Flexible intelligent manufacturing system based on AI intelligent control

Abstract

The application relates to a flexible intelligent manufacturing system based on AI intelligent control, which comprises a rack, a first working module, a second working module, a position detection unit and a control unit. The rack is provided with a standard interface group; the first working module is provided with a module interface group, the standard interface group and the module interface group are matched to enable the first working module to be detachably installed on the rack, the module interface group and the standard interface group have the same matching size and connection form, so that the same type or different type working modules can be interchanged, and the control unit is configured to: acquire an offset change vector in a first unit time according to real-time position data, and predict a predicted offset at a first time node according to the offset change vector; if the predicted offset exceeds a first preset threshold, a preventive operation is performed on each working module. According to the predicted offset, the working module is adjusted in advance, and the problems of continuous production interruption and safety caused by the position offset of the working module are effectively solved.

Description

Technical Field

[0001] This invention relates to the technical field of automated equipment, and in particular to a flexible intelligent manufacturing system based on AI intelligent control. Background Technology

[0002] In the industrial manufacturing sector, traditional non-standard equipment typically employs a fully welded, fixed architecture, welding each working module to the frame. While this ensures stability in high-risk scenarios (acid transport in chemical reactors, semiconductor wafer transfer, and inductor product manufacturing), the fixed functionality necessitates complete disassembly and reassembly for production line changes, resulting in long downtimes and difficulty adapting to the demands of flexible production with diverse product types and small batches. To address this, the applicant developed a universal working machine with detachable working modules, enabling rapid interchangeability of various modules, significantly improving equipment reusability, and perfectly adapting to flexible production scenarios. However, during the development process, the applicant discovered that in certain application scenarios, some working modules of this universal working machine may experience positional shifts due to high-frequency operation, potentially affecting continuous production on flexible manufacturing lines and posing certain safety concerns. Summary of the Invention

[0003] The purpose of this invention is to provide a flexible intelligent manufacturing system based on AI intelligent control, which aims to solve the problem that some working modules in related technologies are prone to positional displacement, which may affect the continuous production of flexible manufacturing lines and pose certain safety issues.

[0004] To address the aforementioned technical problems, a flexible intelligent manufacturing system based on AI intelligent control is provided, comprising:

[0005] The rack is equipped with a standard interface group;

[0006] The first working module is provided with a module interface group. The standard interface group cooperates with the module interface group to allow the first working module to be detachably installed on the frame. The module interface group and the standard interface group have the same mating size and connection form to realize the interchangeability of the first working modules of the same or different types. The first working module includes one of the following: a feeding module, a loading module, a unloading module, and a process function module.

[0007] The second working module is fixedly installed on the frame. The second working module includes one of the following: a feeding module, a loading module, a unloading module, and a process function module. The first working module and the second working module are different types of working modules.

[0008] A position detection unit is used to collect real-time position data of the first working module and the second working module.

[0009] The control unit is electrically connected to the position detection unit and is configured to: obtain the offset change vector within a first unit time based on the real-time position data, and predict the predicted offset at a first time node based on the offset change vector; if the predicted offset exceeds a first preset threshold, perform preventive operations on the first working module and the second working module.

[0010] Furthermore, the second working module is configured as a feeding module, and the first working module is configured as a process function module, a loading module, or a unloading module; the second working module is configured as a loading module, and the first working module is configured as a process function module or an unloading module; the second working module is configured as an unloading module, and the first working module is configured as a process function module.

[0011] Furthermore, the step of performing preventative operations on the first working module and the second working module if the predicted offset exceeds a first preset threshold includes:

[0012] If the predicted offset exceeds the first preset threshold within the first time period, the working power of the first working module and the second working module shall be reduced.

[0013] If the predicted offset exceeds the first preset threshold during the second time period, a shutdown warning is triggered during the third time period. After the third time period, the first working module and the second working module are controlled to stop working.

[0014] Furthermore, the position detection unit includes a vision camera and a laser rangefinder, which are electrically connected to the control unit respectively. The vision camera is disposed above the frame, and the laser rangefinder is disposed beside each working module. The real-time position data acquired by the vision camera is denoted as A, and the real-time position data acquired by the laser rangefinder is denoted as B. Then, the offset change vector C = αA + (1-α)B, where the value of the data weight α can be dynamically adjusted according to the changes in working conditions.

[0015] Furthermore, the values ​​of the first preset thresholds differ depending on the type of the working module, and the first preset threshold corresponding to the process function module is lower than the first preset threshold corresponding to other types of working modules.

[0016] Furthermore, the control unit parses the process path and generates a correlation map. If the predicted offset exceeds a first preset threshold, preventive operations are performed on the corresponding working module and its upstream and downstream working modules.

[0017] Furthermore, the control unit is also configured to: obtain the current working module and the adjacent working modules according to the correlation map, calculate the relative offset change vector between the two working modules, and predict the predicted relative offset at the second time node according to the relative offset change vector; if the predicted relative offset exceeds the second preset threshold, perform preventive operations on each working module.

[0018] Furthermore, the standard interface group includes a mounting hole array, the mounting hole array including mounting holes distributed in an array, and the module interface group includes a through hole group, the through hole group including multiple through holes that mate with the mounting holes, the mounting holes and the through holes being connected by bolts.

[0019] Furthermore, the mounting holes in the mounting hole array are evenly distributed along mutually perpendicular X-axis and Y-axis directions at a first preset spacing to form a gridded positioning reference. The spacing between any two adjacent through holes is a second preset spacing, which is an integer multiple of the first preset spacing.

[0020] Furthermore, the AI-based flexible intelligent manufacturing system also includes a positioning mechanism, which includes multiple positioning pins. Each of the working modules includes a base, which includes a first stop and a second stop arranged adjacent to each other. During assembly, a portion of the positioning pins abut against the first stop and another portion of the positioning pins abut against the second stop.

[0021] Implementing the embodiments of the present invention will have the following beneficial effects:

[0022] The AI-based flexible intelligent manufacturing system in this embodiment achieves rapid interchangeability of different or similar types of work modules through the isomorphic design of the standard interface group of the rack and the modular fixed interface group of the work modules, significantly improving equipment reuse rate and flexible production adaptability. The position detection unit monitors the position of the work modules in real time, and through the cooperation of the position detection unit and the control unit, predicts the offset of the work modules after a unit of time. Based on the predicted offset, the system adjusts the work modules in advance, effectively solving the problems of continuous production interruption and safety issues caused by high-frequency work module position offset. The modular, detachable design and dynamic offset control help to balance the rapid changeover requirements of flexible production with operational safety. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a schematic diagram of the structure of the flexible intelligent manufacturing system based on AI intelligent control according to an embodiment of the present invention;

[0025] Figure 2 for Figure 1 A magnified view of a portion of point A in the middle;

[0026] Figure 3 This is an exploded view of the flexible intelligent manufacturing system based on AI intelligent control as described in an embodiment of the present invention;

[0027] Figure 4 This is a control principle diagram of the flexible intelligent manufacturing system based on AI intelligent control according to an embodiment of the present invention.

[0028] The system includes: 100, a flexible intelligent manufacturing system based on AI intelligent control; 110, a frame; 111, a standard interface group; 1111, mounting holes; 120, a first working module; 121, a module interface group; 1211, a through hole; 122, a first retaining edge; 123, a second retaining edge; 130, a second working module; 140, a position detection unit; 141, a vision camera; 142, a laser rangefinder sensor; 150, a control unit; 160, a positioning mechanism; 161, a positioning pin; 171, a feeding module; 172, a loading module; 173, a unloading module; and 174, a process function module. Detailed Implementation

[0029] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Preferred embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a thorough and complete understanding of the disclosure of the invention.

[0030] It should be noted that when a component is said to be "fixed to" another component, it can be directly attached to the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0031] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0032] Please refer to Figures 1-4This invention provides a flexible intelligent manufacturing system 100 based on AI intelligent control. The flexible intelligent manufacturing system 100 includes a frame 110, a first working module 120, a second working module 130, a position detection unit 140, and a control unit 150. The frame 110 is provided with a standard interface group 111; the first working module 120 is provided with a module interface group 121. The standard interface group 111 and the module interface group 121 cooperate to allow the first working module 120 to be detachably installed on the frame 110. Both the module interface group 121 and the standard interface group 111 have the same mating dimensions and connection form to achieve interchangeability of the same or different types of first working modules 120. The first working module 120 includes one of a feeding module 171, a loading module 172, a unloading module 173, and a process function module 174. The second working module 130 is fixedly mounted on the frame 110. The second working module 130 includes one of a feeding module 171, a loading module 172, a unloading module 173, and a process function module 174. The first working module 120 and the second working module 130 are working modules of different types. The position detection unit 140 is used to collect real-time position data of the first working module 120 and the second working module 130. The control unit 150 is electrically connected to the position detection unit 140. The control unit 150 is configured to: obtain the offset change vector within a first unit time based on the real-time position data, and predict the predicted offset at a first time node based on the offset change vector; if the predicted offset exceeds a first preset threshold, perform preventative operations on the first working module 120 and the second working module 130. For example, it should be noted that the position detection unit 140 can collect the offset of the working module in the X-axis direction and the Y-axis direction. The control unit 150 is electrically connected to the first working module 120 and the second working module 130 respectively, and is used to control the first working module 120 and the second working module 130 to stop, reduce working power, and issue alarms. Specifically, in this embodiment, the process function module 174 of this application can realize conventional processes such as dispensing, bending, cutting, resistance / continuity testing, laser paint stripping, CCD detection, 3D / 2D measurement, labeling / film correction, resistance welding, implantation assembly, and assembly. The feeding module 171 includes a vibratory feeder, vibratory feeder loading, magazine loading, and tray loading. The loading module 172 includes robot loading and three-axis module loading. The unloading module 173 includes a belt conveyor (fixture), heat sealing / cold sealing tape, double-speed chain (fixture), and tray unloading. In addition, it should be noted that the process function module 174, feeding module 171, loading module 172, and unloading module 173 of this application are standardized. That is to say, the process function module 174, flexible feeding module 171, robotic loading module 172, and unloading module 173 have standard bases, which can be adapted to be installed on the workbench, and the equipment can be universally used through disassembly and connection.

[0033] In some application scenarios, such as semiconductor wafer transport, traditional non-standard equipment requires complete disassembly and reassembly of the rack 110 during model changeovers due to the fixed welding of working modules. This can result in downtime of up to several hours. Furthermore, if the transport module experiences even a slight displacement due to high-frequency operation, it may cause wafer scratches or transport interruptions, affecting yield. In contrast, the flexible intelligent manufacturing system of this application enables production line model changeovers through rapid module replacement, significantly reducing downtime. Simultaneously, the module position detection unit 140 monitors the position of the transport module in real time. When the offset exceeds a first preset threshold (e.g., 0.5mm), the control unit 150 immediately stops the transport module and the associated loading / unloading module 173, preventing wafer damage. This improves changeover efficiency while ensuring production safety and product quality.

[0034] The AI-based flexible intelligent manufacturing system 100 includes a main control program, and each working module has a corresponding sub-control program. It should be noted that the main control program controls the overall process flow and execution steps of each working module in the flexible intelligent manufacturing system, while the sub-control programs control the specific execution processes and steps of individual working modules. Since the working modules in this application can be replaced as needed, an AI intelligent docking algorithm module is embedded in the main control program to ensure that the replaced working module can cooperate with the main control program. Based on the existing communication link between the main control and sub-control programs, automatic docking is completed through three steps: data acquisition, feature matching, and decision execution. First, automatic identification occurs after module replacement. When the working module is installed, the main control program automatically sends an identity query command to the sub-control program. The sub-control program returns a preset unique identifier and core hardware parameters. The AI ​​algorithm is based on a historically stored module feature database. The module feature database contains the IDs, functions, and parameters of all compatible modules. It uses a K-nearest neighbor classification algorithm to quickly match module types. If the feedback ID matches the database perfectly, the AI ​​directly confirms the module's identity, skipping parameter verification. If the feedback ID does not match but the core function / hardware parameters match (e.g., adding a process function module 174 of the same type but from a different batch), the AI ​​calculates parameter similarity (e.g., a parameter similarity deviation ≤ 10% indicates compatibility), automatically classifies it as an compatible module, and updates the feature database. If the feedback parameters deviate from the database by > 10% (e.g., mistakenly installing an incompatible module), the AI ​​immediately sends a module incompatibility warning to the main control program, prohibits equipment startup, and displays the reason for the incompatibility. The second step is automatic synchronization of control parameters. After module identity is confirmed, the AI ​​automatically completes parameter configuration based on the historical mapping relationship between module type and process requirements. The AI ​​retrieves the process requirements of the current production task from the main control program, combines the hardware parameters fed back by the sub-control program, and automatically calculates and issues adaptive parameters. The sub-control program receives these parameters and automatically adjusts the hardware status. If a related module is replaced, the AI ​​automatically retrieves the module collaboration rule library and synchronously updates the control logic of the main control program for other related modules to avoid collaboration conflicts. The third step is real-time status closed-loop verification. After the parameters are issued, the AI ​​collects feedback data from the sub-control program in real time (such as sensor signals and drive component operating status) through the main control program to verify the docking validity. If the sub-control program successfully receives the parameters and the hardware is in place, the AI ​​determines that the docking is complete and sends a module ready command to the main control program, allowing the equipment to start running immediately. If the parameters fed back by the sub-control program exceed the limits, the AI ​​automatically calculates the process parameter correction value or sends a process conflict warning to the main control program, prompting manual adjustment of the production task. If parameter drift occurs during operation, the AI ​​monitors the data from the sub-control program's sensors in real time and automatically issues correction commands without manual intervention.

[0035] Figure 1The AI-based flexible intelligent manufacturing system 100 shown in the paper includes four modules: a feeding module 171, a loading module 172, a unloading module 173, and a process function module 174. This is only to show that the AI-based flexible intelligent manufacturing system 100 of this application may include these four types of working modules, and does not mean that it is necessary to include all four types of working modules. In this implementation, it is only necessary to include two of the modules.

[0036] Please refer to Figure 1 , Figure 2 and Figure 3 In this application, all standard interface groups 111 and module interface groups 121 have the same mating dimensions and connection forms. Specifically, all module interface groups 121 of the working modules are completely identical to the standard interface groups 111 of the frame 110 in terms of dimensions (such as aperture and spacing), shape (such as circular or square), and connection method (such as threaded engagement or magnetic attraction). All standard interface groups 111 and module interface groups 121 have the same mating dimensions and connection forms. Specifically, all module interface groups 121 for picking up special components are completely identical to the standard interface groups 111 of the feeding module in terms of dimensions (such as aperture and spacing), shape (such as circular or square), and connection method (such as threaded engagement or magnetic attraction). This is the core technological foundation for achieving module interchangeability.

[0037] The AI-based flexible intelligent manufacturing system 100 of this application solves the problems of insufficient versatility and low changeover efficiency caused by interface fragmentation and fixed positions in existing modular production equipment through the collaborative design of the standard interface group 111 on the frame 110 and the module interface group 121 at the bottom of each working module. Since all working module interface groups adopt a unified mating size and connection form, different types of modules such as feeding, loading, function and unloading (such as the flexible vibration feeding module 171 and the clip feeding module 171, the cutting process function module 174 and the bending process function module 174) can be directly interchanged without the need for customized adapters or modification of interface structure, which significantly improves the reusability of the equipment. Meanwhile, the fixed units (such as meshed threaded holes or electromagnetic arrays) distributed along the preset direction in the standard interface group 111 support the free positioning of the working module on the mounting surface of the frame 110. The position is quickly fixed through physical locking mechanisms (such as bolt engagement or electromagnetic adsorption), completely eliminating the cumbersome disassembly, re-drilling and calibration process in traditional solutions. The module replacement time can be shortened from several hours to several minutes, which can greatly improve the replacement efficiency of each working module and help improve the versatility of the equipment.

[0038] The steps for obtaining the predicted offset include: denoising the real-time position data (Kalman filtering), normalizing it (mapping the displacement values ​​to the 0-1 range), and removing outliers from the installation / disassembly phase. An LSTM model is initially trained using offset data from 100 typical operating conditions (including stable operation, gradual offset, and sudden offset scenarios), and iteratively optimized until the prediction error is ≤0.05mm. The control unit 150 calls the model every second to predict the offset curve for the next 3 minutes based on the latest 30 seconds of data.

[0039] The AI-based flexible intelligent manufacturing system 100 in this embodiment, through the isomorphic design of the standard interface group 111 of the frame 110 and the modular fixed interface group of the working modules, enables rapid interchange of different or the same type of working modules, significantly improving equipment reuse rate and flexible production adaptability. The position detection unit 140 monitors the position of the working modules in real time. Simultaneously, through the cooperation of the position detection unit 140 and the control unit 150, the system predicts the offset of the working modules after a unit of time, and adjusts the working modules in advance based on the predicted offset, effectively solving the problems of continuous production interruption and safety issues caused by high-frequency working module position offset. The modular, detachable design and dynamic offset control help to balance the rapid changeover requirements of flexible production with operational safety.

[0040] Please refer to Figure 1In one possible implementation, the second working module 130 is configured as a feeding module 171, and the first working module 120 is configured as a process function module 174, a loading module 172, or a unloading module 173; the second working module 130 is configured as a loading module 172, and the first working module 120 is configured as a process function module 174 or a unloading module 173; the second working module 130 is configured as an unloading module 173, and the first working module 120 is configured as a process function module 174. Exemplarily, in the first embodiment, the first working module 120 is configured as a process function module 174, and the second working module 130 is configured as a feeding module 171. It is understood that the process function module 174 can be replaced or disassembled, for example, the dispensing module can be replaced with a bending module. This embodiment can realize a continuous process of automatic material supply to the core process, and is suitable for scenarios where there is no intermediate transfer, the process is simple, and the finished product can be directly output from the process function module 174. For example, in the inductor core dispensing process, the feeding module 171 directionally transports the core to the station of the process function module 174. After dispensing, the core slides from below the dispensing machine into the collection box, eliminating the need for a dedicated unloading mechanism. In the second embodiment, the first working module 120 is configured as the loading module 172, and the second working module 130 is configured as the feeding module 171. It is understood that the loading module 172 can be replaced and disassembled. This embodiment enables a pre-processing flow from material storage to precise transfer, suitable for scenarios requiring the transfer of materials from the feeding device to external equipment (such as the process function module of another automated device), serving as a feeding unit for the production line. For example, in the semiconductor chip transfer process, the feeding module 171 carries the chip, and the loading module 172 transfers the chip from the tray to the platform of an external testing device, returning to retrieve the chip after completion. In the third embodiment, the first working module 120 is configured as the unloading module 173, and the second working module 130 is configured as the feeding module 171. It is understood that the unloading module 173 can be replaced and disassembled. This embodiment enables a simplified process from material supply to finished product sorting and recycling, suitable for scenarios where materials require no processing and only need to be transferred from the supply end to different recycling ends (such as sorting and material distribution). For example, in the electronic component specification sorting process, the supply module 171 delivers capacitors of different sizes. After visual judgment, a person triggers the pusher of the unloading module 173 to push the large / small capacitors into the two side boxes respectively. In the fourth embodiment, the first working module 120 is configured as the process function module 174, and the second working module 130 is configured as the loading module 172. It is understood that the process function module 174 can be replaced and disassembled. This embodiment enables a process of precise material transfer to core process handling, suitable for scenarios where materials need to be picked up from external manual / automatic supply points, processed, and directly output (without special unloading).For example, in the automatic PCB board soldering process, the loading module 172 picks up the PCB board from the manually placed rack and puts it into the conveyor chain of the process function module 174. After soldering, it is output from the furnace tail. In the fifth embodiment, the first working module 120 is configured as the unloading module 173, and the second working module 130 is configured as the loading module 172. It is understood that the unloading module 173 can be replaced and disassembled. This embodiment can realize a pure material handling process of material picking → transfer → recycling, which is suitable for scenarios that require material transfer between two external devices (such as the connection unit in the middle of the production line). For example, during production line connection transfer, the loading module 172 picks up the inductor from the upstream winding machine, and the unloading module 173 transfers the inductor to the entrance of the downstream testing equipment, realizing seamless docking of the two devices. In the sixth embodiment, the first working module 120 is configured as the process function module 174, and the second working module 130 is configured as the unloading module 173. It is understood that the process function module 174 can be replaced and disassembled. This embodiment enables the process from core process handling to finished product classification and recycling. It is suitable for scenarios where materials are fed into the process function module 174 manually or by external equipment, and then need to be classified and recycled according to rules (such as qualified / unqualified). For example, during the appearance inspection of inductors, an external conveyor belt feeds the inductors into the process function module 174. Damaged inductors detected are sucked into the NG box by the unloading module 173, while qualified inductors fall into the OK box.

[0041] Please refer to Figure 4 In one possible implementation, if the predicted offset exceeds a first preset threshold, preventative operations are performed on the first working module 120 and the second working module 130, including:

[0042] If the predicted offset exceeds a first preset threshold within a first time period, the operating power of the first working module 120 and the second working module 130 is reduced. Specifically, if the predicted offset within 3 minutes is expected to exceed a first preset threshold (e.g., 0.5 mm), a power reduction command is immediately triggered. That is, in this embodiment, the first time period is 3 minutes, and the first preset threshold is set to 0.5 mm. Of course, in specific applications, the first time period can also be set to 2 minutes, 4 minutes, or 5 minutes. The first preset threshold can also be set to 0.3 mm, 0.4 mm, 0.6 mm, or 0.7 mm. High-frequency operation is one of the core causes of module positional offset (e.g., high-speed motor operation, continuous vibration generated by high-frequency robotic arm movements). Reducing the operating power (e.g., reducing motor speed, reducing output force) can directly reduce the dynamic load and vibration intensity of the module, thereby slowing down the rate of offset growth. One of the core requirements of flexible production is to reduce unplanned downtime. If a shutdown is triggered directly due to "predicted offset exceeding the standard," it may lead to work-in-process inventory buildup and production line rhythm disruption, especially affecting multi-process collaborative production lines (e.g., upstream material accumulation, downstream equipment waiting for materials). Reducing power can maintain the basic operational capability of the working module while controlling the risk of offset. Furthermore, excessive offset directly affects machining accuracy. When the offset is predicted to exceed the limit, reducing power can temporarily improve equipment stability and lower the probability of defective products by decreasing the dynamic error of the working module.

[0043] If the predicted offset exceeds the first preset threshold within the second time period, a shutdown warning is triggered within the third time period. After the third time period, the first working module 120 and the second working module 130 are controlled to stop working. For example, if the predicted offset within 1 minute is expected to exceed the first preset threshold (e.g., 0.5mm), a shutdown warning (e.g., an audible and visual alarm) is triggered, and the alarm lasts for 30 seconds to remind the staff. If the staff does not take any action after 30 seconds, the working modules are controlled to stop working.

[0044] Please refer to Figure 1 , Figure 2 and Figure 4In one possible implementation, the position detection unit 140 includes a vision camera 141 and a laser rangefinder 142, both electrically connected to the control unit 150. The vision camera 141 is mounted above the frame 110, and the laser rangefinder 142 is mounted beside each working module. The real-time position data acquired by the vision camera 141 is denoted as A, and the real-time position data acquired by the laser rangefinder 142 is denoted as B. Then, the offset change vector C = αA + (1-α)B, where the value of the data weight α can be dynamically adjusted according to changes in operating conditions. For example, in this embodiment, the value of α is set to 0.6, and the offset change vector C = 0.6A + 0.4B. The accuracy of the data is ensured by the two measurement values ​​acquired by the laser rangefinder 142 and the vision camera 141. The value of α can be adjusted according to changes in operating conditions. Operating conditions refer to the running state of the working module (such as stationary / moving, speed, load, etc.). For example, for a high-speed moving working module, the value of α can be increased to 0.8, in which case the data detected by the laser rangefinder 142 is more reliable and has a higher weighting. For a low-speed moving or stationary working module, the value of α can be decreased to 0.4, in which case the data detected by the vision camera 141 is more reliable and has a higher weighting. Furthermore, for heavy-load conditions, where the working module deforms or vibrates more, the laser rangefinder 142 is more sensitive to minute deformations, and the weighting α is increased to 0.8. For light-load conditions, where the working module is stable, the data from the laser rangefinder 142 and the vision camera 141 have equal weighting, and the weighting α is set to 0.5. In addition, it should be noted that there are multiple laser rangefinders 142, and at least two laser rangefinders 142 are provided on the first side of each working module to detect the offset in the X-axis direction, and at least two laser rangefinders 142 are provided on the second side of each working module to detect the offset in the Y-axis direction.

[0045] Please refer to Figure 4In one possible implementation, the first preset threshold values ​​differ depending on the type of the working module. The first preset threshold for process function module 174 is lower than the first preset threshold for other types of working modules. For example, a smaller first preset threshold is set for working modules with high precision requirements, and a larger first preset threshold is set for working modules with low precision requirements. For instance, for some high-precision process function modules 174, the first preset threshold is set to 0.3mm, and for the material feeding module 173 with general precision requirements, the first preset threshold is set to 0.5mm. The core of this solution lies in accurately matching the precision requirements and risk levels of different modules to achieve a dynamic balance between safety control and production efficiency. Firstly, it adapts to the precision requirements of different working modules, improving product quality control accuracy. Secondly, it optimizes risk-level control, reducing unnecessary downtime and efficiency losses. One of the core requirements of flexible production is to maximize equipment utilization while ensuring safety. Differentiated thresholds avoid efficiency losses caused by a one-size-fits-all standard by differentiating the risk levels of modules. By accurately identifying the functional importance and risk level of modules, the system control strategy can be refined and personalized. Its core value lies in maximizing equipment operating efficiency and reducing unnecessary intervention while ensuring the precision of core processes and product quality. It embodies the technical advantages of the AI-based flexible intelligent manufacturing system 100, which features "intelligent perception, hierarchical response, and dynamic adaptation," and provides reliable control support for flexible production in multiple scenarios and with high precision.

[0046] Please refer to Figure 4 In one possible implementation, the control unit 150 analyzes the process path and generates a correlation graph. If the predicted offset exceeds a first preset threshold, preventative operations are performed on the corresponding work module and its upstream and downstream work modules. For example, in flexible manufacturing, each work module forms an organic whole through the process path. An offset in a single work module may affect upstream and downstream processes through material transfer. The correlation graph, by analyzing the process path, clarifies the upstream and downstream dependencies between modules, ensuring that when a work module's offset exceeds the limit, the system not only intervenes in that module but also simultaneously controls its upstream modules (to prevent material accumulation due to continuous supply) and downstream modules (to prevent receiving defective products or idling). Therefore, the above solution facilitates end-to-end risk control and prevents cascading failures caused by local offsets.

[0047] Please refer to Figure 4In one possible implementation, the control unit 150 is further configured to: obtain the current working module and adjacent working modules according to the correlation map, calculate the relative offset change vector between the two working modules, and predict the predicted relative offset at a second time node based on the relative offset change vector; if the predicted relative offset exceeds a second preset threshold, preventive operations are performed on each working module. For example, the core objective of this scheme is to compensate for the limitations of absolute offset of a single module by monitoring the relative position accuracy between modules, thus avoiding positioning conflicts during collaboration. Module A's real-time coordinates are denoted as X1, and Module B's real-time coordinates are denoted as X2, with the relative offset change vector = |X1-X2|. If the second preset threshold < predicted relative offset ≤ 1.5 times the second preset threshold, the operating power of the current working module and its upstream and downstream working modules is reduced; if the second preset threshold ≥ 1.5 times the second preset threshold, all working modules on the control rack 110 are stopped from operating. For example, the absolute offset of the feeding module 172A is 0.2mm (not exceeding its own first preset threshold of 0.3mm), and the absolute offset of the process function module 174B is 0.2mm (not exceeding its own first preset threshold of 0.3mm), but the theoretical distance between the two is 150mm, and the actual distance is 150.5mm (relative offset of 0.5mm, exceeding the second preset threshold of 0.3mm). In this embodiment, the second time node is set according to the actual situation, for example, the second time node is set to 3 minutes, or 4 minutes, or 5 minutes.

[0048] Please refer to Figure 1 , Figure 2 and Figure 3 In one possible implementation, the standard interface group 111 includes an array of mounting holes 1111, the array of mounting holes 1111 including mounting holes 1111 distributed in an array, and the module interface group 121 includes a group of through holes 1211, the group of through holes 1211 including a plurality of through holes 1211 that mate with the mounting holes 1111, and the mounting holes 1111 and the through holes 1211 are connected by bolts.

[0049] In one possible implementation, the mounting holes 1111 in the mounting hole array are uniformly distributed along mutually perpendicular X-axis and Y-axis directions at a first preset spacing to form a gridded positioning reference. The spacing between any two adjacent through holes 1211 is a second preset spacing, which is an integer multiple of the first preset spacing. Exemplarily, the mounting holes 1111 and through holes 1211 are connected by bolts. By setting the spacing of the mounting holes 1111 to an integer multiple of the spacing of the through holes 1211, it is ensured that regardless of how the working module moves within the grid, the bottom group of through holes 1211 can be completely aligned with at least one group of mounting holes 1111. This embodiment of the application, through the integer multiple relationship between the spacing of the gridded mounting hole array 1111 and the through hole group 1211, supports rapid alignment of the module at any grid position on the frame 110, enabling flexible positioning of the working module. Furthermore, through the spacing matching rules, multi-module collaborative operation or single-module position migration can be easily achieved, providing rapid expansion capabilities. Simultaneously, it eliminates the need for customized chemical fittings in traditional solutions, significantly improves spare parts versatility, and enables low-cost upgrades. Additionally, it should be noted that the second preset spacing in the X-axis direction may or may not be the same as the second preset spacing in the Y-axis direction. For example, the second preset spacing in the X-axis direction may be 4 times the preset spacing, while the second preset spacing in the Y-axis direction may be 3 times the preset spacing. Furthermore, it should be noted that the X-axis direction and the Y-axis direction are the same. The array of mounting holes 1111 formed on the frame 110 allows the mounting orientation of each working module to be adjusted during installation; for example, the working module can be changed from horizontal (X-axis direction) mounting to vertical (Y-axis direction) mounting.

[0050] Please refer to Figure 1 , Figure 2 and Figure 3In one possible implementation, the AI-based flexible intelligent manufacturing system 100 further includes a positioning mechanism 160. The positioning mechanism 160 includes multiple positioning pins 161. Each working module includes a base, and the base includes a first stop 122 and a second stop 123 arranged adjacently. During assembly, a portion of the positioning pins 161 abut against the first stop 122, and another portion of the positioning pins 161 abut against the second stop 123. For example, the positioning pins 161 are mounted on the frame 110, the base is rectangular, the first stop 122 and the second stop 123 form a right angle, and both the first stop 122 and the second stop 123 are planar. When the working module is assembled, the first stop 122 of the base abuts against a portion of the positioning pins 161, and the second stop 123 of the base abuts against the remaining portion of the positioning pins 161. This indicates that the working module is in position, and then it only needs to be fixed by bolts or magnetic attraction. Similarly, a positioning pin 161 is also provided on the main body of the robot arm for positioning during the replacement of the pick-up component. The positioning pin 161 engages with the base of the working module or the stop on the pick-up component to quickly determine the installation position of the module on the frame 110 and the position of the pick-up component on the main body of the robot arm. This physical positioning method avoids errors that may occur with manual alignment, ensuring the accurate installation positions of the working module and the pick-up component. Traditional equipment typically requires recalibration when changing modules, which is not only time-consuming but may also lead to accuracy issues. The use of the positioning pin 161 allows for rapid positioning of the working module and the pick-up component during installation, reducing calibration time and improving assembly efficiency.

[0051] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A flexible intelligent manufacturing system based on AI intelligent control, characterized in that, include: The rack is equipped with a standard interface group; The first working module is provided with a module interface group. The standard interface group cooperates with the module interface group to allow the first working module to be detachably installed on the frame. The module interface group and the standard interface group have the same mating size and connection form to realize the interchangeability of the first working modules of the same or different types. The first working module includes one of the following: a feeding module, a loading module, a unloading module, and a process function module. The second working module is fixedly installed on the frame. The second working module includes one of the following: a feeding module, a loading module, a unloading module, and a process function module. The first working module and the second working module are different types of working modules. A position detection unit is used to collect real-time position data of the first working module and the second working module. The control unit is electrically connected to the position detection unit and is configured to: obtain the offset change vector within a first unit time based on the real-time position data, and predict the predicted offset at a first time node based on the offset change vector; if the predicted offset exceeds a first preset threshold, perform preventive operations on the first working module and the second working module. The control unit is further configured to: parse the process path, generate an association map, obtain the current working module and adjacent working modules according to the association map, calculate the relative offset change vector between the two working modules, and predict the predicted relative offset at the second time node according to the relative offset change vector; if the predicted relative offset exceeds the second preset threshold, perform preventive operations on each working module.

2. The flexible intelligent manufacturing system based on AI intelligent control according to claim 1, characterized in that, The second working module is configured as a feeding module, and the first working module is configured as a process function module, a loading module, or a unloading module; the second working module is configured as a loading module, and the first working module is configured as a process function module or an unloading module; the second working module is configured as an unloading module, and the first working module is configured as a process function module.

3. The flexible intelligent manufacturing system based on AI intelligent control according to claim 1, characterized in that, If the predicted offset exceeds a first preset threshold, the preventive operation for the first working module and the second working module includes: If the predicted offset exceeds the first preset threshold within the first time period, the working power of the first working module and the second working module shall be reduced. If the predicted offset exceeds the first preset threshold during the second time period, a shutdown warning is triggered during the third time period. After the third time period, the first working module and the second working module are controlled to stop working.

4. The flexible intelligent manufacturing system based on AI intelligent control according to claim 3, characterized in that, The position detection unit includes a vision camera and a laser rangefinder, which are electrically connected to the control unit. The vision camera is mounted above the frame, and the laser rangefinder is mounted beside each working module. The real-time position data acquired by the vision camera is denoted as A, and the real-time position data acquired by the laser rangefinder is denoted as B. Then, the offset change vector C = αA + (1-α)B, where the value of the data weight α can be dynamically adjusted according to changes in operating conditions.

5. The flexible intelligent manufacturing system based on AI intelligent control according to claim 1, characterized in that, The first preset threshold value varies depending on the type of the working module. The first preset threshold value corresponding to the process function module is smaller than the first preset threshold value corresponding to other types of working modules.

6. The flexible intelligent manufacturing system based on AI intelligent control according to claim 1, characterized in that, The control unit parses the process path and generates a correlation map. If the predicted offset exceeds the first preset threshold, preventive operations are performed on the corresponding working module and its upstream and downstream working modules.

7. The flexible intelligent manufacturing system based on AI intelligent control according to claim 1, characterized in that, The standard interface group includes an array of mounting holes, which are arranged in an array. The module interface group includes a group of through holes, which includes a plurality of through holes that mate with the mounting holes. The mounting holes and the through holes are connected by bolts.

8. The flexible intelligent manufacturing system based on AI intelligent control according to claim 7, characterized in that, The mounting holes in the mounting hole array are evenly distributed along mutually perpendicular X-axis and Y-axis directions at a first preset spacing to form a gridded positioning reference. The spacing between any two adjacent through holes is a second preset spacing, which is an integer multiple of the first preset spacing.

9. The flexible intelligent manufacturing system based on AI intelligent control according to claim 1, characterized in that, The AI-based flexible intelligent manufacturing system also includes a positioning mechanism, which includes multiple positioning pins. Each working module includes a base, which includes a first stop and a second stop arranged adjacent to each other. During assembly, a portion of the positioning pins abut against the first stop and another portion of the positioning pins abut against the second stop.

Images