Automatic control method and system based on AI kernel

Abstract

The invention relates to the field of software development, in particular to an automatic control method and system based on an AI kernel, and the method comprises the steps: firstly responding to a production demand input by a user or a feedback end, carrying out the technological process splitting of the production demand based on a preset dynamic planning and task planning AI model, and carrying out the processing of the production demand; the method comprises the following steps: generating corresponding sub-task process parameters and a sub-task sequence, identifying categories of the sub-task process parameters, calling corresponding standardization modules according to the categories, and then replacing initial attribute parameters in the standardization modules with the sub-task process parameters to obtain a plurality of pre-reconstruction modules; and finally, combining the plurality of pre-reconstruction modules according to a subtask sequence, and outputting a production operation flow instruction. Compared with the prior art, the method has the advantages that the production requirements of new products can be met, the corresponding production operation process instruction can be automatically generated, professionals do not need to define a new program framework and logic, and the technical problem that in the prior art, the automatic programming efficiency is low is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of software development. More particularly, the present application relates to an AI kernel-based automatic control method and system. BACKGROUND

[0002] With the rapid advancement and iterative upgrading of industrial automation technology, and the increasing demand for industrial product diversification and individualization in the market, modern production enterprises are facing unprecedented challenges and opportunities. Consumers have higher requirements for product functions, design, and even production speed, prompting production parties to have the ability to quickly respond to market changes and continuously adjust and optimize automated production process flows to ensure efficient and flexible production of industrial products that meet diversified needs. This dynamic adjustment not only concerns product market competitiveness, but also is the key to the continuous innovation and development of enterprises.

[0003] However, in actual production applications, when the automated production process flow is adaptively adjusted, it is usually necessary to rely on highly knowledgeable programmers to manually reconfigure the corresponding program framework or execution logic. This process not only consumes time and effort, involves complex code modification, testing and verification, and system integration, but also is prone to errors due to human factors, increasing project risks and costs. In a rapidly changing market environment, this traditional manual programming method is particularly inefficient, often making it difficult to complete process adjustments and put into production in a short period of time, resulting in missed market opportunities and affecting the overall operational efficiency and customer satisfaction of the enterprise.

[0004] Therefore, the prior art has the technical problem of low efficiency of automated program writing. SUMMARY

[0005] To solve the above technical problem of low efficiency of automated program writing, the present application discloses an AI kernel-based automatic control method and system.

[0006] In a first aspect, the present application discloses an AI kernel-based automatic control method, comprising: In response to production requirements input by a user or a feedback end, based on a preset dynamic planning and task planning AI model, the production requirements are processed to generate corresponding subtask process parameters and subtask sequences; Identify the category of the subtask process parameters, and call the corresponding standardized module according to the category; Replace the initial attribute parameters in the standardized module with the subtask process parameters to obtain a plurality of pre-reconfiguration modules; Combine the plurality of pre-reconfiguration modules according to the subtask sequence, and output the production operation flow instruction.

[0007] Beneficial effects: the method of the present application first splits the process according to the production requirements of the user end or the production requirements of the feedback end in the production process to obtain sub-task process parameters and sub-task sequence, then identifies the categories of the sub-task process parameters, calls the corresponding standardized module according to the categories, replaces the initial attribute parameters in the standardized module with the task process parameters to obtain a plurality of pre-reconstruction modules, and then combines the pre-reconstruction modules in sequence to automatically generate production operation process instructions to drive the corresponding production equipment. Compared with the prior art, the method of the present application can cope with the production requirements of new products, automatically generate corresponding production operation process instructions, and does not require professional personnel to define new program frameworks and logic, thereby solving the technical problem of low efficiency of automatic program writing in the prior art.

[0008] Preferably, the dynamic planning at least includes action sequence planning, delay planning, action trigger planning, action shielding planning, and action path planning.

[0009] Preferably, after generating the corresponding sub-task process parameters and sub-task sequence, the method of the present application further comprises: determining whether the sub-task process parameters exceed a preset action constraint range; if so, screening out the corresponding sub-task process parameters and sub-task sequence, and returning to the step of splitting the process flow according to the production requirements.

[0010] Beneficial effects: the method of the present application can judge the automatically split sub-task process parameters to determine whether they conform to the allowed activity range of the device layer. If they exceed the range, it means that the sub-task process parameters obtained by automatic splitting are unreasonable and need to be adjusted. The splitting step of the task planning AI model is returned to optimize the parameters. Compared with the prior art, the method of the present application has a parameter self-diagnosis function and can maximize the avoidance of execution errors of automatically generated programs.

[0011] Preferably, the action constraint at least includes linear motion device horizontal constraint, rotation device forward and reverse constraint, positive and negative pressure device air pressure constraint, clamping device opening and closing constraint, and transportation device path constraint.

[0012] Preferably, after outputting the production operation process instructions, the method of the present application further comprises: in response to the state signal feedback of the production operation process instructions, generating a production scene dynamic monitoring map.

[0013] Beneficial effects: after outputting the production operation process instructions, the corresponding production scene dynamic monitoring map is immediately refreshed, which can follow up the running situation of the equipment in real time and assist the technical personnel in monitoring the running risks of the automatic programming and automatic execution of the equipment.

[0014] Preferably, the scene dynamic monitoring map carries the whole production line position information, the feeding and discharging position information, the equipment state information and the warehouse storage position information.

[0015] Preferably, the feedback input production demand includes a product taking-out signal, and after outputting the production operation process instruction, the method further comprises: In response to the product taking-out signal, the action path planning of the corresponding conveying device is updated according to the production scene dynamic monitoring map and the transportation device path constraint. When the action path planning is updated, the process of splitting the production demand into the production operation process instruction is executed.

[0016] Preferably, the standardized module at least includes a task layer, a skill layer, a device layer and an execution layer. The task layer is used to define the subtask process. The skill layer is used to bind the subtask process with the basic action function. The device layer is used to bind the basic action function and the corresponding execution device attribute. The execution layer is used to convert the basic action function into the action instruction of the execution device.

[0017] Beneficial effects: the method predefines a set of standardized modules, and encapsulates the standardized modules, each standardized module has a certain independence, can realize "plug and play", Supports calling and combining at any time, and drives the corresponding execution device to work.

[0018] Preferably, the execution device attribute at least includes the number and type parameters of the execution device.

[0019] Beneficial effects: the number and type parameters of the execution device attribute are beneficial to distinguish the execution device position and function, thereby facilitating the accurate positioning of the execution device.

[0020] In the second aspect, the application further discloses an automatic control system based on an AI kernel, which comprises a processor and a memory, and the memory stores computer program instructions.

[0021] The beneficial effects of the application are as follows: (1) Compared with the prior art, the method can cope with the production demand of new products, automatically generate corresponding production operation process instructions, and does not need professional personnel to define new program framework and logic, thereby solving the technical problem of low efficiency of automatic program writing in the prior art.

[0022] (2) The method can judge the automatically split subtask process parameters, judge whether they meet the allowed activity range of the device layer, if they exceed the range, it means that the subtask process parameters obtained by automatic splitting are unreasonable, and need to be adjusted, return to the splitting step of the task planning AI model to optimize the parameters. Compared with the prior art, the method has a parameter self-diagnosis function, which can maximize the avoidance of execution errors of automatically generated programs.

[0023] (3) Compared with the prior art, the method of the present application refreshes the corresponding production scene dynamic monitoring map in real time after outputting the production operation process instruction, which can follow up the running situation of the equipment in real time, and assist the technical personnel to monitor the running risk of the automatic programming and automatic execution of the equipment.

[0024] (4) Compared with the prior art, the method of the present application predefines a set of standardized modules, and encapsulates the standardized modules, each standardized module has a certain independence, can realize "plug and play", supports calling and combining at any time, and drives the corresponding execution equipment to work. BRIEF DESCRIPTION OF DRAWINGS

[0025] The above and other objects, features and advantages of the exemplary embodiments of the present application will be more apparent from the following detailed description read in conjunction with the accompanying drawings, in which several embodiments of the present application are shown by way of example, and wherein the same or corresponding elements are referred to by the same or corresponding reference numerals, in which: Figure 1 is a flow chart of the AI kernel-based automatic control method in embodiment one of the present application; Figure 2 is a structural schematic diagram of the AI kernel-based automatic control system in embodiment two of the present application. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0027] The present embodiment discloses an AI kernel-based automatic control method and system, which is used to solve the technical problem of low efficiency of automatic program writing.

[0028] The specific embodiments of the present application will be described in detail below with reference to the drawings.

[0029] Embodiment one As Figure 1As shown, the embodiment discloses an AI kernel-based automatic control method, which comprises: S10: in response to the production demand input by the user or the feedback end, based on the preset dynamic planning and task planning AI model, the production demand is processed flow splitting to generate corresponding subtask process parameters and subtask sequence.

[0030] In this embodiment, the production demand of the user refers to the product production demand proposed by the artificial (generally a new product or an improved model of an old product), which can be a demand proposed by text or a demand proposed by voice, and the expression of the production demand can be "produce the assembly shown in the design drawing" or "transport the raw materials of model X01 from the warehouse". The production demand of the feedback end is usually proposed by the equipment itself, for example, the new product raw materials lacking in the production line will propose the production demand of "fetching the new product raw materials from the warehouse". The dynamic planning of different devices differs in specific parameters, but the overall framework planning differs little, basically having action sequence planning, delay planning, action trigger planning, action shielding planning and action path planning. The task planning AI model can adopt a BERT-based semantic understanding model, which is adapted to the automatic programming environment of the present method by pre-training.

[0031] Specifically, after the user or the feedback end inputs the production demand, the BERT model is used for semantic understanding of the production demand and combined with the dynamic planning, the process flow that may need to be used in the production demand is combined with the dynamic planning, and the process flow with multiple combinations or sequence actions is split into corresponding subtask process parameters and subtask sequence (for example, when receiving the production demand of screwing, the production demand is decomposed into positioning subtask, tightening subtask and / or loosening subtask).

[0032] Further, in order to avoid errors in the automatically generated execution logic, after the corresponding subtask process parameters and subtask sequence are generated in the above step S10, the method of the embodiment further comprises: S101: judging whether the subtask process parameters exceed the preset action constraint range; S102: if yes, the corresponding subtask process parameters and subtask sequence are screened out, and the step of processing flow splitting of the production demand in step S10 is returned.

[0033] In this embodiment, the above-mentioned action constraint range at least includes linear motion device horizontal constraint, rotation device forward and reverse constraint, positive and negative pressure device air pressure constraint, clamping device opening and closing constraint and transportation device path constraint.

[0034] Through the above steps S100-S200, if the sub-task process parameters exceed the range, it indicates that the automatically split sub-task process parameters are unreasonable and need to be adjusted again, and the splitting step of the task planning AI model is returned to optimize the parameters. Compared with the prior art, the method of the present application has a parameter self-diagnosis function, and the output sub-task process parameters and sub-task sequence can be closely combined with the actual physical environment, avoiding the appearance of instructions exceeding the process range, not meeting the physical rules or not meeting the action sequence, and can maximize the avoidance of execution errors of automatically generated programs (for example, when receiving a task of tightening a screw, the action can be accurately constrained to tightening, not loosening).

[0035] S103: If not, step S20 is performed.

[0036] S20: Identify the category of the sub-task process parameters, and call the corresponding standardization module according to the category.

[0037] In this embodiment, the category of the sub-task process parameters is mainly determined according to the type of the execution equipment, which includes linear motion equipment, rotating equipment, positive and negative pressure equipment, clamping equipment and transportation equipment, etc. The corresponding standardization modules of the sub-task process parameters of different categories are different.

[0038] It needs to be supplemented and explained that although the standardization modules are different, basically the standardization modules have task layer, skill layer, equipment layer and execution layer. The above four levels are used as the standard encapsulation framework of the standardization module to realize the relative independence of the standardization module, and then realize the multi-module assembly of the process flow.

[0039] Specifically, the above-mentioned task layer is used to define the sub-task process flow, that is, the name or project group of the process flow to which the standardized module belongs (such as the process flow belonging to "assembling rubber ring"). The above-mentioned skill layer is used to bind the sub-task process flow with the basic action function, and the basic action function involves basic action functions such as grasping, translating, rotating, inhaling, blowing, heating, cooling, and cutting. The basic action functions required by different sub-task process flows are different. In order to provide the efficiency of function calling, the sub-task process flow is bound with the basic action function in advance in the embodiment. The device layer is used to bind the basic action function and the corresponding execution device (such as a mechanical arm or a mobile chassis) attribute, wherein the execution device attribute at least includes the number and type parameters of the execution device, so as to facilitate the positioning of the corresponding execution device at the first time after the standardized module is called out, and the corresponding output result in the subsequent processing process is placed into the corresponding execution device. The execution layer is used to convert the basic action function into the action instruction of the execution device, so as to convert the host computer code (C language, Java, Python, or graphical code) of the standardized module into the mechanical code (G code, assembly, or industrial computer ladder diagram code, etc.) that can be read by the execution device, thereby meeting the needs of various hardware interfaces.

[0040] Through the above-mentioned predefinition of the standardized module, the embodiment realizes the deep program encapsulation of each sub-task process flow. The method recorded in the embodiment is like building a set of "plug and play" intelligent building blocks for complex industrial automation systems. Each module encapsulates the core logic and function of a specific process flow, which not only ensures the internal delicacy and efficiency, but also provides a simple and unified interface. The above-mentioned standardized module makes it unnecessary to deeply understand the specific implementation details of each sub-task in the subsequent process flow construction or adjustment process. It only needs to flexibly select and combine the corresponding standardized module according to the production demand, so as to quickly build an automated production line that meets the requirements. Moreover, the standardized modules have a certain independence and can be easily migrated and reused in different hardware platforms, software environments, or production scenarios, greatly enhancing the adaptability and expansibility of the embodiment, and providing strong technical support for diversified and variable production demands.

[0041] S30: Replace the initial attribute parameters in the standardized module with the sub-task process parameters to obtain a plurality of pre-reconstruction modules.

[0042] Specifically, the initial attribute parameters of the basic action function in the standardized module are replaced by the sub-task process parameters, so as to realize the scene action adaptation and action optimization of the standardized module. Then, the standardized module that has been action-optimized is pre-stored as a pre-reconstruction module for subsequent combined calling.

[0043] S40: Combine the plurality of pre-reconstruction modules in the order of sub-tasks to output a production operation flow instruction.

[0044] Need to explain the steps, the combination operation in the above step S40 can be realized based on the graphical programming window, to intuitively embody the production operation flow instruction of the combination output, thereby facilitating the real-time monitoring of professionals.

[0045] Through the above steps S10-S40, the method of the embodiment can cope with the production demand of new products, automatically generate corresponding production operation flow instructions, and does not need professionals to define new program framework and logic, solving the technical problem of low efficiency of automatic program writing in the prior art.

[0046] Preferably, after the above step S40, the method of the embodiment further comprises: S50: generating a production scene dynamic monitoring map in response to the state signal feedback of the production operation flow instruction.

[0047] The above step S50 is based on the IO signal fed back by the production operation flow instruction, and the corresponding production scene dynamic monitoring map is refreshed in real time to intuitively show the actual working state of the production line, thereby assisting the technical personnel in monitoring the equipment driven by automatic programming, and avoiding operation risks.

[0048] Preferably, the scene dynamic monitoring map carries the overall position information of the production line, the feeding and discharging position information, the equipment state information and the warehouse storage position information.

[0049] Specifically, the positioning and marking of the position information can be realized by coordinate positioning.

[0050] Through the above technical solution, the overall position information of the production line, the feeding and discharging position information, the equipment state information and the warehouse storage position information can be accurately presented in the production scene dynamic monitoring map, thereby facilitating the production process risk monitoring of the technical personnel.

[0051] Further, the production demand input by the feedback end includes a product taking-out signal, and after the above step S50, the method of the embodiment further comprises: S500: updating the action path planning of the corresponding conveying equipment according to the production scene dynamic monitoring map and the transportation equipment path constraint in response to the product taking-out signal.

[0052] S501: when the action path planning is updated, the step of process flow splitting of the production demand to output the production operation flow instruction is executed.

[0053] Exemplarily, when the host computer receives the product taking-out signal indicating that the feeding device needs to put the raw materials in the warehouse, the method of the embodiment can dynamically monitor the map according to the scene to accurately update the action path planning of the corresponding conveying device. When the action path planning is updated, the steps S10-S40 are executed to drive the conveying device to automatically go to the warehouse to obtain the raw materials and put them into the feeding device.

[0054] Through the steps S500-S501, the method of the embodiment can automatically update the action path planning of the corresponding conveying device in real time according to the dynamic monitoring map and the transportation device path constraint after receiving the product taking-out signal. The updated action path planning is used as the dynamic planning of the corresponding conveying device, and then the steps S10-S40 are executed again, so as to realize the automatic programming and parameter optimization of the process flow of obtaining raw materials in the warehouse, and further improve the efficiency of the method of the embodiment.

[0055] In summary of the above technical description, the method of the embodiment has at least the following advantages: (1) The production line program is generated according to the production demand, and does not need to be reconstructed and parameter adjusted by manual. The production line program reconstruction and parameter optimization can be quickly completed by relying on artificial intelligence technology, and the efficiency is higher.

[0056] (2) The process of automatic reconstruction is limited by setting the dynamic planning and action constraint range, which can maximize the possibility of avoiding the failure of the automatically generated program, and realizes the safe production of the enterprise.

[0057] (3) It can adapt to various engineering application environments and perform self-adaptive adjustment according to specific application environments, and has stronger universality.

[0058] Embodiment Two As shown in Figure 2 , the embodiment discloses an automatic control system based on an AI kernel. The system includes a processor and a memory. The memory stores computer program instructions. When the computer program instructions are executed by the processor, the automatic control method based on the AI kernel recorded in the embodiment is realized.

[0059] The system of the embodiment also includes a communication interface and other components familiar to those skilled in the art. The settings and functions thereof are known in the art, and thus will not be described here.

[0060] In this description, the term "application" also means any composition or device that comprises a program or a piece of code that can be implemented on a computer or networked device. The term "application" comprises software and / or hardware that implements the methods described in this specification. A "computer" or a "computer system" can be any processor-based or digital computer system.

[0061] In the description of the specification, the meaning of "a plurality of" is at least two, for example, two, three or more, and the like, unless specifically defined otherwise.

[0062] While several embodiments of the application have been shown and described in this specification, it is to be understood that these embodiments are merely exemplary of the many possible specific embodiments which exist. Those skilled in the art will recognize that many modifications, changes and substitutions are possible in the present application without departing from the spirit and scope of the application. It is therefore intended that the present application extend to all such modifications, changes and substitutions as well as to the equivalents thereof.

Claims

1. An automated control method based on an AI kernel, characterized in that, include: In response to production demands input by users or feedback terminals, based on preset dynamic planning and task planning AI models, the production demands are broken down into process flows, generating corresponding sub-task process parameters and sub-task order. Identify the category of the subtask process parameters and call the corresponding standardization module according to the category; By replacing the initial attribute parameters in the standardized module with the subtask process parameters, multiple pre-reconstructed modules are obtained; Multiple pre-reconstruction modules are combined according to the sub-task order to output production operation process instructions.

2. The AI-based automated control method according to claim 1, characterized in that, The dynamic programming includes at least action sequence planning, delay planning, action triggering planning, action shielding planning, and action path planning.

3. The AI-based automated control method according to claim 1 or 2, characterized in that, After generating the corresponding sub-task process parameters and sub-task order, the method further includes: Determine whether the process parameters of the subtask exceed the preset action constraint range; If so, filter out the corresponding subtask process parameters and subtask order, and return to the step of breaking down the production requirements into process flows.

4. The AI-based automated control method according to claim 3, characterized in that, The motion constraints include at least the horizontal constraints of linear motion equipment, the forward and reverse rotation constraints of rotating equipment, the air pressure constraints of positive and negative pressure equipment, the opening and closing constraints of clamping equipment, and the path constraints of transport equipment.

5. The AI-based automated control method according to claim 4, characterized in that, After outputting the production operation process instructions, the method further includes: In response to the status signal feedback of the production operation process instructions, a dynamic monitoring map of the production scene is generated.

6. The AI ​​kernel-based automated control method according to claim 5, characterized in that, The scene dynamic monitoring map contains information on the overall location of the production line, loading and unloading locations, equipment status, and warehouse storage locations.

7. The AI ​​kernel-based automated control method according to claim 6, characterized in that, The production demand input at the feedback terminal includes a product retrieval signal. After outputting production operation process instructions, the method further includes: In response to the product retrieval signal, the motion path planning of the corresponding conveying equipment is updated according to the dynamic monitoring map of the production scenario and the path constraints of the transportation equipment. Once the action path planning is updated, the step of breaking down the production requirements into process flow instructions for outputting production operation flow instructions is executed.

8. The AI-based automated control method according to claim 1, characterized in that, The standardized module includes at least a task layer, a skill layer, a device layer, and an execution layer; The task layer is used to define the process flow of sub-tasks; The skill layer is used to bind the sub-task process flow to the basic action function; The device layer is used to bind the basic action function and the corresponding execution device attributes; The execution layer is used to convert the basic action functions into action instructions for the execution device.

9. The AI ​​kernel-based automated control method according to claim 8, characterized in that, The execution device attributes include at least the execution device number and type parameter.

10. An automated control system based on an AI kernel, characterized in that, It includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the AI ​​kernel-based automated control method according to any one of claims 1-9 is implemented.

Images