CN116088842A

CN116088842A - AI algorithm engine system and method applied to Internet of things platform

Info

Publication number: CN116088842A
Application number: CN202310037705.XA
Authority: CN
Inventors: 戴吉平; 李信洪; 袁宜峰
Original assignee: Shenzhen Das Intellitech Co Ltd
Current assignee: Shenzhen Das Intellitech Co Ltd
Priority date: 2023-01-10
Filing date: 2023-01-10
Publication date: 2023-05-09

Abstract

The invention relates to an AI algorithm engine system and method applied to an internet of things platform, comprising the following steps: the algorithm center module and the algorithm application module; the algorithm center module outputs an unexecutable algorithm template management page and an algorithm template definition page according to an algorithm template operation instruction input by a user; the algorithm application module outputs an executable algorithm management page and an algorithm definition page according to an algorithm operation instruction input by a user. All modules of the AI algorithm engine related to the invention can be developed and uniformly managed on the platform of the Internet of things, a preset algorithm template is constructed through the algorithm center module, and then an executable algorithm can be constructed by data binding through the algorithm template of the algorithm center module referenced by the algorithm application module, so that the intervention of developers is not needed, the production efficiency of the algorithm is greatly improved, the automatic production capacity of the algorithm is realized, and the manpower and material resource investment of customized development is reduced.

Description

AI algorithm engine system and method applied to Internet of things platform

Technical Field

The invention relates to the technical field of the Internet of things, in particular to an AI algorithm engine system and method applied to an Internet of things platform.

Background

The traditional internet of things platform is connected with objects, and achieves an integrated platform integrating the capabilities of equipment management, data safety communication, message subscription and the like. Downward supporting connection of mass equipment, and collecting equipment data to cloud; the cloud end API is provided upwards, and the server side can send the instruction to the equipment side by calling the cloud end API, so that remote control is realized. However, in a big data background, it is critical to integrate an AI (artificial intelligence) technology and an IoT (internet of things) technology to implement an intelligent internet of things platform. The AI of the internet of things platform is a necessary trend, the data value is continuously mined, the knowledge service is provided, and the AI capability is used for realizing the closed loop of perception, cognition, decision making and action of people and things.

The existing internet of things platform mainly has the following key problems that the algorithm capability is not provided, or the algorithm capability is provided, but an algorithm system is independent of the internet of things platform, the algorithm production needs customized development, the realization cost is high, and the deployment implementation efficiency is low.

Disclosure of Invention

Aiming at the defects of the prior art, the technical problem to be solved by the invention is to provide an AI algorithm engine system and an AI algorithm engine method applied to an Internet of things platform.

The technical scheme adopted for solving the technical problems is as follows: an AI algorithm engine system for an Internet of things platform is constructed, comprising: the algorithm center module and the algorithm application module;

the algorithm center module is used for outputting an executable algorithm template management page and an algorithm template definition page according to an algorithm template operation instruction input by a user; the algorithm template management page is used for displaying the non-executable algorithm template and the template information of the non-executable algorithm template and providing a man-machine interaction function for a user; the algorithm template definition page is used for a user to execute the definition configuration operation on the non-executable algorithm template;

the algorithm application module is used for outputting an executable algorithm management page and an algorithm definition page according to an algorithm operation instruction input by a user; the algorithm management page is used for displaying executable algorithms and algorithm information of the executable algorithms and providing a man-machine interaction function for a user; the algorithm definition page is used for a user to execute the definition configuration operation on the executable algorithm.

The AI algorithm engine system applied to the platform of the Internet of things, provided by the invention, further comprises: a function library module;

the function library module is used for outputting a function management page and a function definition page of the python function according to a function operation instruction input by a user;

the function management page displays the python function and provides a man-machine interaction function for a user; the function definition page is used for a user to define and configure the python function.

In the AI algorithm engine system applied to the internet of things platform of the present invention, the function library module includes:

the function management unit is used for storing at least one python function and outputting the function management page based on a function operation instruction input by a user;

and the function definition unit is used for providing the function definition page for a user and executing definition and configuration of the python function according to a function definition configuration operation input by the user.

In the AI algorithm engine system for an internet of things platform of the present invention, the python function includes: basic information of functions and scripts of the functions;

the basic information of the function includes: function name, function code, function description, updater, and update time;

the script of the function includes: python edits the script.

In the AI algorithm engine system applied to the internet of things platform of the present invention, the algorithm center module includes:

the algorithm template management unit is used for storing at least one non-executable algorithm template and outputting the algorithm template management page based on an algorithm template operation instruction input by a user;

and the algorithm template definition unit is used for providing the algorithm template definition page for a user and executing the definition and configuration of the non-executable algorithm template according to the algorithm template definition configuration operation instruction input by the user.

In the AI algorithm engine system applied to the internet of things platform of the present invention, the template information of the non-executable algorithm template includes: basic information of an algorithm template and definition information of the algorithm template;

the basic information of the algorithm template comprises: the name of the algorithm template, the coding of the algorithm template, the starting state of the algorithm template, the description of the algorithm template, the updater of the algorithm template and the updating time of the algorithm template;

the definition information of the algorithm template comprises: the basic information of the algorithm template, the algorithm script of the algorithm template and the file attachment of the algorithm template;

the algorithm script of the algorithm template comprises: defining an input parameter set, an output parameter set and a python editing script;

the file attachment of the algorithm template comprises: at least one file uploaded by the user.

In the AI algorithm engine system for the platform of the internet of things, the algorithm application module includes:

the algorithm application management unit is used for storing at least one executable algorithm and outputting the algorithm application management page based on an algorithm operation instruction input by a user;

and the algorithm definition unit is used for providing the algorithm definition page for a user and executing definition and configuration of the executable algorithm according to the algorithm definition configuration operation instruction input by the user.

In the AI algorithm engine system applied to the internet of things platform of the present invention, the algorithm information of the executable algorithm includes: basic information of an algorithm and definition information of the algorithm;

the basic information of the algorithm comprises: the method comprises the steps of (1) name of an algorithm, coding of the algorithm, triggering condition of the algorithm, starting state of the algorithm, latest execution duration of the algorithm, description of the algorithm, updater of the algorithm and updating time of the algorithm;

the definition information of the algorithm comprises: basic information of the algorithm, triggering conditions of the algorithm, configuration of the algorithm and file attachments of the algorithm;

the configuration of the algorithm comprises: after the non-executable algorithm template in the algorithm center module is referred, the input parameter set and the output parameter set of the referred non-executable algorithm template are subjected to data binding;

the triggering conditions of the algorithm include: timing calculation, delay calculation and event calculation;

the file attachment of the algorithm comprises: by referencing a file synchronized when the algorithm is not executable in the algorithm center module, a file uploaded by the user, or a file stored when the algorithm is executed.

In the AI algorithm engine system for the platform of the internet of things of the present invention, the algorithm application management unit includes:

and the algorithm execution subunit is used for performing control, suspension control or log operation on the executable algorithm.

The invention also provides an AI algorithm engine method applied to the platform of the Internet of things, which comprises the following steps:

receiving a function operation instruction input by a user, executing function creation according to the function operation instruction, and outputting a function management page and a function definition page of a python function;

receiving an algorithm template operation instruction input by a user, and executing algorithm template creation according to the algorithm template operation instruction so as to output an unexecutable algorithm template management page and an algorithm template definition page;

and receiving an algorithm operation instruction input by a user, executing algorithm creation according to the algorithm operation instruction, and outputting an executable algorithm management page and an algorithm definition page.

The AI algorithm engine system and the AI algorithm engine method applied to the platform of the Internet of things have the following beneficial effects: the algorithm center module outputs an unexecutable algorithm template management page and an algorithm template definition page according to an algorithm template operation instruction input by a user; the algorithm application module outputs an executable algorithm management page and an algorithm definition page according to an algorithm operation instruction input by a user. All modules of the AI algorithm engine related to the invention can be developed and uniformly managed on the platform of the Internet of things, a preset algorithm template is constructed through the algorithm center module, and then an executable algorithm can be constructed by data binding through the algorithm template of the algorithm center module referenced by the algorithm application module, so that the intervention of developers is not needed, the production efficiency of the algorithm is greatly improved, the automatic production capacity of the algorithm is realized, and the manpower and material resource investment of customized development is reduced.

Drawings

The invention will be further described with reference to the accompanying drawings and examples, in which:

FIG. 1 is a schematic block diagram of an AI algorithm engine system applied to an Internet of things platform provided by an embodiment of the invention;

FIG. 2 is a schematic diagram of a function list of a function management page according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of basic information of a function definition page according to an embodiment of the present invention;

FIG. 4 is a schematic diagram of a script of a function definition page provided by an embodiment of the present invention;

FIG. 5 is a schematic diagram of an algorithm template list of an algorithm template management page provided by an embodiment of the present invention;

FIG. 6 is a basic information diagram of an algorithm template definition page according to an embodiment of the present invention;

FIGS. 7 and 8 are schematic diagrams of algorithm scripts of an algorithm template definition page provided by an embodiment of the present invention;

FIG. 9 is a schematic diagram of a file attachment of an algorithm template definition page provided by an embodiment of the present invention;

FIG. 10 is a schematic diagram of an algorithm list of an algorithm application management page provided by an embodiment of the present invention;

FIG. 11 is a schematic diagram of an algorithm execution log in an algorithm application management page according to an embodiment of the present invention;

FIG. 12 is a schematic view of checking the execution result of an algorithm execution log in an algorithm application management page according to an embodiment of the present invention;

FIG. 13 is a basic information diagram of an algorithm definition page according to an embodiment of the present invention;

FIG. 14 is a schematic diagram of trigger conditions for an algorithm of an algorithm definition page provided by an embodiment of the present invention;

FIG. 15 is a schematic diagram of an algorithm configuration of an algorithm definition page provided by an embodiment of the present invention;

FIG. 16 is a schematic illustration of a file attachment of an algorithm definition page provided by an embodiment of the present invention;

FIG. 17 is a schematic diagram of basic information created by an algorithm template according to an embodiment of the present invention;

FIG. 18 is a schematic diagram of creating a parameter set under input parameters and output parameters for an algorithm template provided by an embodiment of the present invention;

FIG. 19 is a schematic diagram of editing a python script in the creation of an algorithm template provided by an embodiment of the present invention;

FIG. 20 is a schematic diagram of uploading file attachments in the creation of an algorithm template according to an embodiment of the present invention;

FIG. 21 is a schematic diagram showing an algorithm management page after a certain algorithm template provided by the embodiment of the present invention is created;

FIG. 22 is a schematic diagram of basic information created by an algorithm provided by an embodiment of the present invention;

FIG. 23 is a schematic diagram of trigger conditions in the creation of an algorithm provided by an embodiment of the present invention;

FIG. 24 is a schematic diagram of an algorithm configuration in an algorithm creation provided by an embodiment of the present invention;

FIG. 25 is a schematic diagram of file attachments for synchronization after referencing an algorithm template in the creation of an algorithm provided by an embodiment of the present invention;

FIG. 26 is a schematic diagram showing an algorithm application management page after a certain algorithm is created according to an embodiment of the present invention;

FIG. 27 is a schematic diagram of an algorithm execution log operation provided by an embodiment of the present invention;

FIG. 28 is a schematic diagram of log view results of an algorithm according to an embodiment of the present invention;

fig. 29 is a flowchart of an AI algorithm engine method applied to an internet of things platform according to an embodiment of the present invention.

Detailed Description

The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

Referring to fig. 1, the invention provides an AI algorithm engine system applied to an internet of things platform, the AI algorithm engine system can be applied to all internet of things platforms, all modules of the AI algorithm engine in the AI algorithm engine system can realize development and unified management on the internet of things platform, a preset algorithm template is built in an AI algorithm center of the AI algorithm engine, algorithm application support reference algorithm binding data construct executable algorithm, an application module does not need to be intervened by a developer, only needs to call equipment instance binding data of the internet of things platform, algorithm production efficiency is greatly improved, algorithm automatic production capacity is also realized, meanwhile, the AI algorithm engine is integrated on the traditional internet of things platform to truly endow the platform algorithm capacity without independent deployment, and manpower and material resource investment of customized development is reduced.

Specifically, as shown in fig. 1, the AI algorithm engine system applied to the internet of things platform includes: an algorithm center module 20 and an algorithm application module 30.

The algorithm center module 20 is used for outputting an executable algorithm template management page and an algorithm template definition page according to an algorithm template operation instruction input by a user; the algorithm template management page is used for displaying the non-executable algorithm template and the template information of the non-executable algorithm template and providing a man-machine interaction function for a user; the algorithm template definition page is used for a user to execute a definition configuration operation on the non-executable algorithm template.

The algorithm application module 30 is configured to output an executable algorithm management page and an algorithm definition page according to an algorithm operation instruction input by a user; the algorithm management page is used for displaying executable algorithms and algorithm information of the executable algorithms and providing a man-machine interaction function for a user; the algorithm definition page is used for a user to execute definition configuration operation on the executable algorithm.

Further, as shown in fig. 1, the AI algorithm engine system applied to the internet of things platform further includes: a function library module 10.

The function library module 10 is used for outputting a function management page and a function definition page of the python function according to a function operation instruction input by a user. The function management page displays the python function and provides a man-machine interaction function for a user; the function definition page is used for a user to define and configure the python function.

It will be appreciated that in the embodiment of the present invention, the functions in the function library module 10 are set to facilitate the algorithm template creation by the algorithm center module 20, and thus, in other embodiments, the function library module 10 may not be set. For example, both the a algorithm template and the B algorithm template need to use the function 1#, so that the function 1# can be created in the function library module 10, and the a algorithm template and the B algorithm template can be directly called without re-writing the script of the function 1#. Because one algorithm template can call less than ten functions and more than ten functions when writing scripts, the functions are common or have specific functions, and therefore the efficiency of creating the algorithm template can be greatly facilitated.

Optionally, in an embodiment of the present invention, the function library module 10 includes: a function management unit 11 and a function definition unit 12.

Wherein the function management unit 11 is configured to store at least one python function and output a function management page based on a function operation instruction input by a user. Specifically, as shown in fig. 2, in the embodiment of the present invention, the function management page output by the function management unit 11 includes all python function lists. The function management page also displays basic information of each python function for a user to view. Alternatively, in embodiments of the present invention, the python function includes, but is not limited to: model functions, data processing functions, error calculation functions, etc.

The function definition unit 12 is used to provide a function definition page to a user, and perform definition and configuration of the python function according to a function definition configuration operation input by the user. The function definition unit 12 may be used to define basic information and scripts of any one python function for a development user to perform definition and configuration operations. The function definition unit 12 provides the user with a function definition page for the user to define basic information of the python function, as shown in fig. 3, and may also provide the user with a script definition and configuration, as shown in fig. 4.

Optionally, in an embodiment of the present invention, the python function includes: basic information of functions and scripts of functions. It should be noted that the function referred to in the present invention is the python function.

Wherein, the basic information of the function comprises: function name, function code, function description, updater, and update time.

The script of the function includes: python edits the script.

In the embodiment of the invention, the python function can be used for creating and calling the algorithm template of the algorithm center module 20, and the function can be successfully applied by editing the importer function code in the algorithm script during the calling.

Optionally, in an embodiment of the present invention, the algorithm center module 20 includes: an algorithm template management unit 21 and an algorithm template definition unit 22.

The algorithm template management unit 21 is configured to store at least one non-executable algorithm template and output an algorithm template management page based on an algorithm template operation instruction input by a user. In the embodiment of the invention, the algorithm template management page can comprise all the algorithm template lists. As shown in fig. 5, the algorithm template management page displays all the algorithm templates included therein. All algorithm templates displayed by the algorithm template management page are non-executable empty templates, namely, empty templates which are not executable without binding data (wherein the algorithm templates can comprise, but are not limited to, model training algorithm templates and optimization calculation algorithm templates).

The algorithm template definition unit 22 is configured to provide an algorithm template definition page to a user, and perform definition and configuration of an algorithm template that is not executable according to an algorithm template definition configuration operation instruction input by the user. The algorithm template defining unit 22 may be used for defining and configuring basic information, algorithm script and file attachment of any algorithm template by a user. As shown in fig. 6, the algorithm template definition page provides a schematic diagram of basic information definition of the algorithm template, so that a user can define and configure the basic information of the algorithm template.

Optionally, in an embodiment of the present invention, the template information of the non-executable algorithm template includes: basic information of the algorithm template and definition information of the algorithm template.

The basic information of the algorithm template comprises: the name of the algorithm template, the coding of the algorithm template, the enabling state of the algorithm template, the description of the algorithm template, the updater of the algorithm template and the updating time of the algorithm template.

The definition information of the algorithm template comprises: basic information of the algorithm template, an algorithm script of the algorithm template and a file attachment of the algorithm template.

In the embodiment of the invention, the algorithm script of the algorithm template comprises: input parameter sets, output parameter sets, and python editing scripts are defined.

In the embodiment of the invention, the file attachment of the algorithm template comprises: at least one file uploaded by the user. Specifically, the file attachment in the algorithm template is at least one file uploaded by the user. When an algorithm template is referenced by an algorithm in algorithm application module 30, the file synchronization of the algorithm template is referenced. Further, in the embodiment of the present invention, the file in the file attachment of the algorithm template may be empty, that is, the algorithm template is an algorithm template that does not call the file, or in other words, the algorithm template is an algorithm template without a file.

In this embodiment, the enabled state of the algorithm template is whether the algorithm template can be referenced by the algorithm application module 30, and the user can switch the enabled and disabled states. If in an enabled state, the algorithm application module 30 may directly reference the algorithm template when creating the algorithm; in the disabled state, the algorithm application module 30 may not be referenced to the algorithm template when creating the algorithm.

In an embodiment of the present invention, the algorithm script provided in the algorithm template definition page includes a definition input parameter set and output parameter set, and a python editing script.

The creation of the input parameter set and the output parameter set is shown in fig. 7 and 8, and includes a parameter set name, a parameter set code, and parameter information (including all parameter names and parameter codes). For example, chiller-chilled water supply temperature (T1), and chilled water return temperature (T2). It should be noted that, in the embodiment of the present invention, the input parameter set and the output parameter set may not be created, that is, the algorithm template is an algorithm template that does not invoke binding data.

Further, in the embodiment of the present invention, the input parameter set and the output parameter set function to support script call binding data. Specifically, the data binding can be completed by defining a read value function and a write value function called by the Http interface (the function can be successfully applied by editing an import function code in a script). The read value function can acquire the bound instantiation device parameters under the parameter set and the corresponding parameters of the parameter set only by carrying in the parameter set code and the parameter code. For example, the read value function read.value (), the parameter set and the corresponding parameter code are the child/T1, the bound device and the corresponding parameter are the 1# host/chilled water supply temperature, and the value of the 1# host/chilled water supply temperature at the current time can be directly read by writing the script read.value (child/T1) into the algorithm template. Wherein the write value function is the same principle as the read value function.

As shown in fig. 9, the role of the file attachment of the algorithm template is to support script invocation of the user uploaded file. Specifically, the function of reading and writing the user interface file can be realized by defining a read file address function called by the Http interface (the function can be successfully applied by editing an import function code in a script), and the read file address function can be successfully called only by writing without uploading parameters. For example, the read file address function rule.readpath () reads the file attachment address of the algorithm when the algorithm template is referenced by the algorithm in the algorithm application module 30, and can directly call the file in the algorithm template.

Optionally, in an embodiment of the present invention, the algorithm application module 30 includes: an algorithm application management unit 31 and an algorithm definition unit 32.

The algorithm application management unit 31 is configured to store at least one executable algorithm and output an algorithm application management page based on an algorithm operation instruction input by a user. In the embodiment of the invention, the algorithm application management page can be used for the common user to carry out algorithm creation. As shown in fig. 10, the algorithm application management page may include all the algorithm lists. Wherein, the algorithm application manages all algorithms of the display of the page to be executable algorithms, namely algorithms which are executable by binding data (wherein, the algorithms can comprise, but are not limited to, model training algorithms and optimization calculation algorithms).

The algorithm definition unit 32 is used for providing an algorithm definition page to a user, and performing definition and configuration of an executable algorithm according to an algorithm definition configuration operation instruction input by the user. The algorithm definition page can be used for defining and configuring basic information, trigger conditions, algorithm configuration, file attachments and the like of any algorithm by a user.

Optionally, the algorithm information of the executable algorithm includes: basic information of the algorithm and definition information of the algorithm.

The basic information of the algorithm comprises: the method comprises the steps of calculating the name of an algorithm, encoding the algorithm, triggering conditions of the algorithm, enabling state of the algorithm, latest execution duration of the algorithm, description of the algorithm, updating person of the algorithm and updating time of the algorithm. As shown in fig. 13, basic information of an algorithm definition page provided by the present invention is schematically shown.

The definition information of the algorithm includes: basic information of the algorithm, trigger conditions of the algorithm, configuration of the algorithm and file attachment of the algorithm.

In the embodiment of the invention, the function of the enabling state of the algorithm is to judge whether the algorithm is executed or not, and the user can switch the enabling state and the disabling state. If the starting state is the starting state, the algorithm can be automatically executed according to the defined triggering condition, and the manual execution and the manual suspension of the user are supported; if the state is in the disabled state, the algorithm is not executed, and manual execution and manual suspension of the algorithm are not supported.

Optionally, in an embodiment of the present invention, the triggering conditions of the algorithm include: timing calculations, delay calculations, and event calculations. Specifically, the trigger condition of the algorithm is a condition defining the algorithm to be automatically executed. Wherein the conditions for automatic execution of the algorithm include; timing calculations (timing triggers), delay calculations (delay triggers), event calculations (event triggers, e.g., triggers greater than a certain set rated power or triggers greater than a certain set temperature), etc. Any one may be selected when the trigger condition definition setting of the algorithm is performed. For example, the user may select a timing trigger, input a specified time; alternatively, the user may select a delay trigger, entering a time interval for execution; or, the user can select event triggering, namely selecting an object model device point position or an instance device point position of the internet of things platform, editing a conditional expression and a conditional true and false triggering type. After the trigger condition definition setting is completed, the algorithm can be automatically executed according to the selected trigger condition. As shown in fig. 14, a schematic diagram of trigger conditions of an algorithm defining a page is provided for the algorithm of the present invention.

The configuration of the algorithm comprises: after referencing the non-executable algorithm template in the algorithm center module 20, data binding is performed on the input parameter set and the output parameter set of the referenced non-executable algorithm template. The configuration of the algorithm is that after the algorithm template in the algorithm center module 20 is referenced, the input parameter set and the output parameter set of the algorithm template are subjected to data binding, and the instantiation binding is supported according to the equipment model or the equipment instance mode. As shown in fig. 15, an algorithm configuration diagram of the algorithm definition page provided by the present invention is shown.

Specifically, the user selects the algorithm template to be cited first, after completing the selection of the algorithm template, the algorithm template is automatically brought into the input parameter set and the output parameter set of the algorithm template, and then, data binding is needed to be sequentially carried out on each input parameter set and each output parameter set. When data binding is performed, two modes of device model or device instance can be selected for instantiation. For example, when a user selects an equipment model, a corresponding equipment model can be selected from the internet of things platform according to the name of the parameter set, then an equipment instance under the equipment model is checked, finally the parameter points of the equipment model are matched according to the one-to-one mapping relation of the parameters under the parameter set (for example, the parameter set and the parameters of the algorithm template are that a cold water host machine is selected as the cold water host machine, the cold water supply temperature is selected as the cold water host machine, the equipment instance is selected as the 1# host machine and the 2# host machine, and the cold water supply temperature of the equipment model and the cold water return temperature of the equipment model are selected according to the cold water supply temperature). When a user selects an equipment instance, the one-to-one mapping relation of all parameters under the parameter set is used for matching the parameter points of the equipment instance of the internet of things platform (for example, the parameter set and the parameters of the algorithm template are that a cold water host machine-frozen water supply temperature and a cooling backwater temperature, and the internet of things platform is that a 1# host machine/frozen water supply temperature is selected according to the frozen water supply temperature and a 1# host machine/cooling backwater temperature is selected according to the cooling backwater temperature).

In the embodiment of the invention, the file attachment of the algorithm comprises: by referencing files synchronized when the algorithm is not executable in the algorithm center module 20, files uploaded by the user, or files stored when the algorithm is executed. By setting the file attachment of the algorithm, the capability of interaction between the algorithm and the file is realized, the user uploads file data to the algorithm from the file attachment, and the algorithm reads the file and stores the file in the file attachment after executing the file. As shown in FIG. 16, a schematic of a file attachment for an algorithm defining a page for an algorithm provided by the present invention.

Further, in the embodiment of the present invention, the algorithm application management unit 31 includes: and the algorithm execution subunit is used for performing control, suspension control or log operation on the executable algorithm.

The execution control of the executable algorithm is that a user can manually trigger an unexecuted algorithm, the suspension control is that the user can manually suspend the executed algorithm, and the execution log operation is that the user can check log records of historical execution, including start execution time, end execution time, execution duration, execution results and the like. FIG. 11 is a schematic diagram of an algorithm execution log in an algorithm application management page. Fig. 12 is a schematic diagram showing an execution result of an algorithm execution log.

It should be understood that, in the present invention, the function library module 10 is mainly used for constructing an algorithm template, that is, the function library 10 is applied to the construction of the algorithm template in the algorithm center module 20, and the algorithm template in the algorithm center module 20 is applied to the algorithm application in the algorithm application module 30 for execution.

Referring to fig. 29, the present invention provides a flowchart of a preferred embodiment of an AI algorithm engine method applied to an internet of things platform.

The AI algorithm engine method applied to the Internet of things platform can be realized based on the AI algorithm engine system applied to the Internet of things platform disclosed by the embodiment of the invention.

Specifically, as shown in fig. 29, the AI algorithm engine method applied to the internet of things platform includes the following steps:

and step S10, receiving a function operation instruction input by a user, executing function creation according to the function operation instruction, and outputting a function management page and a function definition page of the python function.

And step S20, receiving an algorithm template operation instruction input by a user, and executing algorithm template creation according to the algorithm template operation instruction so as to output an unexecutable algorithm template management page and an algorithm template definition page.

Step S30, receiving an algorithm operation instruction input by a user, executing algorithm creation according to the algorithm operation instruction, and outputting an executable algorithm management page and an algorithm definition page.

In the following, a specific embodiment will be described.

Specifically, referring to fig. 17 to 28, an algorithm for load prediction is configured for the user, and then the output embodiment looks at log execution results, as shown in fig. 17 to 26.

Function creation is first performed. It should be noted that, the function is not necessarily used to create the algorithm template. For example, creating an algorithmic template (named LSTM+att) that predicts cold contribution does not require the use of functions in the function library module 10.

Then, an algorithm template creation is performed.

Specifically, as shown in fig. 17, the basic information of the algorithm template is first filled in. Next, as shown in fig. 18, a parameter set is created under the input parameter and the output parameter. For example, a parameter group name, a parameter group code, parameter information (parameter name, parameter code), and the like are configured. Next, as shown in fig. 19, the python script is edited. Then, as shown in fig. 20, the file attachment is uploaded. It should be noted that some algorithm templates do not need to upload file attachments, and are mainly determined according to whether the defined algorithm template needs a file or not. As shown in FIG. 21, once the creation of the algorithm template is completed, the created algorithm template (LSTM+att) is visible in the algorithm center management page.

As shown in fig. 22, a basic information diagram created for a certain algorithm (lstm+att).

As shown in fig. 22, basic information of an algorithm is defined first when the algorithm is created. Next, as shown in fig. 23, the trigger condition of the algorithm is set. Next, as shown in fig. 24, an algorithm configuration is performed, wherein the algorithm template referred to by the algorithm is the algorithm template of lstm+att created previously. After the algorithm template of lstm+att is selected, the algorithm template of lstm+att is automatically synchronized from the attached file attachments as shown in fig. 25. Finally, as shown in FIG. 26, after the algorithm creation is completed, the LSTM+ATT algorithm and related information is visible on the algorithm application management page.

Further, as shown in fig. 27, an operation diagram of the execution log of the lstm+att algorithm is shown, and based on the diagram, the execution log of the lstm+att algorithm can be viewed for a period of time. Wherein, fig. 28 shows a schematic diagram of the execution log viewing result of the lstm+att algorithm.

The AI algorithm engine system and the AI algorithm engine method applied to the Internet of things platform can realize development and unified management on the Internet of things platform, a preset algorithm template is built in an AI algorithm center, an algorithm application support quotation algorithm binds data to build an executable algorithm, an application module does not need to be intervened by a developer, only needs to call equipment instance binding data of the Internet of things platform, algorithm production efficiency is greatly improved, algorithm automatic production capacity is also realized, and meanwhile, the AI algorithm engine is integrated with the traditional Internet of things platform to truly endow the platform with algorithm capacity without independent deployment, so that manpower and material resource investment of customized development is reduced.

Specifically, the invention realizes the algorithm capability of the internet of things platform, applies the AI algorithm engine to the traditional internet of things platform (namely the connection platform of the things), endows the internet of things and the things with the calculation capability of the AI algorithm, is not simple basic operation any more, and promotes the change of the quality of the internet of things platform. The unification of the internet of things platform and the algorithm platform is realized, all modules of the AI algorithm engine are developed and managed by the internet of things platform, the problem that the internet of things platform and the algorithm application are mutually independent is solved, the cost is low, the dependence is low, and the deployment efficiency is high; the algorithm capability of low-code mass production is realized, a general algorithm template is created in an AI algorithm engine algorithm center module of the Internet of things platform, an algorithm application module can refer to a plurality of projects and multiplex the algorithm template to create an algorithm, and the algorithm can be executed only by binding data, so that the efficiency of algorithm production and deployment is greatly improved. The automatic operation and visual monitoring of the algorithm are realized, the automatic execution of the trigger condition is defined by an AI algorithm engine algorithm application module of the platform of the Internet of things, and the manual execution and the manual suspension of the algorithm are also supported; in addition, the algorithm execution log can check the historical execution record for a period of time, namely the start execution time, the end execution time, the execution duration and the execution result. The operation and maintenance deployment efficiency of the algorithm and the convenience of algorithm debugging are improved. The method realizes the mutual call of the algorithm and the file, and the interface function of the file attachment of the AI algorithm engine of the platform of the Internet of things can be used for uploading file data from the file attachment to the algorithm and downloading the algorithm output file from the file attachment. The algorithm function and the expansibility of the algorithm are improved, and the application of the big data algorithm and the application of the picture audio video recognition algorithm are supported.

In the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.

Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative elements and steps are described above generally in terms of functionality in order to clearly illustrate the interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.

The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. The software modules may be disposed in Random Access Memory (RAM), memory, read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

The above embodiments are provided to illustrate the technical concept and features of the present invention and are intended to enable those skilled in the art to understand the content of the present invention and implement the same according to the content of the present invention, and not to limit the scope of the present invention. All equivalent changes and modifications made with the scope of the claims should be covered by the claims.

Claims

1. An AI algorithm engine system applied to an internet of things platform, comprising: the algorithm center module and the algorithm application module;

2. The AI algorithm engine system for an internet of things platform of claim 1, further comprising: a function library module;

3. The AI algorithm engine system for an internet of things platform of claim 2, wherein the function library module comprises:

4. The AI algorithm engine system for application to an internet of things platform of claim 3 wherein the python function includes: basic information of functions and scripts of the functions;

the script of the function includes: python edits the script.

5. The AI algorithm engine system for an internet of things platform of claim 1, wherein the algorithm center module comprises:

6. The AI algorithm engine system for an internet of things platform of claim 5, wherein the template information of the non-executable algorithm template comprises: basic information of an algorithm template and definition information of the algorithm template;

7. The AI algorithm engine system for application to an internet of things platform of claim 1, wherein the algorithm application module comprises:

8. The AI algorithm engine system for an internet of things platform of claim 7, wherein the algorithm information of the executable algorithm comprises: basic information of an algorithm and definition information of the algorithm;

9. The AI algorithm engine system applied to the internet of things platform according to claim 7, wherein the algorithm application management unit includes:

10. The AI algorithm engine method applied to the platform of the Internet of things is characterized by comprising the following steps: