源接入场景下的微电网能量管理系统研发与教学实践;②为实现微电网功率稳定控制与经济高效运行提供先进的智能控制解决方案;③支撑高水平学术论文复现、科研课题攻关及实际工程项目的仿真验证与方案优化。; 阅读建议:建议结合提供的Simulink模型与相关代码进行动手实践,重点关注ANFIS控制器的设计流程、规则库构建与参数调优方法,并通过与传统PID或MPC控制策略的对比实验,深入理解其在动态响应与鲁棒性方面的优势。同时可进一步拓展文中提出的优化调度逻辑,应用于多目标、多约束的复杂实际应用场景中。
内容概要:本文档聚焦于“直流电机双闭环控制Matlab仿真”,系统阐述了基于Matlab/Simulink平台实现直流电机双闭环控制系统(主要包括速度环与电流环)的设计与仿真全过程。通过构建直流电机的数学模型,结合PI控制器进行调控,实现对电机转速和电枢电流的高精度动态控制,验证控制策略的稳定性与响应性能。文档详细介绍了仿真模型的搭建流程、关键参数的整定方法、系统动态波形的分析手段以及仿真结果的有效性验证,体现了经典自动控制理论在实际电机系统中的工程应用,是电机控制与电力电子技术相结合的典型研究案例。; 适合人群:具备自动控制原理、电机与拖动基础、电力电子技术和Matlab/Simulink仿真能力的电气工程、自动化、机电一体化等专业的本科生、研究生及从事电机驱动系统研发的工程技术人员。; 使用场景及目标:①作为高校课程设计或实验教学材料,帮助学生深入理解双闭环调速系统的工作机理与工程实现;②服务于科研项目,为新型电机控制算法(如滑模、模糊PID等)的开发与性能对比提供基础仿真验证平台;③作为工业界产品前期设计的仿真工具,用于评估不同控制策略在动态响应、抗干扰能力和稳态精度方面的可行性。; 阅读建议:建议读者在学习过程中紧密结合自动控制理论知识,亲手在Simulink环境中搭建完整的双闭环仿真模型,通过反复调整PI控制器的比例与积分参数,观察并分析转速、电流的阶跃响应曲线,从而深刻理解反馈控制的本质、系统稳定性条件以及参数整定对动态性能的影响,进而掌握电机控制系统的设计精髓。
内容概要:本文研究了基于Benders分解与输电网运营商(TSO)和配电网运营商(DSO)协调机制的不确定环境下输配电网双层优化模型,旨在提升高比例可再生能源接入背景下电网系统的协调性与鲁棒性。模型上层以系统整体经济性为目标进行优化调度,下层采用Benders分解实现TSO与DSO之间的信息交互与协同决策,通过引入割平面迭代机制保障求解的收敛性与全局最优性。研究充分考虑新能源出力与负荷需求的不确定性,构建了具有强适应性的双层优化框架,并基于Matlab完成了模型的编程实现与仿真验证,有效解决了多主体、多层级、多不确定性因素耦合下的电力系统优化调度难题。; 适合人群:具备电力系统分析、运筹学与优化理论基础,熟悉Matlab编程环境,从事智能电网、能源互联网、分布式能源集成、电力市场等方向的研究生、科研人员及工程技术人员。; 使用场景及目标:①研究高渗透率可再生能源条件下输配电网协同优化调度策略;②掌握Benders分解在电力系统双层优化建模中的应用方法与实现技巧;③构建TSO-DSO多主体协调机制,实现跨层级电网资源的高效互动与决策解耦;④提升对不确定性建模、分解算法设计及大规模优化问题求解能力。; 阅读建议:建议读者结合Matlab代码逐模块剖析模型构建流程,重点理解Benders割的生成逻辑、主从问题的信息传递机制及收敛判据设定,推荐在标准IEEE测试系统上复现实验以深入掌握模型特性与算法性能。
内容概要:本文系统研究了基于灰狼优化算法(GWO)优化Elman神经网络的方法,并提供了完整的Matlab代码实现。研究重点在于利用灰狼优化算法强大的全局搜索能力,对Elman神经网络的关键参数进行智能优化,从而克服传统训练方法易陷入局部最优的缺陷,显著提升模型在时序预测与非线性系统建模任务中的精度与稳定性。文章详细阐述了Elman网络的动态反馈机制及其在处理时间序列数据方面的优势,构建了GWO与Elman相结合的混合预测框架,涵盖了从模型搭建、参数寻优、仿真测试到结果分析的全流程,特别适用于风电功率预测、电力负荷预测等具有强时变性和不确定性的工程应用场景。; 适合人群:具备一定Matlab编程能力和神经网络基础知识,从事智能优化算法、时间序列预测、电力系统分析或新能源出力预测等相关领域的研究生、科研人员及工程技术人员。; 使用场景及目标:①掌握灰狼优化算法在神经网络超参数优化中的具体实施路径与技术细节;②深入理解Elman递归神经网络与群体智能优化算法融合的建模范式;③将其应用于风电、光伏等新能源发电功率预测及复杂动态系统的建模与仿真,提升预测性能。; 阅读建议:建议读者结合所提供的Matlab代码进行动手实践,重点关注GWO算法与Elman网络的接口设计、适应度函数构建及参数优化迭代过程,可通过调整数据集或迁移至其他预测场景以深化理解和验证模型泛化能力。
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值