CN113315794A - Hardware architecture of computing system network for online intelligent analysis of blast furnace production - Google Patents

Hardware architecture of computing system network for online intelligent analysis of blast furnace production Download PDF

Info

Publication number
CN113315794A
CN113315794A CN202010118246.4A CN202010118246A CN113315794A CN 113315794 A CN113315794 A CN 113315794A CN 202010118246 A CN202010118246 A CN 202010118246A CN 113315794 A CN113315794 A CN 113315794A
Authority
CN
China
Prior art keywords
blast furnace
computing
node
network
hardware architecture
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
Application number
CN202010118246.4A
Other languages
Chinese (zh)
Inventor
李云涛
毛晓明
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Baoshan Iron and Steel Co Ltd
Original Assignee
Baoshan Iron and Steel Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Baoshan Iron and Steel Co Ltd filed Critical Baoshan Iron and Steel Co Ltd
Priority to CN202010118246.4A priority Critical patent/CN113315794A/en
Publication of CN113315794A publication Critical patent/CN113315794A/en
Pending legal-status Critical Current

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network arrangements or protocols for supporting network services or applications
    • H04L67/01Protocols
    • H04L67/12Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/25Integrating or interfacing systems involving database management systems
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00Machine learning
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network arrangements or protocols for supporting network services or applications
    • H04L67/01Protocols
    • H04L67/10Protocols in which an application is distributed across nodes in the network

Abstract

The invention discloses a hardware architecture of a computing system network for online intelligent analysis of blast furnace production, which is characterized in that a computing hardware platform is formed by a computing subsystem and a production data communication subsystem on the existing blast furnace industrial production control network, and a super-fusion intelligent computing architecture shared by a CPU and a GPU (graphics processing unit) is realized. The computing subsystem comprises a management node, a computing node, a data storage node and a high-speed switch, and is connected by a high-speed switching network, so that different efficiency requirements of software operation in the blast furnace production intelligent system are met; the production data communication subsystem consists of a management node, a database node and a switch, and meets the real-time transmission requirement of production data by adopting gigabit network connection. And a transfer machine NAT communication mode is adopted with a blast furnace industrial production control network, and a firewall and a transfer server are added between the systems, so that the production network and the computing network are isolated, and the communication safety is ensured.

Description

Hardware architecture of computing system network for online intelligent analysis of blast furnace production
Technical Field
The invention relates to a basic hardware framework of intelligent manufacturing process calculation in metallurgical industry, in particular to a hardware framework of a computing system network for online intelligent analysis of blast furnace production.
Background
The blast furnace is a key production procedure of a full-flow iron and steel enterprise, provides high-quality molten iron for iron and steel products, and is the most economic molten iron production process so far. The blast furnace is invented as the most successful smelting container in history, and the production has the characteristics of large time lag and nonlinearity. Meanwhile, the industry accepted "black box" in the blast furnace body is that the metallurgical processes such as the physical process and the chemical reaction in the blast furnace in the production process have no direct detection and observation means, and the internal state of the blast furnace is presumed mainly by means of detection information outside the blast furnace in the actual operation. In order to provide a basis for the production operation of the blast furnace, the domestic modern super-huge blast furnace is basically provided with a relatively complete detection device such as a thermocouple temperature measurement system, a cooling water system, a pressure sensor system, a chemical composition on-line analysis system, a furnace top charge level monitoring system and the like, the three types of data such as temperature, speed and flow in the production of the blast furnace are improved, and the data detection point is about 10000 points. At present, blast furnace production enterprises at home and abroad comprehensively develop intelligent manufacturing work, new technologies based on mature metallurgical theory calculation, production data mining and the like are mainly adopted, accurate judgment and production control on the state of a blast furnace are realized through development of a visual model and a machine learning model, and the current blast furnace production control system architecture is challenged.
The blast furnace control system basically adopts a two-layer architecture of L1 (basic automation layer) + L2 (process automation layer), as shown in fig. 1, wherein L1 mainly adopts PLC/DCS to realize the basic automation of Industrial production, and L2 process automation control adopts an Industrial Personal Computer (IPC Industrial control Computer), i.e. a single conventional Industrial server, which is responsible for completing data acquisition and process model calculation and on-line operation.
However, with the application of new technologies such as industrial big data and artificial intelligence in industrial production, especially in the steel industry, the existing L2-layer process control computer (process machine) is far from meeting the requirements of hardware devices required by visual and intelligent models in blast furnace intelligence in terms of data processing and software computing capabilities, and a hardware architecture with integrated safety, computing power and functionality is required to replace the existing process machine and be integrated with the existing industrial control network, so that a hardware foundation is provided at the level of L2 or above. Aiming at the specific characteristics of large time lag and nonlinearity of blast furnace production data, a new hardware architecture is required to be adopted, the analysis and utilization of the blast furnace production data are realized through machine learning and deep learning modeling, intelligent manufacturing is served, and the automatic control of the blast furnace production is realized.
Disclosure of Invention
Aiming at the defects in the prior art, the invention aims to provide a hardware architecture of a computing system network for online intelligent analysis of blast furnace production, which realizes real-time mining of blast furnace production data and online operation of a model, enhances the accuracy and timeliness of automatic judgment of furnace conditions in the blast furnace production process, improves the stability of blast furnace production and the smooth state of furnace burden, reduces the fuel ratio of blast furnace production and the incidence rate of abnormal furnace conditions, and improves the technical and economic indexes of the blast furnace.
In order to achieve the purpose, the invention adopts the following technical scheme:
a hardware architecture of a computing system network for online intelligent analysis of blast furnace production forms two-way communication with a blast furnace industrial control network, and comprises:
the production data communication subsystem is responsible for completing the two-way communication with the industrial control network of the blast furnace, and a firewall is arranged between the networks;
the calculation subsystem is responsible for providing the calculation power of the GPU and the CPU for machine learning of blast furnace production data and efficient calculation of a functional model;
the software application subsystem is responsible for deploying the machine learning model software of the blast furnace;
the computing subsystem comprises a management node, a control node, a database node and two switches, wherein the two switches are respectively connected with the management node, the control node and the database node to form two independent internal networks.
The two switches are also connected to the production data communication subsystem through a firewall.
One of the two switches is a gigabit switch and the other is a high speed switch.
The software application subsystem comprises a management server, an App server and a graphic server.
The bidirectional communication adopts a transit machine NAT mode.
The management node supports communication with DB2, MongoDB, Oracle and MySQL databases.
The database node is composed of a database server and a large-capacity storage.
The control node is not less than one 1CPU chip server.
The computing node is at least one server configured by 1CPU chip and 1GPU computing card.
The high-speed switch is a Mellanox EDR IB switch.
In the above technical solution, the hardware architecture of the computing system network for online intelligent analysis of blast furnace production provided by the present invention further has the following beneficial effects:
1) the invention isolates the existing process machine network through a firewall, thereby ensuring the safety of the industrial network;
2) the invention flexibly adopts the GPU and the CPU to process the units, can greatly improve the computing capacity of the industrial control system, ensures the response time in the process of various industrial software models, and meets the industrial on-line operation requirement;
3) the super-fusion framework can realize the online operation of artificial intelligence and deep learning algorithm, and obviously improve the intelligent level of blast furnace production data analysis.
Drawings
FIG. 1 is a schematic diagram of a prior art blast furnace industrial control network;
fig. 2 is a network topology diagram of the hardware architecture of the present invention.
Detailed Description
The technical scheme of the invention is further explained by combining the drawings and the embodiment.
Referring to fig. 2, the hardware architecture of the computing system network for online intelligent analysis of blast furnace production provided by the present invention forms a two-way communication with the blast furnace industrial control network, determines the type and the number of computing units of the processor for system computation according to the computation requirements of the industrial model, particularly the blast furnace control model, and determines the number of nodes of the computing units. The hardware architecture of the present invention comprises:
the production data communication subsystem is responsible for completing the two-way communication with the industrial control network of the blast furnace and adopts a NAT mode including but not limited to a transit machine to carry out cross-network communication;
the calculation subsystem is responsible for providing the calculation power of the GPU and the CPU for machine learning of blast furnace production data and efficient calculation of a functional model;
and the software application subsystem is responsible for the deployment of the machine learning model software of the blast furnace.
The production data communication subsystem comprises a switch, and a management node and a database node which are respectively connected with the switch. The exchanger is a kilomega exchanger and is connected with the existing industrial control network of the blast furnace to realize the communication with the L1 controller.
The database nodes, the kilomega switch and the firewall are connected with the existing blast furnace industrial control network through super kilomega network lines to form a 1GLAN local area network, and the connecting network lines comprise six types of lines, super six types of lines and seven types of lines. The existing blast furnace industrial control network can be but is not required to be provided with a data transfer server which is connected with a local area network database server to realize data communication.
The computing subsystem comprises a control node, a computing node and a data storage node which are internally connected by adopting a high-speed switch network, forms an InfiniBand architecture high-speed connection network and forms an InfiniBand network. The network is used for connecting a computing node and a data storage node, the network data transmission speed reaches 40-100 Gb/s, and high-performance computing of the system is achieved. Infiniband network adopts Switch-IBTMSwitches, model numbers include but are not limited to Mellanox FDR, EDR InfiniBand Switch.
And the software application subsystem comprises a management server, an App server and a graphic server.
The bidirectional communication adopts a transit machine NAT mode.
The management node supports communication with DB2, MongoDB, Oracle and MySQL databases.
The database node is composed of a database server and a large-capacity storage.
The control node is not less than one 1CPU chip server.
The computing node is not less than one 1CPU chip +1GPU computing card configuration server.
The hardware architecture of the invention at least comprises three servers, which realize the functions of the computing node, the storage node and the control node, wherein the database service and the transfer node required by the system can be realized by one server at the control node.
The inner side of the firewall is provided with at least two connection ports for connecting the data server and the control server, and the outer end of the firewall is connected with the database transit server or the L1 controller.
The invention is further illustrated below with reference to 3 examples.
The parameter settings for the 3 examples are shown in table 1:
table 1 example parameter set-up
Example 1
Hardware architecture major gather 2800m3The capacities of a data server and a sharing server are both 16T capacity mechanical hard disks; the computing node adopts a commercial wave 2U architecture server NF5288M5, and the server is configured with a 2CPU +1GPU computing core and a memory 128G; the control node adopts a wave 2U architecture server NF5280M5, double CPUs and a memory 64G; the high-speed network adopts an MSB7790-ES2F switch, and the gigabit network adopts an RJ45 switch. The system and the production control system communicate with each other for 10min, real-time production data are modeled by adopting a CNN algorithm, the real-time requirement of intelligent monitoring of the blast furnace operation type for 1min is met, and 10min frequency data are stored in the whole furnace service period.
Example 2
Hardware architecture mainly collects 1 seat 4966m3The thermal power even data, the thermal load data and the blast furnace production state information of the super-huge type blast furnace body, and the capacities of a data server and a shared server are both 72T capacity hard disks; the computing node adopts 3 commercial wave 2U architecturesA server NF5288M5, a server configuration 2CPU +2GPU computing core and a memory 128G; the control node adopts 1 wave 2U architecture server NF5280M5, double CPUs and a memory 96G; the high-speed network adopts EDR InfiniBand Switch, and the gigabit network adopts RJ45 Switch, and the network line is supporting. The system and the production control system communicate with each other at the frequency of 1S, monitor and respond to the real-time production data by 1S, and meet the intelligent monitoring requirement of the blast furnace.
Example 3
Hardware architecture mainly collects 4 seats 4966m3The capacity of a data server and a shared server adopts a 72T capacity hard disk; the computing node adopts 6 commercial wave 2U architecture servers NF5288M5, and the servers are configured with a 2CPU +2GPU computing core and a memory 128G; the control node adopts 2 wave 2U architecture servers NF5280M5, double CPUs and a memory 96G; the high-speed network adopts EDR InfiniBand Switch, and the gigabit network adopts RJ45 Switch, and the network line is supporting. The system and the production control system communicate with each other at the frequency of 1S, monitor and respond to the real-time production data by 1S, and meet the intelligent monitoring requirement of the blast furnace.
The hardware architecture of the present invention is implemented as follows:
firstly, determining the node storage capacity of a data server, the capacity of a shared storage node and the data server according to the number of blast furnaces needing to acquire production data and data points;
secondly, determining the number of computing nodes and the computational power configuration of a node server according to the computational requirements of an intelligent analysis model algorithm, wherein the number of computing nodes and the computational power configuration of the node server are mainly the number of CPUs (central processing units) and the number of GPUs (graphic processing units);
thirdly, counting the number of the running software, and determining the number of the management and application server nodes;
fourthly, all the node servers are configured with different network cards according to the figure 1, and the shared storage node, the computing node and the management application node are connected through an IB switch to establish an internal high-speed network;
fifthly, all the node servers are connected through the gigabit switch to establish an internal gigabit network;
and sixthly, the kilomega switch and the data transfer server are connected with the firewall to establish a communication loop of the internal kilomega network and the industrial control system network, so that the production data message bidirectional communication is realized.
It should be understood by those skilled in the art that the above embodiments are only for illustrating the present invention and are not to be used as a limitation of the present invention, and that changes and modifications to the above described embodiments are within the scope of the claims of the present invention as long as they are within the spirit and scope of the present invention.

Claims (10)

1. A hardware architecture of a computing system network for online intelligent analysis of blast furnace production forms two-way communication with a blast furnace industrial control network, and is characterized by comprising the following components:
the production data communication subsystem is responsible for completing the two-way communication with the industrial control network of the blast furnace, and a firewall is arranged between the networks;
the calculation subsystem is responsible for providing the calculation power of the GPU and the CPU for machine learning of blast furnace production data and efficient calculation of a functional model;
the software application subsystem is responsible for deploying the machine learning model software of the blast furnace;
the computing subsystem comprises a management node, a control node, a database node and two switches, wherein the two switches are respectively connected with the management node, the control node and the database node to form two independent internal networks.
2. The hardware architecture of a computing system network for online intelligent analysis of blast furnace production according to claim 1, wherein: the two switches are also connected to the production data communication subsystem through a firewall.
3. The hardware architecture of a computing system network for online intelligent analysis of blast furnace production according to claim 1, wherein: one of the two switches is a gigabit switch and the other is a high speed switch.
4. The hardware architecture of a computing system network for online intelligent analysis of blast furnace production according to claim 1, wherein: the software application subsystem comprises a management server, an App server and a graphic server.
5. The hardware architecture of a computing system network for online intelligent analysis of blast furnace production according to claim 1, wherein: the bidirectional communication adopts a transit machine NAT mode.
6. The hardware architecture of a computing system network for online intelligent analysis of blast furnace production according to claim 2, wherein: the management node supports communication with DB2, MongoDB, Oracle and MySQL databases.
7. The hardware architecture of a computing system network for online intelligent analysis of blast furnace production according to claim 2, wherein: the database node is composed of a database server and a large-capacity storage.
8. The hardware architecture of a computing system network for online intelligent analysis of blast furnace production as claimed in claim 3, wherein: the control node is not less than one 1CPU chip server.
9. The hardware architecture of a computing system network for online intelligent analysis of blast furnace production as claimed in claim 3, wherein: the computing node is at least one server configured by 1CPU chip and 1GPU computing card.
10. The hardware architecture of a computing system network for online intelligent analysis of blast furnace production as claimed in claim 3, wherein: the high-speed switch is a Mellanox EDR IB switch.
CN202010118246.4A 2020-02-26 2020-02-26 Hardware architecture of computing system network for online intelligent analysis of blast furnace production Pending CN113315794A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010118246.4A CN113315794A (en) 2020-02-26 2020-02-26 Hardware architecture of computing system network for online intelligent analysis of blast furnace production

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010118246.4A CN113315794A (en) 2020-02-26 2020-02-26 Hardware architecture of computing system network for online intelligent analysis of blast furnace production

Publications (1)

Publication Number Publication Date
CN113315794A true CN113315794A (en) 2021-08-27

Family

ID=77369947

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010118246.4A Pending CN113315794A (en) 2020-02-26 2020-02-26 Hardware architecture of computing system network for online intelligent analysis of blast furnace production

Country Status (1)

Country Link
CN (1) CN113315794A (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107733696A (en) * 2017-09-26 2018-02-23 南京天数信息科技有限公司 A kind of machine learning and artificial intelligence application all-in-one dispositions method
CN107942912A (en) * 2017-12-19 2018-04-20 西安科技大学 A kind of coalmine production control management integrated system
CN208013975U (en) * 2018-04-23 2018-10-26 苏州超集信息科技有限公司 The hardware device of on-line intelligence ability platform
CN108763550A (en) * 2018-06-01 2018-11-06 东北大学 Blast furnace big data application system
CN109739203A (en) * 2019-02-25 2019-05-10 南京世界村云数据产业集团有限公司 A kind of industrial network Border Protection system
CN110647399A (en) * 2019-09-22 2020-01-03 南京信易达计算技术有限公司 High-performance computing system and method based on artificial intelligence network

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107733696A (en) * 2017-09-26 2018-02-23 南京天数信息科技有限公司 A kind of machine learning and artificial intelligence application all-in-one dispositions method
CN107942912A (en) * 2017-12-19 2018-04-20 西安科技大学 A kind of coalmine production control management integrated system
CN208013975U (en) * 2018-04-23 2018-10-26 苏州超集信息科技有限公司 The hardware device of on-line intelligence ability platform
CN108763550A (en) * 2018-06-01 2018-11-06 东北大学 Blast furnace big data application system
CN109739203A (en) * 2019-02-25 2019-05-10 南京世界村云数据产业集团有限公司 A kind of industrial network Border Protection system
CN110647399A (en) * 2019-09-22 2020-01-03 南京信易达计算技术有限公司 High-performance computing system and method based on artificial intelligence network

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
李金柱: ""通钢七号高炉专家系统的开发与应用"", 《中国优秀硕士学位论文全文数据库》 *

Similar Documents

Publication Publication Date Title
CN102915027B (en) Blast furnace smelting expert system built based on pattern recognition technology and method thereof
CN103268069B (en) Based on the adaptive prediction control method of Hammerstein model
CN102263436B (en) On-line monitoring system for energy-saving scheduling and real-time coal consumption of power grid
CN100550003C (en) The implementation method of chip-on communication of built-in isomerization multicore architecture interconnection organisational level
CN113315794A (en) Hardware architecture of computing system network for online intelligent analysis of blast furnace production
CN111556032A (en) Industrial big data processing system based on artificial intelligence algorithm
CN201804331U (en) Date deduplication system based on co-processor
CN104699654B (en) One kind is based on interconnection interconnection adaption system and method between CHI on-chip interconnection buss and QPI pieces
CN108107849A (en) A kind of automatic Control System For Hot Strip Mill Line
CN203116168U (en) Header-system heating power plant automatic generation control (AGC) system
CN206310967U (en) Twin-stage cooling device
CN110197320B (en) Power distribution network overload risk assessment method with multi-terminal flexible multi-state switch
CN107091972A (en) A kind of active power distribution network Fault Locating Method based on improvement population
CN206728062U (en) Combustion engine remote online diagnostic system data transmission link device
CN205844878U (en) Electrolytic aluminium automatization centralized control system
CN206003081U (en) A kind of synchronous serial interface association of industrial computer processes managing device and industrial computer
CN108733769A (en) A kind of real-time fuel consumption based on car running computer data and intelligent search algorithm
CN105183013B (en) The balance of steam method of steaminess pipe network system and steaminess pipe network system
CN110012075A (en) Data transmission method and relevant device
CN106789228A (en) Increase the mapping method of interchanger and controller in a kind of software defined network newly
CN102520701B (en) Control method of solution preparation center shared by multiple continuous steel annealing production lines
CN107324216A (en) Intelligent speed changing control method, apparatus and system
CN113234874B (en) Blast furnace fault monitoring and self-healing control system based on dynamic Bayesian network
CN201886333U (en) Double-microcomputer system of excitation regulator capable of realizing function of dual-computer hot-standby
CN210015341U (en) Remote centralized monitoring system of urban water supply pipe network

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