CN108153200A - A kind of stereo garage control method of three-layer neural network path planning - Google Patents
A kind of stereo garage control method of three-layer neural network path planning Download PDFInfo
- Publication number
- CN108153200A CN108153200A CN201711482990.7A CN201711482990A CN108153200A CN 108153200 A CN108153200 A CN 108153200A CN 201711482990 A CN201711482990 A CN 201711482990A CN 108153200 A CN108153200 A CN 108153200A
- Authority
- CN
- China
- Prior art keywords
- neural network
- path planning
- controller
- layer
- stereo garage
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/04—Programme control other than numerical control, i.e. in sequence controllers or logic controllers
- G05B19/042—Programme control other than numerical control, i.e. in sequence controllers or logic controllers using digital processors
- G05B19/0423—Input/output
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/20—Pc systems
- G05B2219/25—Pc structure of the system
- G05B2219/25257—Microcontroller
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Feedback Control In General (AREA)
Abstract
The present invention provides a kind of stereo garage control methods of three-layer neural network path planning；Using following steps：1. training pattern：Using the historical data of multi-storied garage parking position movement as training set, the three-layer neural network model of path planning is established；2. data are written：Node parameter in neural network model is written into controller, and the sequence number where write-in in each node parameter, for every node layer, the parameter of two nodes is at most written in each controller；3. receive instruction：Live host RX path planning tasks data；4. subpackage is sent；5. it receives and wraps；It is performed 6. sending.The present invention efficiently uses field resources, can greatly reduce cost, improve resource utilization, and the practical application mode for neural network model provides new thought by the hidden node being considered as controller in neural network model.
Description
Technical field
The present invention relates to a kind of stereo garage control methods of three-layer neural network path planning.
Background technology
In stereo garage path planning control aspect, traditional mode essentially consists in traversal possible path, the spy based on CPU
Property, the mode of traversal take it is longer, and once calculate it is excessively complicated if be easy to cause system stuck, need to make in engineering
It is solved with other a large amount of means, R＆D costs are high.
Development based on neural network algorithm and realization, at present deep learning for programming evaluation problem application increase severely,
By neural network model solve stereo garage path planning problem prove in the lab it is feasible, but put into practice in, due to nerve
Network model is related to the calculating of a large amount of neurons, and GPU special disposals are usually used, and cost is higher, and resource utilization is low.
Invention content
In order to solve the above technical problems, a kind of stereo garage control the present invention provides three-layer neural network path planning
Method, the stereo garage control method of the three-layer neural network path planning is by the way that controller is considered as in neural network model
Node can efficiently use field resources, greatly reduce cost.
The present invention is achieved by the following technical programs.
A kind of stereo garage control method of three-layer neural network path planning provided by the invention；Using following steps：
1. training pattern：Using the historical data of multi-storied garage parking position movement as training set, three layers of god of path planning are established
Through network model；
2. data are written：Node parameter in neural network model is written into controller, and in each node parameter
Sequence number where middle write-in, for every node layer, the parameter of two nodes is at most written in each controller；
3. receive instruction：Live host RX path planning tasks data；
4. subpackage is sent：Path planning task data is sent to controller by live host, and controller is according to the section of write-in
Point parameter, is sorted according to layer where node, is calculated with the calculation of neural network model；
5. it receives and wraps：Result of calculation is sent to live host by the controller for being written with third layer node parameter, scene
Host obtains program results according to result of calculation；
It is performed 6. sending：Live host sends execute instruction, while live host waits for according to program results to controller
It receives next path planning task and enters step 3..
The step 2. in, the parameter of 1~4 neural network model node is written in each controller altogether.
The proximal end server RX path planning tasks data that the scene host is connected from direct communication.
The proximal end server receives stereo garage movement request by internet.
The neural network model is obtained for Kohonen neural network algorithms.
The three-layer neural network model, it is three layers to refer to intermediate hidden layer.
The controller completes control after execute instruction is received, by the executing agency for controlling connection.
The controller is STM32F4 series monolithics.
The beneficial effects of the present invention are：Pass through the hidden node being considered as controller in neural network model, effectively profit
With field resources, cost can be greatly reduced, improves resource utilization, and the practical application mode for neural network model provides
New thought.
Description of the drawings
Fig. 1 is the applicable connection diagram of the present invention；
In figure：101- scenes host, 102- proximal ends server, 103- remote hosts, group of routes inside 20-, inside 201-
Router, 30- controller groups, 301- controllers, 302- executing agencies.
Specific embodiment
Be described further below technical scheme of the present invention, but claimed range be not limited to it is described.
The present invention provides a kind of stereo garage control method of three-layer neural network path planning, for as shown in Figure 1
Three-dimensional garage control system, using following steps：
1. training pattern：Using the historical data of multi-storied garage parking position movement as training set, three layers of god of path planning are established
Through network model；
2. data are written：Node parameter in neural network model is written into controller 301, and is joined in each node
Sequence number where write-in in number, for every node layer, the parameter of two nodes is at most written in each controller 301；
3. receive instruction：Live 101 RX path planning tasks data of host；
4. subpackage is sent：Path planning task data is sent to controller 301,301 basis of controller by live host 101
The node parameter of write-in is sorted according to layer where node, is calculated with the calculation of neural network model；
5. it receives and wraps：Result of calculation is sent to live host by the controller 301 for being written with third layer node parameter
101, live host 101 obtains program results according to result of calculation；
It is performed 6. sending：Live host 101 sends execute instruction, while scene is main according to program results to controller 301
3. next path planning task to be received such as machine 101 simultaneously enters step.
The step 2. in, the parameter of 1~4 neural network model node is written in each controller 301 altogether.
The 102 RX path planning tasks data of proximal end server that the scene host 101 is connected from direct communication.
The proximal end server 102 receives stereo garage movement request by internet.
The neural network model is obtained for Kohonen neural network algorithms.
The three-layer neural network model, it is three layers to refer to intermediate hidden layer.
The controller 301 completes control after execute instruction is received, by the executing agency 302 for controlling connection.
The controller 301 is STM32F4 series monolithics.
In general, what individual node in neural network model carried out is all relatively simple calculating, calculation amount is small, and
In the scheme to each parking stall setting independent control 301, in order to ensure that system is stablized, the performance of controller 301 is generally all
There is redundancy, and be changed to from common STM32F1 series monolithics using STM32F4 series monolithics, cost increases insufficient
20%, performance boost can be more than 80%, therefore by the way that controller 301 is considered as neural network node, efficiently use controller 301
Performance redundancy, cost performance is high.
Claims (8)
1. a kind of stereo garage control method of three-layer neural network path planning, it is characterised in that：Using following steps：
1. training pattern：Using the historical data of multi-storied garage parking position movement as training set, three layers of nerve net of path planning are established
Network model；
2. data are written：Node parameter in neural network model is written into controller (301), and in each node parameter
Sequence number where middle write-in, for every node layer, the parameter of two nodes is at most written in each controller (301)；
3. receive instruction：Live host (101) RX path planning tasks data；
4. subpackage is sent：Path planning task data is sent to controller (301), controller (301) root by live host (101)
According to the node parameter of write-in, sort according to layer where node, calculated with the calculation of neural network model；
5. it receives and wraps：Result of calculation is sent to live host by the controller (301) for being written with third layer node parameter
(101), live host (101) obtains program results according to result of calculation；
It is performed 6. sending：Live host (101) sends execute instruction, while scene is main according to program results to controller (301)
3. next path planning task to be received such as machine (101) simultaneously enters step.
2. the stereo garage control method of three-layer neural network path planning as described in claim 1, it is characterised in that：It is described
Step 2. in, the parameter of 1~4 neural network model node is written in each controller (301) altogether.
3. the stereo garage control method of three-layer neural network path planning as described in claim 1, it is characterised in that：It is described
Proximal end server (102) RX path planning tasks data that live host (101) connects from direct communication.
4. the stereo garage control method of three-layer neural network path planning as claimed in claim 3, it is characterised in that：It is described
Proximal end server (102) receives stereo garage movement request by internet.
5. the stereo garage control method of three-layer neural network path planning as described in claim 1, it is characterised in that：It is described
Neural network model is obtained for Kohonen neural network algorithms.
6. the stereo garage control method of three-layer neural network path planning as described in claim 1, it is characterised in that：It is described
Three-layer neural network model, it is three layers to refer to intermediate hidden layer.
7. the stereo garage control method of three-layer neural network path planning as described in claim 1, it is characterised in that：It is described
Controller (301) completes control after execute instruction is received, by the executing agency (302) for controlling connection.
8. the stereo garage control method of three-layer neural network path planning as described in claim 1, it is characterised in that：It is described
Controller (301) is STM32F4 series monolithics.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711482990.7A CN108153200A (en) | 2017-12-29 | 2017-12-29 | A kind of stereo garage control method of three-layer neural network path planning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711482990.7A CN108153200A (en) | 2017-12-29 | 2017-12-29 | A kind of stereo garage control method of three-layer neural network path planning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108153200A true CN108153200A (en) | 2018-06-12 |
Family
ID=62460355
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711482990.7A Pending CN108153200A (en) | 2017-12-29 | 2017-12-29 | A kind of stereo garage control method of three-layer neural network path planning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108153200A (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2008542859A (en) * | 2005-05-07 | 2008-11-27 | エル ターラー、ステフエン | Device for autonomous bootstrapping of useful information |
| CN103112680A (en) * | 2013-01-31 | 2013-05-22 | 西安科技大学 | Stereo logistics system access cargo path optimization control system and method |
| CN104145281A (en) * | 2012-02-03 | 2014-11-12 | 安秉益 | Neural network computing apparatus and system, and method therefor |
| CN104978601A (en) * | 2015-06-26 | 2015-10-14 | 深圳市腾讯计算机系统有限公司 | Neural network model training system and method |
| CN105986693A (en) * | 2015-02-05 | 2016-10-05 | 深圳怡丰机器人科技有限公司 | Intelligent three-dimensional parking system based on carrying robot and car-loading plate |
| CN106647404A (en) * | 2016-12-27 | 2017-05-10 | 贵州航天南海科技有限责任公司 | Path planning control system for easily extended stereo garage |
| CN106948634A (en) * | 2017-05-15 | 2017-07-14 | 中国电子科技集团公司第三十八研究所 | The operation managing and control system of the tower-type space garage of parallelization Transport Vehicle |
| CN107269054A (en) * | 2017-06-08 | 2017-10-20 | 昆明奥多智能科技有限公司 | A kind of intelligent stereo garage system and control method |
| CN107347069A (en) * | 2017-07-10 | 2017-11-14 | 北京理工大学 | A kind of optimal attack paths planning method based on Kohonen neutral nets |
-
2017
- 2017-12-29 CN CN201711482990.7A patent/CN108153200A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2008542859A (en) * | 2005-05-07 | 2008-11-27 | エル ターラー、ステフエン | Device for autonomous bootstrapping of useful information |
| CN104145281A (en) * | 2012-02-03 | 2014-11-12 | 安秉益 | Neural network computing apparatus and system, and method therefor |
| CN103112680A (en) * | 2013-01-31 | 2013-05-22 | 西安科技大学 | Stereo logistics system access cargo path optimization control system and method |
| CN105986693A (en) * | 2015-02-05 | 2016-10-05 | 深圳怡丰机器人科技有限公司 | Intelligent three-dimensional parking system based on carrying robot and car-loading plate |
| CN104978601A (en) * | 2015-06-26 | 2015-10-14 | 深圳市腾讯计算机系统有限公司 | Neural network model training system and method |
| CN106647404A (en) * | 2016-12-27 | 2017-05-10 | 贵州航天南海科技有限责任公司 | Path planning control system for easily extended stereo garage |
| CN106948634A (en) * | 2017-05-15 | 2017-07-14 | 中国电子科技集团公司第三十八研究所 | The operation managing and control system of the tower-type space garage of parallelization Transport Vehicle |
| CN107269054A (en) * | 2017-06-08 | 2017-10-20 | 昆明奥多智能科技有限公司 | A kind of intelligent stereo garage system and control method |
| CN107347069A (en) * | 2017-07-10 | 2017-11-14 | 北京理工大学 | A kind of optimal attack paths planning method based on Kohonen neutral nets |
Non-Patent Citations (1)
| Title |
|---|
| 吉星宇: "立体车库控制系统设计构想", 《科技创新与应用》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107947845A (en) | Unmanned plane based on communication relay, which is formed into columns, cooperates with target assignment method | |
| CN105894087A (en) | System and method for training parameter set in neural network | |
| CN109491790A (en) | Industrial Internet of Things edge calculations resource allocation methods and system based on container | |
| CN103533097B (en) | A kind of web crawlers download parsing method and device | |
| CN111147307A (en) | Service function chain reliable deployment method based on deep reinforcement learning | |
| CN103365727B (en) | Host load forecasting method in cloud computing environment | |
| CN103731277B (en) | Power-economizing method and energy-saving control apparatus in software defined network | |
| CN109472520A (en) | Multiple no-manned plane method for allocating tasks based on block chain | |
| CN113671987B (en) | Multi-machine distributed time sequence task allocation method based on non-deadlock contract net algorithm | |
| CN108111335B (en) | A kind of method and system of scheduling and link virtual network function | |
| CN110442022A (en) | Grouping consistency control method with time lag isomery multi-agent system | |
| CN109840533A (en) | A kind of applied topology figure recognition methods and device | |
| CN110102058A (en) | The generation method of Virtual Building model, device, processor and terminal in game | |
| CN110533178A (en) | A kind of neural network model training method, apparatus and system | |
| CN109039959A (en) | A kind of the consistency judgment method and relevant apparatus of SDN network rule | |
| CN113037546A (en) | Security virtual network mapping method based on deep reinforcement learning | |
| CN115049071A (en) | Federal learning method based on DAG block chain | |
| CN108880909A (en) | A kind of network energy-saving method and device based on intensified learning | |
| CN111343095B (en) | Method for realizing controller load balance in software defined network | |
| CN109787240A (en) | Distributed Power Flow optimization method and relevant device are coordinated in a kind of main distribution | |
| CN108898221A (en) | The combination learning method of feature and strategy based on state feature and subsequent feature | |
| CN109063978A (en) | Energy Internet resources dispatching method and system, readable storage medium storing program for executing and terminal | |
| CN111461284A (en) | Data discretization method, device, equipment and medium | |
| CN110020046A (en) | A kind of data grab method and device | |
| CN108153200A (en) | A kind of stereo garage control method of three-layer neural network path planning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180612 |