Domestic industrial intelligent edge server and adaptation method, device and equipment thereof

Abstract

The application discloses a kind of localization industrial intelligent edge servers and its adaptation method, device and equipment, the method includes: to each group of localization central processing unit and localization AI image processor is carried out hardware performance index Weighted score;With the weight score result meets the preset score condition localization central processing unit sample and localization AI image processor sample, as the localization central processing unit and localization AI image processor of edge server;According to the preset hardware adaptation strategy, the hardware adaptation scheme of both localization AI image processor and localization central processing unit is generated;The industrial application software and industrial protocol of edge server are configured with hardware adaptation scheme adaptation;The system software of edge server is configured with hardware adaptation scheme, industrial application software and industrial protocol three adaptations.This application can automatically adapt each component of the system architecture of edge server and realize edge server localization.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, and in particular to a localized industrial intelligent edge server and an adaptation method, device and equipment thereof. BACKGROUND

[0002] In many industry application scenarios, such as road management and automatic driving of intelligent transportation, quality detection and equipment monitoring of intelligent manufacturing, disease monitoring and auxiliary diagnosis of intelligent medicine, energy saving and carbon reduction and digital energy control of intelligent carbon reduction, and intelligent ocean, the explosive growth of the number of Internet of Things terminal devices and the increasing demand for video monitoring in these application scenarios require that the total amount of data generated by each Internet of Things terminal device is increasingly large, especially image data, which puts forward higher technical requirements for the real-time and reliability of data processing. At present, edge servers are generally used to process the data of each Internet of Things terminal device in these application scenarios. The main features of edge computing of edge servers include: low delay, low bandwidth operation and security; because the computing power is deployed near the device side, the response real-time performance is strong; because it is close to the user, it does not require high transmission bandwidth; and because data is collected, analyzed and processed locally, the opportunity of data exposure in public network is reduced, data privacy is protected and it is safer.

[0003] At present, the core components of edge servers are all from abroad, which may be subject to foreign technology blockade. In order to avoid the risk of foreign technology blockade of the core components of edge servers, it is necessary to adjust the system architecture of traditional edge servers and realize localization. However, if manual localization adaptation is performed on each component of the system architecture of the edge server, the efficiency of the localization adaptation of the edge server is low. SUMMARY

[0004] The embodiment of the present application provides a localized industrial intelligent edge server and an adaptation method, device and equipment thereof, which can automatically adapt each component of the system architecture of the edge server to realize localization of the edge server.

[0005] An embodiment of the present application provides an adaptation method of a localized industrial intelligent edge server, which comprises:

[0006] The weight score of the hardware performance index is performed on each set of localized hardware samples in the set of localized hardware samples to be evaluated; the set of localized hardware samples comprises: a plurality of sets of samples of localized central processing units and localized AI image processors;

[0007] The localized central processing unit sample and the localized AI image processor sample that meet the preset score condition of the weight score result are used as the localized central processing unit and the localized AI image processor of the localized industrial intelligent edge server.

[0008] According to the preset hardware adaptation strategy, a hardware adaptation scheme of the domestic AI image processor and the domestic central processor is generated;

[0009] The industrial application software and the industrial protocol of the domestic industrial intelligent edge server are configured to adapt to the hardware adaptation scheme;

[0010] The system software of the domestic industrial intelligent edge server is configured to adapt to the hardware adaptation scheme, the industrial application software and the industrial protocol.

[0011] As an improvement of the above-mentioned scheme, after the system software of the domestic industrial intelligent edge server is configured to adapt to the hardware adaptation scheme, the industrial application software and the industrial protocol, the method further comprises:

[0012] The performance stability test and the data processing pressure test are performed on the adapted domestic industrial intelligent edge server under the preset test scenario.

[0013] As an improvement of the above-mentioned scheme, the performance stability test comprises: whether the performance indicators of the domestic industrial intelligent edge server under the preset test scenario are abnormal, whether the domestic industrial intelligent edge server under the preset test scenario is down or application is suspended; the performance indicators comprise: the highest working frequency of the domestic central processor, the utilization rate of the domestic central processor, the temperature of the domestic central processor, the power consumption of the domestic central processor, the peak computing power of the domestic Al image processor, the utilization rate of the domestic Al image processor, the temperature of the domestic Al image processor, and the power consumption of the domestic Al image processor;

[0014] The data processing pressure test comprises: whether the data processing indicators of the domestic industrial intelligent edge server under the preset test scenario are abnormal; the data processing indicators comprise: data transmission speed, data transmission speed stability, and data processing response time.

[0015] As an improvement of the above-mentioned scheme, the hardware performance indicators comprise: the performance indicators of the domestic central processor, the performance indicators of the domestic Al image processor, and the compatible performance indicators of the domestic central processor and the domestic Al image processor;

[0016] The scoring conditions comprise: the total score of the hardware performance indicators is greater than a preset score threshold or the total score is maximum.

[0017] As an improvement of the above-mentioned scheme, the weight score of the hardware performance indicators for each group of domestic hardware samples in the group of domestic hardware samples to be evaluated comprises:

[0018] score weight distribution is performed on each hardware performance index of each set of the domestic hardware sample to be evaluated, so as to obtain a score weight coefficient of each hardware performance index;

[0019] According to the score weight coefficient of each hardware performance index of each set of the domestic hardware sample, the total score of each hardware performance index is counted.

[0020] As an improvement of the above-mentioned scheme, the hardware adaptation scheme of the domestic Al image processor and the domestic central processor is generated according to the preset hardware adaptation strategy, comprising:

[0021] According to the packaging mode, the number of packaging pins and the model of the north bridge chip of the domestic central processor, a mainboard packaging scheme of the domestic central processor is generated;

[0022] According to the performance characteristics of the PClE interface of the domestic Al image processor, the parameters of the IP core of the PClE interface are set;

[0023] According to the DMA transmission mode, a design scheme of the PCIE bus connecting the domestic Al image processor and the domestic central processor is generated;

[0024] According to the operation function interface of the equipment file of the domestic industrial intelligent edge server, the linking function function of the I / O controller of the domestic Al image processor is linked.

[0025] As an improvement of the above-mentioned scheme, the industrial application software and the industrial protocol of the domestic industrial intelligent edge server adapted to the hardware adaptation scheme are configured, comprising:

[0026] The data acquisition protocol of the domestic industrial intelligent edge server adapted to the hardware adaptation scheme is configured;

[0027] The Haproxy configuration service is performed for the domestic industrial intelligent edge server, and a start script is created to form the IP load balancing mechanism of the domestic industrial intelligent edge server;

[0028] The image data processing algorithm of the domestic industrial intelligent edge server adapted to the hardware adaptation scheme is configured.

[0029] Another embodiment of the present application provides a domestic industrial intelligent edge server adapted by the adaptation method of the domestic industrial intelligent edge server of any of the above-mentioned embodiments, comprising:

[0030] A mainboard;

[0031] A data acquisition module is arranged on the mainboard and is configured to acquire external data transmitted by an external device, wherein the external data comprises image data and non-image data.

[0032] A domestic AI image processor is arranged on the mainboard and is connected with the data acquisition module, and is configured to perform data processing on the image data.

[0033] A domestic central processor is arranged on the mainboard and is connected with the domestic AI image processor and the data acquisition module, and is configured to perform data processing on the non-image data, and perform corresponding operations according to the data processing result of the non-image data and the data processing result of the image data.

[0034] The weight scores of the hardware performance indicators of the domestic AI image processor and the domestic central processor satisfy a preset score condition.

[0035] The hardware adaptation schemes of the two satisfy a preset hardware adaptation strategy.

[0036] The domestic industrial intelligent edge server is adapted with system software, industrial application software and industrial protocols corresponding to the hardware adaptation schemes of the two.

[0037] As an improvement of the above-mentioned scheme, a heat dissipation device is arranged on the mainboard, and the heat dissipation device comprises a heat dissipation fan, a heat dissipation cover and a cooling liquid pipe; the heat dissipation cover is tightly attached to the outer surface of the domestic central processor, the cooling liquid pipe is arranged on the outer surface of the heat dissipation cover, and the air outlet side of the heat dissipation fan is aligned with the outer surface of the heat dissipation cover.

[0038] As an improvement of the above-mentioned scheme, a power module and a power management chip are arranged on the mainboard; the input end of the power management chip is connected with the power module, the output end of the battery management chip is connected with the data acquisition module, the domestic AI image processor and the domestic central processor, and the power management chip is configured to monitor the power supply state of the power module and control the power supply to the data acquisition module, the domestic AI image processor and the domestic central processor according to the power supply state of the power module.

[0039] Another embodiment of the present application provides a domestic industrial intelligent edge server adaptation device, which comprises:

[0040] A weight score module is configured to perform weight scoring on the hardware performance indicators of each domestic hardware sample in a domestic hardware sample group to be evaluated, wherein the domestic hardware sample group comprises a plurality of groups of samples of domestic central processors and domestic AI image processors.

[0041] The hardware selection module is configured to select, as the domestic central processing unit and the domestic AI image processor of the domestic industrial intelligent edge server, the domestic central processing unit sample and the domestic AI image processor sample whose weight score result satisfies a preset score condition.

[0042] The hardware adaptation module is configured to generate, according to a preset hardware adaptation strategy, a hardware adaptation scheme of the domestic Al image processor and the domestic central processing unit.

[0043] The software protocol configuration module is configured to configure the industrial application software and the industrial protocol of the domestic industrial intelligent edge server which are adapted to the hardware adaptation scheme.

[0044] The system configuration module is configured to configure the system software of the domestic industrial intelligent edge server which is adapted to the hardware adaptation scheme, the industrial application software and the industrial protocol.

[0045] Another embodiment of the present application provides a kind of adaptation equipment of domestic industrial intelligent edge server, including processor, memory and the computer program stored in the memory and being configured to be executed by the processor, the processor executes the computer program when realizing the adaptation method of the domestic industrial intelligent edge server described in above-mentioned application embodiment described in above-mentioned application embodiment.

[0046] Another embodiment of the present application provides a kind of computer readable storage medium, the computer readable storage medium includes stored computer program, wherein, when the computer program runs, control the adaptation method of the domestic industrial intelligent edge server described in above-mentioned application embodiment described in above-mentioned application embodiment is executed by the equipment where the computer readable storage medium is located.

[0047] Compared with the prior art, one of the above technical solutions has the following advantages:

[0048] The localization central processing unit sample and the localization AI image processor sample whose weight score results meet the preset score condition are taken as the localization central processing unit and the localization AI image processor of the localization industrial intelligent edge server, so that the hardware performance and compatibility of the localization central processing unit and the localization AI image processor of the localization industrial intelligent edge server are both good; then, according to a preset hardware adaptation strategy, a hardware adaptation scheme of the localization AI image processor and the localization central processing unit is generated; then, the industrial application software and the industrial protocol of the localization industrial intelligent edge server that are adapted to the hardware adaptation scheme are configured; finally, the system software of the localization industrial intelligent edge server that is adapted to the hardware adaptation scheme, the industrial application software and the industrial protocol is configured; in this way, the system architecture components of the edge server can be automatically adapted. In addition, the localization edge server not only has security in data processing, but also the localization industrial intelligent edge server is specially used for image data processing through the localization AI image processor, and the localization central processing unit is specially used for non-image data processing, so that the localization industrial intelligent edge server can meet the processing demand of massive data.

[0049] Of course, it is not necessary for any product embodying the present application to achieve all of the above advantages simultaneously. BRIEF DESCRIPTION OF DRAWINGS

[0050] Figure 1 FIG. 1 is a flow diagram of a localization industrial intelligent edge server adaptation method according to an embodiment of the present application;

[0051] Figure 2 FIG. 1 is a flow diagram of a localization industrial intelligent edge server adaptation method according to an embodiment of the present application;

[0052] Figure 3 FIG. 1 is a flow diagram of a localization industrial intelligent edge server adaptation method according to an embodiment of the present application;

[0053] Figure 4 FIG. 1 is a flow diagram of a localization industrial intelligent edge server adaptation method according to an embodiment of the present application;

[0054] Figure 5 FIG. 1 is a flow diagram of a localization industrial intelligent edge server adaptation method according to an embodiment of the present application; DETAILED DESCRIPTION

[0055] With reference to the drawings and the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments of the present application, all the other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of the present application.

[0056] The terms "first", "second", "third" in the present document are only for descriptive purpose, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features.

[0057] In the present document, the phrase "embodiment" means that the specific features, structures or properties described in connection with the embodiment can be included in at least one embodiment of the present application. The phrase appears at various places in the specification does not necessarily all refer to the same embodiment, nor does it necessarily refer to a separate or alternative embodiment in isolation from other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0058] As an example, the present embodiment and the server described below can be a domestic industrial intelligent edge server. The main features of edge computing of the edge server include low delay, low bandwidth operation and security. Because the computing power is deployed near the device side, the response is real-time. Because it is close to the user, it does not require high transmission bandwidth. Because data is collected, analyzed and processed locally, the opportunity of data exposure in public network is reduced, data privacy is protected and it is safer. The edge computing mode of the edge server can process data without network connection (such as without streaming or in the case of cloud data storage) by using AI algorithms processed locally on hardware devices, so that the data of the device can be processed locally and timely without relying on cloud processing. At present, with the large-scale deployment of 5G network, the characteristics of super-high speed, large concurrency and ultra-low latency of 5G network make the cost of edge computing service decrease and the demand increase. The local ecological system composed of 5G+edge computing can greatly meet the network, computing and data processing needs of enterprises and factories, and promote the digital innovation of the industry.

[0059] Specifically, the core of the localized industrial intelligent edge server is a localized central processing unit 4. Currently, the central processing units and other devices such as AI image processors of edge servers basically use foreign products. For example, the central processing unit 4 uses Intel, and the AI image processor 3 uses Nvidia. If all core parts of the edge server use foreign products, there is a risk of being blocked by foreign technology. Therefore, it is more important to complete the hardware construction and software adaptation of the edge server localization. Therefore, the core components of the localized industrial intelligent edge server of the embodiment of the application can use domestic components. For example, the localized central processing unit 4 can use the 3A5000L, 3A5000 or 3C5000 processor of the LoongArch for the localized industrial intelligent edge server of the LoongArch, which is developed and designed based on the LoongArch self-instruction system architecture. It can realize powerful application capability in data center, cloud computing and digital infrastructure. The localized central processing unit 4 can also use the Shenwei 26010 CPU or the Kunpeng CPU of HiSilicon, etc. In addition, the localized AI image processor 3 can use the CAISA chip of Kunyun Technology. The CAISA chip is equipped with four CAISA 3.0 engines, with more than 16,000 MAC (multiply-accumulate) units, and the peak performance can reach 10.9TOPs. The chip uses 28nm process, communicates with the host processor through PCIe 3.0x4 interface, and has double DDR channel, which can provide more than 340Gbps bandwidth for each CAISA chip. It can be understood that the localized AI image processor 3 can also use the DeepEye1000 chip of CloudEye or the Jinghong7100 chip of Saifang Technology, etc. without specific limitation. When the localized industrial intelligent edge server uses the localized central processing unit 4 and the localized AI image processor 3, the hardware and software of the localized industrial intelligent edge server need to be adapted, so as to realize the localization of the localized industrial intelligent edge server. Therefore, as one of the purposes of the embodiment of the application, the purpose of the following embodiment of the application is how to realize the software and hardware adaptation of the localized industrial intelligent edge server.

[0060] Referring to Figure 1 is a flowchart of a localized industrial intelligent edge server adaptation method provided by an embodiment of the application. The method is executed by a localized industrial intelligent edge server adaptation device. The localized industrial intelligent edge server adaptation device can be a personal computer or a localized industrial intelligent edge server, etc. without specific limitation. The method comprises steps S10-S14:

[0061] S10, performing weight scoring of hardware performance indicators on each set of localized hardware samples in the set of localized hardware samples to be evaluated;

[0062] The domestic hardware sample group includes: a plurality of samples of domestic central processing units 4 and domestic AI image processors 3; and the hardware performance indicators include: performance indicators of the domestic central processing units 4, performance indicators of the domestic AI image processors 3, and compatibility performance indicators of the domestic central processing units 4 and the domestic AI image processors 3.

[0063] Specifically, the step S10 includes:

[0064] S100, score weight distribution is performed on each hardware performance indicator of each domestic hardware sample in the domestic hardware sample group to be evaluated, to obtain a score weight coefficient of each hardware performance indicator;

[0065] S101, according to the score weight coefficient of each hardware performance indicator of each domestic hardware sample, total scores of the hardware performance indicators are counted.

[0066] Specifically, the domestic central processing units 4 and the domestic AI image processors 3 meeting the requirements of industrial application are selected through the steps S100 and S101. The two cores (domestic central processing units 4 and domestic AI image processors 3) of the domestic industrial intelligent edge server have requirements for running efficiency and stability, and have compatibility requirements for the two cores. In view of the above requirements, the preferred idea of the hardware of the two cores of the domestic industrial intelligent edge server is as follows:

[0067] (1) A plurality of domestic central processing unit 4 samples and a plurality of domestic AI image processor 3 samples are found respectively as domestic hardware sample groups to be evaluated. It is assumed that the domestic central processing unit 4 samples of the domestic hardware sample group are A1…A n , a total of n; and the domestic AI image processor 3 samples are B1…B m , a total of m.

[0068] (2) In view of the requirements of the domestic central processing units 4 and the domestic AI image processors 3 for industrial application, the performance indicators of the domestic central processing units 4, the performance indicators of the domestic AI image processors 3, and the compatibility performance indicators of the two cores of the domestic central processing units 4 and the domestic AI image processors 3 are set. Among them, the performance indicators of the domestic central processing units 4 have requirements such as single-core running frequency and running core number; the performance indicators of the domestic AI image processors 3 have requirements such as peak computing performance and number of supported image deep learning algorithm; and the compatibility performance indicators of the two cores have requirements such as instruction set compatibility of the two cores. It is assumed that the domestic central processing unit 4 has 1 indicator, and the weight coefficients of these indicators are α1…α l , then The performance indicators of the domestic AI image processor 3 are collectively p, and the weight coefficients of these indicators are β1…β p Then The compatibility indicators are collectively q, and the weight coefficients of these indicators are γ1…γ q Then The weight vectors of the performance indicators of the domestic central processor 4, the performance indicators of the domestic AI image processor 3, and the compatibility performance indicators of the two cores are a = [α1…α l ] T , β = [β1…β p ] T , γ = [γ1…γ g ] T .

[0069] (3) Let be the score of the performance indicators of the domestic central processor 4, the performance indicators of the domestic AI image processor 3, and the compatibility performance indicators of the two cores, respectively. Then, the vectors of the score of the performance indicators of the domestic central processor 4, the performance indicators of the domestic AI image processor 3, and the compatibility performance indicators of the two cores are and the score weight coefficients of the three types of indicators of the domestic central processor 4, the performance indicators of the domestic AI image processor 3, and the compatibility performance indicators of the two cores are ω1, ω2, ω3, respectively, and the weight vector of the overall score is ω = [ω1, ω2, ω3] T , the total score p of the three types of indicators of each group of core samples is

[0070]

[0071] (4) Select one domestic central processor 4 sample and one domestic AI image processor 3 sample from the domestic hardware sample group as a group of domestic hardware samples, a total of m×n groups. The score of the domestic central processor 4 sample A i and the domestic AI image processor 3 sample B j is According to the formula in (3), m×n scores are calculated. Select the domestic central processor 4 and the domestic AI image processor 3 samples with the highest score to complete the optimization of the hardware.

[0072] According to the requirements of (3), the score of each item can be continuously scored according to different degrees according to specific requirements, or can be ladder scored according to different levels. According to the requirements of (2) and (3), the weights α, β, γ, and ω can be obtained according to expert experience.

[0073] S11, the weight score result of the localization central processor 4 sample and the localization AI image processor 3 sample meeting the preset score condition is taken as the localization central processor 4 and the localization AI image processor 3 of the localization industrial intelligent edge server;

[0074] The score condition includes that the total score of the hardware performance index is greater than a preset score threshold or the total score is maximum. That is, based on the above performance index, the localization central processor 4 sample and the localization AI image processor 3 sample meeting the preset score condition are selected, so as to complete the optimization of the two core hardware of the localization industrial intelligent edge server.

[0075] S12, according to the preset hardware adaptation strategy, the hardware adaptation scheme of the localization AI image processor 3 and the localization central processor 4 is generated;

[0076] Specifically, the step S12 includes steps S120-S123:

[0077] S120, according to the packaging mode, the number of packaging pins and the model of the north bridge chip of the localization central processor 4, the mainboard 1 packaging scheme of the localization central processor 4 is generated;

[0078] Specifically, according to the packaging mode (such as LGA, BGA, PGA packaging mode) of the selected localization central processor 4 and the corresponding number of packaging pins, and considering the model of the north bridge chip which directly communicates with the localization central processor 4, the mainboard 1 packaging scheme of the localization central processor 4 is determined, and in the subsequent assembly of the localization industrial intelligent edge server, the localization central processor 4 can be inserted into the mainboard 1 through the localization central processor 4 packaging interface on the mainboard 1 according to the mainboard 1 packaging scheme. The localization AI image processor 3 is inserted into the mainboard 1 through the hardware interface (PCle) of the localization AI image processor 3 on the mainboard 1. Through the mainboard 1 packaging scheme of the localization central processor 4, the purpose is mainly to prevent the localization central processor 4 packaged on the mainboard 1 from interfering with other devices. The technical personnel can test the packaging mode, the number of packaging pins and the model of the north bridge chip of different localization central processors 4 in advance, and determine which mainboard 1 packaging scheme is suitable for the localization central processor 4, and then set the correspondence between the three and the mainboard 1 packaging scheme in advance. Subsequently, the mainboard 1 packaging scheme of the localization central processor 4 can be configured according to the correspondence and the step S120.

[0079] S121, according to the performance characteristics of the PCIE interface of the localization AI image processor 3, the parameters of the IP core of the PCIE interface are set;

[0080] In the process of software and hardware adaptation of the domestic industrial intelligent edge server, based on the management requirements of the system device file of the domestic industrial intelligent edge server, a unified operation function interface is provided for all system device files, and a driver of the domestic AI image processor 3 is written. According to the specific functions required by the domestic AI image processor 3 (such as closing and opening interfaces, interface reading and writing), the corresponding function pointers of the data structure in the system function interface are called, and the corresponding function logic structure and sequence are written, so that the operating system of the domestic industrial intelligent edge server can call the corresponding I / O interface.

[0081] Specifically, the interface of the domestic AI image processor 3 can generally be a PCIE interface. Based on the data packet, serial and point-to-point high-performance characteristics of the PCIE interface, the parameters of the IP core of the interface (basic parameter settings such as LaneWidth, LinkSpeed and InterfaceFrequency, base address register settings, etc.) are set. As an example, the PCIE interface IP core contains all the functions required by the processing layer, link layer and physical layer, and most of the optional functions. Among them, a full-featured IP core module can be generated through simple parameter setting in IPCompiler, and Avalon-ST interface or AvaIon-MM interface adapter can be used to map the application layer to the TLP of the processing layer of the IP core. The Avalon-ST adaptation layer maps the Avalon-ST interface of the application layer to the TLPs of the processing layer of the IP core.

[0082] S122, according to the DMA transmission mode, a design scheme of connecting the PCIE bus of the domestic AI image processor 3 and the domestic central processor 4 is generated;

[0083] Specifically, Direct Memory Access (DMA) is a data transmission mode that almost does not require the intervention of the domestic central processor 4 during transmission. The domestic AI image processor 3 can realize direct communication with the domestic central processor 4 through the PCIE bus, and the domestic AI image processor 3 can also communicate with the domestic central processor 4 through the interface with AXI protocol.

[0084] The PCIE bus between the domestic AI image processor 3 and the domestic central processor 4 is designed based on the DMA transmission mode. As an example, the designed PCIE bus (such as PCI9054) integrates two independent DMA channels, each of which supports Block DMA and Scatter / Gather DMA. Channel 0 also supports the Demand DMA transmission mode. Since the two DMA channels of the PCIE bus are composed of a DMA controller and a dedicated bidirectional FIFO, when each channel performs DMA transmission, it is the master device for the PCIE bus and the local bus. That is, the DMA controller will initiate operations on the local bus and the PCIE bus. If it is required to transmit data from the local space to the PCI space, the PCIE bus first performs a read operation on the local bus. There can be other devices accessing the local bus on the local bus, so the PCIE bus needs to apply for access to the local bus. When the PCIE bus obtains the access right, the local data is read into the FIFO of the PCIE bus. At the same time, the PCIE bus applies for PCIE bus access to the PCI bus arbitrator, and writes data from the FIFO to the PCIE bus space. Once the DMA transmission is completed, the PCIE bus sets the DMA "end of transmission bit" to end the transmission. If the interrupt is enabled, an interrupt will be output to the PCIE bus or the local according to the setting, and data can also be transmitted from the PCIE bus to the local bus space.

[0085] S123, according to the operation function interface of the device file of the domestic industrial intelligent edge server, link the link function function of the I / O controller of the domestic AI image processor 3.

[0086] Specifically, based on the operation function interface of the device file of the domestic industrial intelligent edge server, the link function function of the I / O controller of the domestic AI image processor 3 is linked to load the software function of the domestic AI image processor 3. Finally, the functions of the control logic of the domestic AI image processor 3 are debugged, and the reliability of the operation and communication of the domestic AI image processor 3 is evaluated, and the software and hardware adaptation of the domestic central processor 4 and the domestic AI image processor 3 is completed.

[0087] S13, configuring the industrial application software and the industrial protocol of the domestic industrial intelligent edge server adapted to the hardware adaptation scheme;

[0088] Specifically, the step S13 comprises:

[0089] S130, configuring the data acquisition protocol of the domestic industrial intelligent edge server adapted to the hardware adaptation scheme;

[0090] Specifically, the efficiency of industrial data collection in smart factories needs to be improved, and the data collection protocol of the localized industrial intelligent edge server needs to be configured to adapt to the hardware adaptation scheme. To meet the needs of industrial data collection, first, the localized industrial intelligent edge server is physically connected to each industrial device (CNC, robot, PLC, etc.). The localized industrial intelligent edge server is equipped with Ethernet, serial port, CAN port, IO port, etc. Each industrial device can generate multiple collection nodes of industrial data (including image data and non-image data), and the localized industrial intelligent edge server can complete direct connection with industrial devices through corresponding hardware interfaces. To connect more industrial devices, the localized industrial intelligent edge server is connected to one or more industrial gateway devices, and each gateway device is used to connect to multiple industrial devices.

[0091] The localized industrial intelligent edge server must comply with certain industrial core protocols when communicating with industrial devices. From the interface aspect, the localized industrial intelligent edge server is designed to support Ethernet, serial port, CAN port, IO port, etc. and RS485 / RS232 / RS422 electrical interfaces to adapt the physical layer of the communication protocol from the hardware. In order to realize remote transmission in wireless communication mode (NB-IOT, 4G, 5G, WIFI), the localized industrial intelligent edge server is equipped with a wireless module that matches the wireless communication protocol to complete the wireless physical adaptation. In addition, based on the software SDK of multiple industrial devices, the device driver is connected to the localized industrial intelligent edge server. And using the IDE (Integrated Development Environment) software, the industrial protocol API adapted by the industrial device is written into the industrial application software of the localized industrial intelligent edge server. The localized industrial intelligent edge server collects and converts the industrial data generated by the connected industrial devices through the industrial application software, and realizes the configuration of the industrial protocol (such as Modbus, Ethernet / lp, ProfiNet, Powerlink, Ethercat, etc.) of the localized industrial intelligent edge server to support the access of most industrial devices, thereby realizing the matching of the standardized industrial protocol / interface of the localized industrial intelligent edge server and the industrial device.

[0092] As an example, the localized industrial intelligent edge server uses a multi-core (>4 core) multi-threaded localized central processor 4 hardware, which can realize parallel execution of industrial program multi-threading through reasonable allocation of the operating system configured in the subsequent step, improve the task processing efficiency of the system, and have the ability of high efficiency and high reliability of multi-channel industrial I / 0 data processing.

[0093] Specifically, in the application scenario of data acquisition of multi-channel industrial equipment, when the gateway port connected by the domestic industrial intelligent edge server receives industrial data of the same communication protocol, the domestic industrial intelligent edge server controls the industrial gateway to receive data of the corresponding equipment node according to the node number, IP address and gateway port number of the collected equipment, so as to realize multi-channel signal acquisition in industrial acquisition.

[0094] S131, the domestic industrial intelligent edge server is configured with a Haproxy service, and a start script is created to form an IP load balancing mechanism of the domestic industrial intelligent edge server;

[0095] Specifically, if there are a large number of industrial equipment connected to the domestic industrial intelligent edge server at the same time, the domestic industrial intelligent edge server has a particularly large amount of data to process in a short time, which poses a great challenge to the data processing speed and stability of the domestic industrial intelligent edge server. In the face of the data transmission and data processing requirements of a large number of industrial equipment and the domestic industrial intelligent edge server, a multi-domestic industrial intelligent edge server running mechanism can be designed. However, the running of multiple domestic industrial intelligent edge servers will bring the problem of allocation of information resources of multiple industrial equipment, so the load (industrial equipment information resources) can be balanced and distributed to multiple domestic industrial intelligent edge servers for running, thereby realizing load balancing of multiple domestic industrial intelligent edge servers.

[0096] Among them, according to the network topology of the domestic industrial intelligent edge server and the industrial equipment and the multi-IP address allocation form, a specific TCP / IP technology can be used to realize the load balancing of the domestic industrial intelligent edge server. Specifically, haproxy is used as a proxy domestic industrial intelligent edge server, and a minimum connection number algorithm is applied to configure the Haproxy service, and finally a start script is created to realize software-based IP load balancing.

[0097] S132, configure the image data processing algorithm of the domestic industrial intelligent edge server adapted to the hardware adaptation scheme.

[0098] The image data processing algorithm is used to detect appearance defects of products in a factory, such as scratches, indentations, dents, and corrugations. The accuracy of appearance defect detection directly affects the final quality of the products. Traditional manual detection cannot meet the requirements of detection accuracy and efficiency. The embodiment configures the image data processing algorithm on the domestic industrial intelligent edge server, which can accurately and quickly detect the appearance defects of products in real time and accurately capture various surface defects, thereby improving the accuracy of defect detection.

[0099] Specifically, the image data processing algorithm of the domestic industrial intelligent edge server is configured according to the image data processing requirements of the smart factory (such as product appearance defect detection types) and the hardware adaptation scheme. For example, the image data processing algorithm can be R-CNN algorithm, SSD algorithm, and YOLO algorithm for target detection, and AlexNet algorithm, VGG algorithm, and GoogLeNet algorithm for target classification.

[0100] It can be understood that the correspondence among the image data processing requirements, the hardware adaptation scheme, and the image data processing algorithm can be set in advance by the adaptation personnel of the domestic industrial intelligent edge server and in the adaptation device of the domestic industrial intelligent edge server. During the adaptation of the domestic industrial intelligent edge server, the adaptation device of the domestic industrial intelligent edge server can configure the image data processing algorithm of the domestic industrial intelligent edge server according to the hardware adaptation scheme and the image data processing requirements. For example, analyzing product appearance defect types (scratches, indentations, dents, and corrugations) can be attributed to the classification problem of Al images, and AlexNet algorithm, VGG algorithm, or GoogLeNet algorithm can be selected.

[0101] Among them, as an example, the image data processing algorithm provided by the embodiment can be trained and optimized in the following way: an industrial CCD camera takes multiple product surface images and sends them to a domestic industrial intelligent edge server, and the domestic AI image processor 3 of the domestic industrial intelligent edge server extracts different defect features from the images using the image data processing algorithm. Specifically, the defect features can be extracted by edge detection algorithm (Sobel, Canny or Laplacian edge detection operator), so as to obtain the defect features of multiple product surface images. Then all the images are divided into two groups according to the ratio of 6:4, the defect features in the first group of images are the training data group, which is used to train the group image data processing algorithm, and the defect features in the second group of images are the test data group, which is used to evaluate the image data processing algorithm later. When evaluating the image data processing algorithm in a specific industrial scene, the test data set is used for testing, and relevant technical evaluation indexes (such as average error or average variance) are set to evaluate the pros and cons of each AI algorithm before and after optimization. In order to ensure the stability of the detection environment, the light intensity of the product detection environment should be consistent as much as possible, and the positions of the light source, industrial CCD camera and product must be fixed, and the color and intensity of the light source must be consistent.

[0102] In addition, based on the evaluation results of the image data processing algorithm, the image processing algorithm is improved, and the improvement measures include selecting more stable and more training samples, optimizing network parameters and optimizing network structure. Among them, regarding the optimization of network parameters, according to a certain defect type with low accuracy, the discrimination error of the (cross-entropy) loss function is adjusted to change the analysis standard and weight of the image data processing algorithm for defect classification. Regarding the optimization of network structure, according to the characteristics of the network multi-layer calculation of the image processing algorithm, the network convolution layer that loses more characteristic information is analyzed according to the characteristics of the defect type with low accuracy, and the network structure is optimized, such as replacing one s convolution layer with n k convolution layers (s < k).

[0103] In addition, the image processor applied in the traditional AI image data processing is usually a serial instruction set architecture, which needs to wait for the calculation processing result of the previous one before executing the next instruction, so the chip utilization rate is low, causing the disadvantage of delay. In order to make the domestic AI image processor 3 more effectively execute the image data processing algorithm, the domestic AI image processor 3 of the embodiment adopts data flow architecture, so that the calculation flow and data flow run in parallel, that is, the data is accessed while the data is calculated, reducing the idle time of the calculation unit of the domestic AI image processor 3 and improving the utilization rate of the domestic AI image processor 3. In addition, the calculation architecture of the high-performance domestic AI image processor 3 can be set based on the deep learning neural network, and the domestic AI image processor 3 controls the calculation order through the flow order of the data flow, eliminating the additional time overhead caused by instruction operation.

[0104] S14, configuring system software of the localized industrial intelligent edge server adapted to the hardware adaptation scheme, the industrial application software, and the industrial protocol.

[0105] Specifically, after determining the localized industrial intelligent edge server adapted to the hardware adaptation scheme, the industrial application software, and the industrial protocol, the corresponding system software of the localized industrial intelligent edge server can be configured according to the correspondence between the three and the system software. The correspondence between the three and the system software can be set by adaptation personnel of the localized industrial intelligent edge server according to experience in advance. The system software includes an operating system of the localized industrial intelligent edge server, and the operating system is a process operating system, which can be a Kylin operating system, a UC operating system, a Red Flag Linux operating system, or a Huawei Euler operating system.

[0106] In the process of configuring the system software, an application development framework and environment of the operating system can be first constructed. Specifically, a visual development environment of the application software compatible with mainstream languages C++, Java, and C# is configured for the operating system, for example, a domestic SDK software development kit is provided, and the following software tool kits are also provided: an interface interaction development tool kit, a plug-in development framework kit, a distributed development framework kit, a package management development tool kit, a kernel development and debugging tool kit, and a visual monitoring tool graphical tool, so that the constructed localized industrial intelligent edge server has the abilities of code writing, analysis, compilation, and debugging. In addition, based on the tool functions of compilation and debugging, and considering the characteristics of the operating system (for example, a domestic Linux operating system), the core editor design is completed. The editor is the main interface directly faced by programming, and the interface display form (text, semi-visualization, visualization) is determined to realize basic editing functions (paste, undo, delete, save, error prompt, fold, and format). In addition, the visual development environment of the operating system can design the functions and arrangement of the main interface according to the ease of use of programmers, such as the main graph, navigation bar, tool bar, state bar color, and structure arrangement. To enhance the ease of use of the application development of the operating system, the operating system can be adapted to mainstream IDE development environments (Eclipse, Android Studio, PyCharm, Qt, Vim, etc.), so that the source code of the application software on the original IDE can run directly on the system, which greatly improves the rapid migration ability of the application.

[0107] In the embodiment of the present application, by taking the weight score result of the localization central processor sample and the localization AI image processor sample that meet the preset score condition as the localization central processor and the localization AI image processor of the localization industrial intelligent edge server, the hardware performance and compatibility of the localization central processor and the localization AI image processor of the localization industrial intelligent edge server are ensured to be good; then, according to the preset hardware adaptation strategy, the hardware adaptation scheme of the localization AI image processor and the localization central processor is generated; then, the industrial application software and the industrial protocol of the localization industrial intelligent edge server that are adapted to the hardware adaptation scheme are configured; finally, the system software of the localization industrial intelligent edge server that is adapted to the hardware adaptation scheme, the industrial application software and the industrial protocol is configured; in this way, the system architecture of the edge server can be automatically adapted. In addition, the localization edge server not only has security in data processing, but also the localization industrial intelligent edge server is specially used for image data processing through the localization AI image processor, and the localization central processor is specially used for non-image data processing, so that the localization industrial intelligent edge server can meet the processing demand of massive data.

[0108] In the above-mentioned embodiment of the present application, further, after the step S14, the method further comprises:

[0109] S15, the performance stability test and the data processing pressure test are performed on the localization industrial intelligent edge server after adaptation and in the preset test scene.

[0110] Specifically, the performance stability test comprises whether the performance index of the localization industrial intelligent edge server in the preset test scene is abnormal, whether the localization industrial intelligent edge server in the preset test scene is down or application is suspended; the performance index comprises the highest working frequency of the localization central processor 4, the utilization rate of the localization central processor 4, the temperature of the localization central processor 4, the power consumption of the localization central processor 4, the peak computing power of the localization AI image processor 3, the utilization rate of the localization AI image processor 3, the temperature of the localization AI image processor 3, and the power consumption of the localization AI image processor 3.

[0111] In addition, the data processing pressure test comprises whether the data processing index of the localization industrial intelligent edge server in the preset test scene is abnormal; the data processing index comprises the data transmission speed, the stability of the data transmission speed, and the data processing response time.

[0112] In this embodiment, specifically, when the soft and hard fully cooperate, the typical scene of the intelligent factory is tested and evaluated. The test and evaluation content mainly includes performance stability test and data processing stress test.

[0113] (1) Performance stability test

[0114] According to the test scene of the intelligent factory, the domestic industrial intelligent edge server is connected to multiple devices (CNC, mechanical arm) through a gateway, and the stability of industrial applications such as data acquisition and AI image processing is mainly tested. In the industrial application test, the performance indicators of the system are mainly considered, including the maximum frequency, utilization rate, temperature, and power consumption of the domestic central processor 4, the peak computing power, utilization rate, temperature, and power consumption of the domestic AI image processor 3, whether these parameter values change abnormally, and whether they can remain stable. In addition, it is necessary to analyze whether abnormal situations such as system downtime and application suspension occur during the test period.

[0115] (2) Data processing stress test

[0116] According to the test scene of the intelligent factory, the domestic industrial intelligent edge server is connected to multiple devices (CNC, mechanical arm) through a gateway, and the stability of industrial applications such as data acquisition and AI image processing is mainly tested. In the industrial application test, the performance indicators of the system are mainly considered, including the maximum frequency, utilization rate, temperature, and power consumption of the domestic central processor 4, the peak computing power, utilization rate, temperature, and power consumption of the domestic AI image processor 3, whether these parameter values change abnormally, and whether they can remain stable. In addition, it is necessary to analyze whether abnormal situations such as system downtime and application suspension occur during the test period.

[0117] Reference Figure 2, is a structural schematic diagram of a localized industrial intelligent edge server provided by an embodiment of the present application, the localized industrial intelligent edge server is adapted by any of the adaptation method embodiments of the localized industrial intelligent edge server described above, and the localized industrial intelligent edge server comprises a mainboard 1, a data acquisition module 2, a localized AI image processor 3, and a localized central processor 4. The data acquisition module 2 is arranged on the mainboard 1 and is used to acquire external data sent by an external device. The external data comprises image data and non-image data. The localized AI image processor 3 is arranged on the mainboard 1 and is connected with the data acquisition module 2, and is used to perform data processing on the image data. The localized central processor 4 is arranged on the mainboard 1 and is connected with the localized AI image processor 3 and the data acquisition module 2, and is used to perform data processing on the non-image data, and perform corresponding operations according to the data processing result of the non-image data and the data processing result of the image data.

[0118] In the present application, the system architecture of the localized industrial intelligent edge server is adapted, and the localized AI image processor 3 is specially used for processing image data, so that the localized central processor 4 can be used for processing non-image data, thereby enabling the adapted localized industrial intelligent edge server to meet the processing requirements of massive data. The localized industrial intelligent edge server with the adapted system architecture can meet the processing requirements of massive data by using the localized AI image processor 3 for processing image data and using the localized central processor 4 for processing non-image data.

[0119] In the present application, the system architecture of the localized industrial intelligent edge server is adapted, and the localized AI image processor 3 is specially used for processing image data, so that the localized central processor 4 can be used for processing non-image data, thereby enabling the adapted localized industrial intelligent edge server to meet the processing requirements of massive data. The localized industrial intelligent edge server with the adapted system architecture can meet the processing requirements of massive data by using the localized AI image processor 3 for processing image data and using the localized central processor 4 for processing non-image data.

[0120] Specifically, the mainboard 1 is provided with a heat dissipation device (not shown in the figure), the heat dissipation device includes a heat dissipation fan (not shown in the figure), a heat dissipation cover (not shown in the figure) and a cooling liquid pipe (not shown in the figure); the heat dissipation cover is close to the outer surface of the localized central processing unit 4, the cooling liquid pipe is arranged on the outer surface of the heat dissipation cover, and the air outlet side of the heat dissipation fan is aligned with the outer surface of the heat dissipation cover. The heat dissipation cover provided with the cooling liquid pipe can quickly dissipate the working heat of the localized central processing unit 4, and the heat dissipation fan can further quickly exchange the heat on the surface of the heat dissipation cover with air in a convection manner, so that the localized central processing unit 4 can be effectively cooled.

[0121] Specifically, referring to Figure 3 , the mainboard 1 is provided with a power module 5 and a power management chip 6; the input end of the power management chip 6 is connected with the power module 5, and the output end of the battery management chip is connected with the data acquisition module 2, the localized AI image processor 3 and the localized central processing unit 4; the power management chip 6 is used for monitoring the power supply state of the power module 5, and controlling the power supply to the data acquisition module 2, the localized AI image processor 3 and the localized central processing unit 4 according to the power supply state of the power module 5. Wherein, the power management chip 6 can be LMG3410R050, UCC12050, BQ25790, HIP6301, IS6537, RT9237, ADP3168, KA7500, TL494, etc.

[0122] Referring to Figure 4 , it is a structural schematic diagram of an adaptive device of a localized industrial intelligent edge server provided by an embodiment of the application, the adaptive device of the localized industrial intelligent edge server comprises:

[0123] A weight scoring module 10 is used for performing weight scoring on the hardware performance indicators of each group of localized hardware samples in a localized hardware sample group to be evaluated; the localized hardware sample group comprises: multiple groups of samples of localized central processing units 4 and localized AI image processors 3;

[0124] A hardware selection module 11 is used for selecting the samples of the localized central processing unit 4 and the samples of the localized AI image processor 3, which meet the preset scoring conditions, as the localized central processing unit 4 and the localized AI image processor 3 of the localized industrial intelligent edge server;

[0125] A hardware adaptation module 12 is used for generating a hardware adaptation scheme of the localized AI image processor 3 and the localized central processing unit 4 according to a preset hardware adaptation strategy;

[0126] a software protocol configuration module 13, configured to configure industrial application software and industrial protocols of the localized industrial intelligent edge server, which are adapted to the hardware adaptation scheme;

[0127] a system configuration module 14, configured to configure system software of the localized industrial intelligent edge server, which is adapted to the hardware adaptation scheme, the industrial application software and the industrial protocols.

[0128] In the embodiment of the present application, the localized central processing unit sample and the localized AI image processor sample whose weight score results meet the preset score condition are taken as the localized central processing unit and the localized AI image processor of the localized industrial intelligent edge server, so as to ensure that the hardware performance and compatibility of the localized central processing unit and the localized AI image processor of the localized industrial intelligent edge server are both good. Then, the hardware adaptation scheme of the localized AI image processor and the localized central processing unit is generated according to the preset hardware adaptation strategy. Then, the industrial application software and the industrial protocols of the localized industrial intelligent edge server, which are adapted to the hardware adaptation scheme, are configured. Finally, the system software of the localized industrial intelligent edge server, which is adapted to the hardware adaptation scheme, the industrial application software and the industrial protocols, is configured. In this way, the system architecture components of the edge server can be automatically localized and adapted. In addition, the localized edge server not only has security in data processing, but also the localized industrial intelligent edge server is specially used for image data processing through the localized AI image processor, and the localized central processing unit is specially used for non-image data processing. In this way, the localized industrial intelligent edge server can meet the processing demand of massive data.

[0129] For more functions of the function modules of the adaptation device of the localized industrial intelligent edge server in the embodiment, refer to the description of the adaptation device method of the localized industrial intelligent edge server in the embodiments of the present application, which will not be repeated here.

[0130] Referring to Figure 5 is a schematic diagram of the adaptation device of the localized industrial intelligent edge server provided by an embodiment of the present application. The adaptation device of the localized industrial intelligent edge server in the embodiment comprises a processor 100, a memory 101, and a computer program, such as an adaptation program of the localized industrial intelligent edge server, stored in the memory 101 and executable on the processor 100. The processor 100 implements the steps in each of the above-mentioned adaptation methods of the localized industrial intelligent edge server when executing the computer program. Alternatively, the processor 100 implements the functions of each module / unit in each of the above-mentioned device embodiments when executing the computer program.

[0131] For example, the computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present application. The one or more modules / units can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program in the adaptation device of the domestic industrial intelligent edge server.

[0132] The adaptation device of the domestic industrial intelligent edge server can be a desktop computer, a notebook computer, a palm computer, a cloud domestic industrial intelligent edge server and the like. The adaptation device of the domestic industrial intelligent edge server can include, but is not limited to, a processor and a memory. Those skilled in the art can understand that the schematic diagram is only an example of the adaptation device of the domestic industrial intelligent edge server, and does not constitute a limitation on the adaptation device of the domestic industrial intelligent edge server, and can include more or fewer components than the diagram, or combine certain components, or different components, for example, the adaptation device of the domestic industrial intelligent edge server can also include an input / output device, a network access device, a bus and the like.

[0133] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor and the like. The processor is the control center of the adaptation device of the domestic industrial intelligent edge server, and connects all parts of the adaptation device of the domestic industrial intelligent edge server through various interfaces and lines.

[0134] The memory can be used to store the computer program and / or modules, and the processor realizes various functions of the adaptation device of the localized industrial intelligent edge server by running or executing the computer program and / or modules stored in the memory, and calling data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application program required by a function (such as a sound playing function, an image playing function, etc.), and the like; and the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), and the like. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, for example, a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state memory devices.

[0135] If the modules / units integrated by the adaptation device of the localized industrial intelligent edge server are realized in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer-readable storage medium. The computer program can implement the steps of each method embodiment when executed by a processor. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or some intermediate forms, etc. The computer-readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the contents included in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0136] It should be noted that the apparatus embodiments described above are merely illustrative, and the units described as separate units can or can not be physically separate, and the units shown as units can or can not be physical units, i.e., can be located in one place, or can be distributed to multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiment according to actual needs. In addition, the connection relationship between the modules in the apparatus embodiment provided by the present application indicates that there is a communication connection between them, which can be implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement it without creative labor.

[0137] The above is the preferred embodiment of the present application. It should be noted that those skilled in the art can make several improvements and refinements without departing from the principles of the present application, and these improvements and refinements are also considered within the scope of protection of the present application.

Claims

1. A method for adapting a localized industrial intelligent edge server, characterized in that, The method comprises the following steps: weight scoring of hardware performance indicators for each set of localized hardware samples in a set of localized hardware samples to be evaluated; the set of localized hardware samples comprises multiple sets of samples of localized central processing units and localized AI image processors; localized central processing unit samples and localized AI image processor samples that meet preset scoring conditions in the weight scoring results are selected as localized central processing units and localized AI image processors of the localized industrial intelligent edge server; a hardware adaptation scheme for the localized AI image processor and the localized central processing unit is generated according to a preset hardware adaptation strategy; the hardware adaptation scheme for the localized AI image processor and the localized central processing unit is generated according to the preset hardware adaptation strategy, which comprises the following steps: a mainboard packaging scheme for the localized central processing unit is generated according to the packaging mode, the number of packaging pins and the model of the north bridge chip of the localized central processing unit; parameters of an IP core of a PCIE interface are set according to the performance characteristics of the PCIE interface of the localized AI image processor; a design scheme of a PCIE bus connecting the localized AI image processor and the localized central processing unit is generated according to a DMA transmission mode; a linking function function of an I / O controller of the localized AI image processor is linked according to an operation function interface of a device file of the localized industrial intelligent edge server; industrial application software and industrial protocols of the localized industrial intelligent edge server that are adapted to the hardware adaptation scheme are configured; system software of the localized industrial intelligent edge server that is adapted to the hardware adaptation scheme, the industrial application software and the industrial protocols is configured. 2.The adaptation method of a localized industrial intelligent edge server of claim 1, wherein, After the system software of the localized industrial intelligent edge server that is adapted to the hardware adaptation scheme, the industrial application software and the industrial protocols is configured, the method further comprises the following steps: performance stability testing and data processing pressure testing are performed on the localized industrial intelligent edge server in a preset test scenario.

3. The adaptation method of the localized industrial intelligent edge server according to claim 2, wherein the performance stability testing comprises whether performance indicators of the localized industrial intelligent edge server in the preset test scenario are abnormal, whether the localized industrial intelligent edge server in the preset test scenario is down or application is suspended, and the performance indicators comprise the maximum working frequency of the localized central processing unit, the utilization rate of the localized central processing unit, the temperature of the localized central processing unit, the power consumption of the localized central processing unit, the peak computing power of the localized AI image processor, the utilization rate of the localized AI image processor, the temperature of the localized AI image processor and the power consumption of the localized AI image processor; the data processing pressure testing comprises whether data processing indicators of the localized industrial intelligent edge server in the preset test scenario are abnormal, and the data processing indicators comprise data transmission speed, stability of the data transmission speed and data processing response time. 4.The adaptation method of a localized industrial intelligent edge server of claim 1, wherein, The hardware performance indicators include performance indicators of the localized central processor, performance indicators of the localized AI image processor, and compatibility performance indicators of both the localized central processor and the localized AI image processor. The scoring conditions include that a total score of the hardware performance indicators is greater than a preset score threshold or the total score is maximum. 5.The adaptation method of a localized industrial intelligent edge server of claim 1, wherein, The weight scoring of each set of localized hardware samples in the set of localized hardware samples to be evaluated includes: The weight scoring of each set of localized hardware samples in the set of localized hardware samples to be evaluated includes: The total score of each hardware performance indicator is calculated according to the weight scoring coefficient of each hardware performance indicator of each set of localized hardware samples. 6.The adaptation method of a localized industrial intelligent edge server of claim 1, wherein, The industrial application software and industrial protocol of the localized industrial intelligent edge server adapted to the hardware adaptation scheme include: The data acquisition protocol of the localized industrial intelligent edge server adapted to the hardware adaptation scheme is configured. The Haproxy configuration service is provided for the localized industrial intelligent edge server, and a start script is created to form an IP load balancing mechanism of the localized industrial intelligent edge server. The image data processing algorithm of the localized industrial intelligent edge server adapted to the hardware adaptation scheme is configured.

7. A localized industrial intelligent edge server, characterized in that, The localized industrial intelligent edge server adapted by the adaptation method of the localized industrial intelligent edge server of any one of claims 1-6 includes: a mainboard; a data acquisition module disposed on the mainboard, configured to acquire external data transmitted by an external device, wherein the external data includes image data and non-image data; a localized AI image processor disposed on the mainboard and connected with the data acquisition module, configured to perform data processing on the image data; and a localized central processor disposed on the mainboard and connected with the localized AI image processor and the data acquisition module, configured to perform data processing on the non-image data, and perform corresponding operations according to the data processing result of the non-image data and the data processing result of the image data.

8. The localized industrial intelligent edge server of claim 7, wherein, The mainboard is provided with a heat dissipation device, which includes a heat dissipation fan, a heat dissipation cover and a cooling liquid pipe; the heat dissipation cover is in close contact with the outer surface of the localized central processor, the cooling liquid pipe is disposed on the outer surface of the heat dissipation cover, and the air outlet side of the heat dissipation fan is aligned with the outer surface of the heat dissipation cover.

9. The localized industrial intelligent edge server of claim 7, wherein, The mainboard is provided with a power module and a power management chip; the input end of the power management chip is connected with the power module, the output end of the power management chip is connected with the data acquisition module, the localized AI image processor and the localized central processor, and the power management chip is configured to monitor the power supply state of the power module and control the power supply to the data acquisition module, the localized AI image processor and the localized central processor according to the power supply state of the power module.

10. An adaptation device for a local industrial intelligent edge server, characterized in that, The weight scoring module is configured to score the weight of each set of localized hardware samples in the set of localized hardware samples to be evaluated. ​ The domestic hardware sample set comprises: a plurality of domestic central processor and domestic AI image processor samples; The hardware selection module is configured to select the domestic central processor sample and the domestic AI image processor sample whose weight score result satisfies a preset score condition as the domestic central processor and the domestic AI image processor of the domestic industrial intelligent edge server. The hardware adaptation module is configured to generate a hardware adaptation scheme of the domestic AI image processor and the domestic central processor according to a preset hardware adaptation strategy. The hardware adaptation scheme of the domestic AI image processor and the domestic central processor according to the preset hardware adaptation strategy comprises: According to the packaging mode, the number of pins and the model of the north bridge chip of the domestic central processor, a mainboard packaging scheme of the domestic central processor is generated. According to the performance characteristics of the PCIE interface of the domestic AI image processor, the parameters of the IP core of the PCIE interface are set. According to the DMA transmission mode, a design scheme of a PCIE bus connecting the domestic AI image processor and the domestic central processor is generated. According to the operation function interface of the device file of the domestic industrial intelligent edge server, a linkage function function of an I / O controller of the domestic AI image processor is linked. The software protocol configuration module is configured to configure the industrial application software and the industrial protocol of the domestic industrial intelligent edge server which are adapted to the hardware adaptation scheme. The system configuration module is configured to configure the system software of the domestic industrial intelligent edge server which is adapted to the hardware adaptation scheme, the industrial application software and the industrial protocol.

11. An adaptation device for a local industrial intelligent edge server, characterized in that, The computer readable storage medium comprises a stored computer program, wherein when the computer program is running, the device where the computer readable storage medium is located is controlled to execute the adaptation method of the domestic industrial intelligent edge server according to any one of claims 1 to 6.

12. A computer-readable storage medium, characterized in that, The computer readable storage medium comprises a stored computer program, wherein when the computer program is running, the device where the computer readable storage medium is located is controlled to execute the adaptation method of the domestic industrial intelligent edge server according to any one of claims 1 to 6.

Images