Cooperative control system of intelligent numerical control machine manufacturing equipment

Abstract

The invention discloses an intelligent numerical control machine manufacturing equipment cooperative control system, and particularly relates to the field of industrial automation control, which comprises a system architecture module, a cooperative control algorithm module, a real-time communication optimization module, a predictive maintenance module, a man-machine cooperative interaction module and a plurality of numerical control equipment terminals, the central control module uniformly receives task instructions, performs dynamic decomposition and distribution, issues cooperative control instructions to each terminal device, integrates data acquisition, cloud collaboration, adaptive optimization and safety protection functions, monitors the device state in real time, optimizes the processing path, predicts faults, and improves the processing efficiency. The problems of low multi-device cooperation efficiency, poor heterogeneous device compatibility and inflexible resource scheduling are effectively solved, modular design is adopted, a good man-machine interaction interface is provided, system function extension and customization are supported, and the overall production efficiency, the machining precision and the system reliability of an intelligent manufacturing production line are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of industrial automation control technology, and more specifically, to a collaborative control system for intelligent numerical control mechanical manufacturing equipment. Background Technology

[0002] With the deepening of Industry 4.0 and intelligent manufacturing, CNC mechanical manufacturing equipment has become the core equipment of modern factories. In order to achieve more efficient and precise production goals, multiple CNC equipment are interconnected and coordinated for control, forming the basis of flexible manufacturing units and intelligent production lines. Therefore, research on intelligent CNC mechanical manufacturing equipment collaborative control systems is of great significance to improving the overall level of the manufacturing industry.

[0003] Currently, most existing equipment collaborative control schemes adopt a centralized control architecture, typically consisting of a master computer connected to several CNC devices via an industrial bus (or industrial Ethernet). The master computer is responsible for receiving tasks from the upper-level production management system, parsing them into a sequence of control instructions for each device, and then distributing them to the subordinate devices for execution in a predetermined order. The collaborative logic is usually pre-programmed, and there is a lack of real-time information interaction and adaptive adjustment capabilities between devices.

[0004] However, in actual use, it still has some drawbacks, such as the centralized architecture leading to poor system fault tolerance. Once the main controller fails, the entire production line will be paralyzed. The system lacks intelligent scheduling capabilities and cannot dynamically adjust task allocation according to the real-time status of the equipment (such as load and fault warning). The equipment utilization rate is low. The existing system is not flexible enough and it is difficult to be compatible with heterogeneous equipment of different brands and protocols. Adding new equipment requires a lot of complex reprogramming and debugging. The system has poor scalability and adaptability. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present invention provide a collaborative control system for intelligent CNC mechanical manufacturing equipment, which solves the problems mentioned in the background art through the following solutions.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a collaborative control system for intelligent CNC mechanical manufacturing equipment, comprising a system architecture module: through a central control unit coordinating the communication, data processing, decision execution, and feedback of multiple CNC devices, to achieve efficient collaborative control between devices; Collaborative control algorithm module: Based on deep reinforcement learning, a collaborative control algorithm is constructed to build a multi-device state space model, analyze the workload, machining accuracy and energy consumption of each CNC device in real time, dynamically allocate machining tasks and optimize the collaboration timing between devices; Real-time communication optimization module: It adopts a hybrid communication architecture that combines the Time-Sensitive Networking (TSN) protocol with 5G communication technology to establish a deterministic low-latency communication channel. Through dynamic bandwidth allocation mechanism and priority queue management, it controls the transmission latency of instructions and transmitted data, while supporting redundant backup and fault self-recovery of communication links. Predictive maintenance module: Establishes degradation models for specified components of each CNC machine, combines historical operating data and real-time monitoring parameters to achieve early warning of equipment failure, and automatically adjusts equipment operating parameters based on prediction results to reduce unplanned downtime; Human-machine collaborative interaction module: Through visualization technology, the system displays the collaborative working status, processing progress and abnormal information of multiple devices in real time. Operators can intervene and adjust the system through gesture recognition and voice commands. The system provides intelligent operation suggestions and risk warnings based on the operator's behavior patterns and historical decision data.

[0007] Preferably, the data collected in the system architecture module includes basic equipment information, processing task information and real-time operating status information. The central control unit adopts an industrial-grade dual-redundant server with a computing speed of no less than 200 TFLOPS and supports automatic switching between dual power supplies.

[0008] Preferably, the workload selection parameter is the current load rate of the device. For equipment number, , Total number of devices, number of tasks to be processed Remaining time for the current processing task of the equipment The machining accuracy selection parameter is the dimensional deviation of the workpieces processed by the equipment within 3 months. , Number the workpiece. , The number of workpieces processed within 3 months, surface roughness Machining accuracy pass rate The energy consumption status selection parameter is the power consumption of the equipment per unit time. Energy consumption per unit of workpiece processed Equipment standby power consumption .

[0009] Preferably, the collaborative control algorithm adopts the improved Deep Deterministic Policy Gradient (DDPG) algorithm, which introduces a priority experience replay mechanism and a dual Critic network structure. The algorithm training constructs a reward function based on "processing efficiency, processing accuracy pass rate, and total energy consumption".

[0010] Preferably, in the hybrid communication architecture, the TSN protocol is responsible for device-level communication within the workshop, achieved through a time synchronization mechanism. With precise time synchronization, 5G communication employs independent network slicing technology, allocating dedicated bandwidth for remote monitoring and cross-workshop collaboration.

[0011] Preferably, the dynamic bandwidth allocation mechanism divides data into three priority levels: control commands > precision detection data > log data. When high-priority data traffic exceeds the threshold, it preempts low-priority bandwidth. The communication link adopts a dual-redundancy design of TSN main link and 5G backup link, and the switching response time is ≤100ms when the link fails.

[0012] Preferably, the predictive maintenance module constructs a "two-factor degradation model" for spindle, cutting tool, and ball screw components. Combining three years of historical fault data with real-time monitoring parameters, the module calculates the real-time degradation level through the degradation model, sets a three-level early warning mechanism, and determines the early warning level based on the relationship between the real-time degradation level and the threshold.

[0013] Preferably, the visualization system of the human-machine collaborative interaction module supports 2D / 3D view switching, uses color gradients to mark the equipment status (green for normal, yellow for warning, red for fault), and dynamically displays the workpiece flow trajectory and processing progress. The multimodal interaction uses a TOF depth camera and a noise-canceling microphone, and supports 5 types of customized gestures and 10 languages ​​of voice commands.

[0014] The technical effects and advantages of this invention are as follows: This invention improves the reliability and fault tolerance of the system through system architecture and collaborative control. The central control module and each equipment terminal have a certain degree of independent operation and decision-making capabilities. When a single node fails, the system can continue to operate through task redistribution or degradation mode, avoiding the problem of the entire line being paralyzed due to the failure of the main controller in the existing centralized architecture, and ensuring the continuity and stability of production. This invention continuously monitors the operating status, processing progress, and load of each device by real-time monitoring parameters, and uses built-in intelligent algorithms for real-time analysis and decision-making, thereby dynamically adjusting task allocation and processing parameters, achieving equipment load balancing and production process optimization, and overcoming the shortcomings of rigid scheduling and low equipment utilization in existing technologies. This invention expands the system's scalability and compatibility through human-machine collaborative interaction, defines a unified interface protocol and data interaction standard, and integrates heterogeneous CNC equipment of different brands and models. When adding new equipment, only protocol adaptation is required without reconstructing the entire system. This design effectively solves the problems of poor scalability and integration difficulties in existing systems, and meets the needs of modern smart factories for flexible production and rapid upgrades. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the overall structure of the present invention; Figure 2This is a schematic diagram of the intelligent dynamic task allocation and timing optimization of the present invention; Figure 3 This is a schematic diagram of the predictive maintenance process of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] As attached Figure 1 Appendix Figure 2 and appendix Figure 3 The intelligent CNC mechanical manufacturing equipment collaborative control system shown includes a system architecture module: through a central control unit, the communication, data processing, decision execution and feedback of multiple CNC devices are coordinated to achieve efficient collaborative control between devices.

[0018] It should be noted that the data collected in the system architecture module includes basic equipment information, processing task information and real-time operating status information. The central control unit adopts an industrial-grade dual-redundant server with a computing speed of no less than 200 TFLOPS and supports automatic switching between dual power supplies.

[0019] It should be further explained that the basic equipment information includes the model of each CNC machine, factory parameters (maximum spindle speed, maximum machining diameter, positioning accuracy), installation location, machining process, and historical maintenance records. The reason for selecting this type of data is that the basic equipment information is a prerequisite for realizing multi-machine collaborative control. The central control unit determines the machining capability by the model and factory parameters of the equipment (for example, a CNC milling machine with a positioning accuracy of ±0.005mm is suitable for milling high-precision parts). The installation location of the equipment optimizes the workpiece transfer path and reduces the transfer time of the workpiece between machines. The machining process and historical maintenance records provide basic data to help analyze the failure patterns of vulnerable parts of the equipment.

[0020] The processing task information includes the model, quantity, 3D model data, and process requirements (such as processing accuracy level, surface roughness, material properties, and processing procedure flow) of the workpiece to be processed. The reason for selecting this type of data is that the processing task information directly determines the direction of task allocation. The central control unit obtains the quantity requirements, external dimensions, and key features (such as holes, slots, and threads) corresponding to different models of equipment by parsing the equipment model, quantity, and 3D model data. Then, it determines the processing steps that each equipment needs to complete. The processing accuracy level and surface roughness data in the process requirements are used to match equipment with corresponding processing capabilities to avoid substandard processing quality due to insufficient equipment accuracy. Material properties affect the setting of processing parameters (for example, a higher cutting speed needs to be selected when processing 45 steel, while the cutting speed needs to be appropriately reduced when processing aluminum alloy to prevent material from sticking to the tool).

[0021] Real-time operating status information includes the spindle speed, feed rate, depth of cut, equipment load rate, component temperature (spindle bearing temperature, motor temperature), tool wear, machining task progress, and equipment fault alarm signals (overload alarm, over-temperature alarm) of each CNC machine. The reason for selecting this type of data is that real-time operating status information can reflect the current working condition of the equipment. The central control unit monitors the spindle speed, feed rate, and depth of cut to determine whether the equipment is operating according to the preset process requirements. Equipment load rate data is used to avoid overload operation of the equipment. When the load rate of a certain machine exceeds 80%, its machining task needs to be adjusted in time to prevent equipment damage. Component temperature and tool wear data are important bases for predicting equipment failures (for example, when the spindle bearing temperature exceeds 80°C, it may indicate that the bearing is worn and needs to be maintained in time). Machining task progress provides support for dynamic adjustment of task allocation to ensure that the machining tasks of each machine can be completed synchronously.

[0022] Collaborative control algorithm module: Based on deep reinforcement learning, a collaborative control algorithm is constructed to build a multi-device state space model, analyze the workload, machining accuracy and energy consumption of each CNC device in real time, dynamically allocate machining tasks and optimize the collaboration timing between devices.

[0023] It should be noted that the workload selection parameter is the current load rate of the device. For equipment number, , Total number of devices, number of tasks to be processed Remaining time for the current processing task of the equipment The machining accuracy selection parameter is the dimensional deviation of the workpieces processed by the equipment within 3 months. , Number the workpiece. , The number of workpieces processed within 3 months, surface roughness Machining accuracy pass rate The energy consumption status selection parameter is the power consumption of the equipment per unit time. Energy consumption per unit of workpiece processed Equipment standby power consumption .

[0024] The cooperative control algorithm adopts the improved deep deterministic policy gradient (DDPG) algorithm, introduces a priority experience replay mechanism and a dual Critic network structure, and constructs a reward function based on "processing efficiency, processing accuracy pass rate, and total energy consumption" during algorithm training.

[0025] Further explanation is needed regarding the selection of the Deep Deterministic Policy Gradient (DDPG) algorithm (adapted to continuous action space decision-making), with two improvements: First, a priority experience replay mechanism is introduced, prioritizing samples based on their TD error (higher error, higher priority) to improve sample utilization. Second, a dual-Critic network structure is employed: action values ​​are calculated through two isomorphic, heterogeneous Critic networks, with the minimum value taken as the final value, mitigating the overestimation problem of traditional DDPG. Third, 12 months of historical production line operation data are collected, covering normal production, faults, and peak / valley order conditions, and the data is processed through outlier removal, linear interpolation to fill missing values, and standardization mapping. [0,1]” preprocessing ensures training stability. The network parameters Actor / Critic are both 3-layer fully connected networks. For Actor: input layer nodes = total number of state space parameters, hidden layers (512→256), output layer nodes = number of devices (task allocation variable), activation function is ReLU (hidden layer) + Sigmoid (output layer). For Critic: input layer nodes = total number of state parameters + total number of action parameters, hidden layers are the same as Actor, output layer nodes = 1 (action value), activation function is ReLU (hidden layer) + linear function (output layer), learning rate 0.001, experience replay buffer 1×10. 6 The batch sample size is 256, the discount factor is 0.95, and the soft update coefficient is 0.001. The training process is based on the PyTorch framework, and the iteration cycle includes "exploration (Actor output with Gaussian noise, standard deviation decreasing from 0.1 to 0.01) - update (batch sample training updates Critic / Actor, soft update target network)". The reward function is constructed based on "processing efficiency + accuracy pass rate - energy consumption". ( =Processing efficiency, =Pass rate =Total energy consumption), achieving multi-objective optimization, convergence condition: 1×10 iterations. 5 If the reward function value fluctuates by less than 5% over 1000 consecutive iterations, stop training and save the model.

[0026] The task allocation process is as follows: When a new processing task enters the production line, the collaborative control algorithm module first obtains the current status data (workload, processing accuracy, energy consumption status) of each device and the parameters of the new task (processing difficulty coefficient, required time, priority) from the system architecture module. This data is then input into the Actor network of the trained DDPG algorithm. The Actor network outputs the task allocation ratio for each device (for example, in a scenario with 10 devices and 50 tasks to be processed, the Actor network may output an allocation scheme where device 1 is allocated 5 tasks, device 2 is allocated 6 tasks, and device 10 is allocated 4 tasks). Based on the task priority and processing time required, the tasks allocated to each device are sorted, with higher priority and shorter processing time tasks placed first to ensure that urgent tasks are processed first. At the same time, the feasibility of the task allocation scheme is verified by checking whether the total number of tasks allocated to each device exceeds its maximum processing capacity (the maximum processing capacity is determined based on the device's workload, processing accuracy, and energy consumption status). If it exceeds the maximum processing capacity, the excess tasks are allocated to devices with lower load rates by adjusting the task allocation ratio to ensure the feasibility of the task allocation scheme.

[0027] During task execution, the collaborative control algorithm module acquires real-time processing progress data of each device (e.g., device 1 completes 70% of the current task, device 2 completes 50%), and optimizes the workpiece flow sequence between devices by constructing a "processing cycle synchronization coefficient" model. The processing cycle synchronization coefficient K is defined as the ratio of the actual processing cycles of adjacent devices. , For the processing cycle of the preceding equipment, (for the processing cycle of subsequent equipment), when When the value exceeds the preset reasonable range of [0.9, 1.1], a dynamic timing compensation mechanism is triggered (for example, when the processing cycle of the CNC lathe (preceding equipment) in the production line is exceeded). =2min, CNC milling machine (subsequent equipment) processing cycle =2.3min, calculated =0.87<0.9, at this time the system automatically starts compensation measures: on the one hand, it sends a parameter adjustment command to the PLC of the CNC milling machine, increasing its feed speed from 800mm / min to 850mm / min, shortening the feed rate. To 2.1 minutes, on the other hand, the conveyor robotic arm connecting the two devices was adjusted to change the gripping interval from 2 minutes / time to 2.1 minutes / time to avoid waiting for subsequent equipment or workpiece accumulation in the preceding equipment.

[0028] Real-time communication optimization module: It adopts a hybrid communication architecture that combines the Time-Sensitive Networking (TSN) protocol with 5G communication technology to establish a deterministic low-latency communication channel. Through dynamic bandwidth allocation mechanism and priority queue management, it controls the transmission delay of instructions and data, while supporting redundant backup and fault self-recovery of communication links.

[0029] It should be specifically noted that in the hybrid communication architecture, the TSN protocol is responsible for device-level communication within the workshop, which is achieved through a time synchronization mechanism. With precise time synchronization, 5G communication employs independent network slicing technology, allocating dedicated bandwidth for remote monitoring and cross-workshop collaboration.

[0030] The dynamic bandwidth allocation mechanism divides data into three priority levels: control commands > precision detection data > log data. When high-priority data traffic exceeds the threshold, it preempts low-priority bandwidth. The communication link adopts a dual redundancy design of TSN main link and 5G backup link, and the switching response time is ≤100ms when the link fails.

[0031] It should be further explained that, based on the impact of the data on the production line operation, the transmitted data is divided into three categories. The specific content, transmission requirements, and priorities of each category are as follows: Control command data includes spindle speed adjustment commands, feed rate correction commands, workpiece transfer start / stop commands, and emergency stop commands, with a delay of ≤10ms and a packet loss rate of ≤10ms. Priority is P1, the highest level. Key status data include spindle bearing temperature, tool wear, machining dimensional deviation, equipment load rate, and fault warning signals. Delay ≤ 50ms, packet loss rate ≤ Priority is P2. Non-critical data includes equipment historical maintenance records, processing task completion statistics, and workshop environmental temperature and humidity. Latency ≤ 1000ms, packet loss rate ≤ Its priority is P3, the lowest level.

[0032] The dynamic bandwidth allocation mechanism is based on data priority and real-time transmission requirements, and adopts a "bandwidth pre-allocation + dynamic adjustment" strategy. The system pre-allocates 40% of the fixed bandwidth for P1 data, while P2 and P3 data share the remaining 60% bandwidth. When the P2 data traffic exceeds the preset threshold, the transmission bandwidth of P3 data is automatically compressed (with a minimum of 10% reserved) to prioritize the needs of P2 data (for example, when a device suddenly experiences a fault warning (P2 data), the bandwidth allocation ratio is automatically adjusted to P1:40%, P2:50%, and P3:10%).

[0033] Priority queue management constructs a three-level priority queue in the TSN switch and 5G module. P1 type data enters the high priority queue, and the switch prioritizes forwarding high priority queue data until the queue is empty before processing low priority data. P2 type and P3 type data enter the medium and low priority queues respectively, with a weight ratio of 3:1, to avoid low priority data being blocked for a long time. Link redundancy and fault self-recovery are implemented through real-time monitoring of the connectivity status of the TSN wired link (sending a "heartbeat packet" for detection every 20ms). If no heartbeat response is received for three consecutive times, a link fault is determined, and the 5G wireless link is immediately switched over. The switching process takes ≤10ms, ensuring uninterrupted data transmission. At the same time, the IEEE 802.1CB protocol is used to implement frame duplication for P1 type data. That is, the central control unit sends the same control command frame to both the TSN and 5G links simultaneously. The receiving end (numerical control equipment) automatically eliminates duplicate frames and retains only the first arriving frame, further improving the reliability of data transmission.

[0034] Predictive maintenance module: Establishes degradation models for specified components of each CNC machine, combines historical operating data and real-time monitoring parameters to achieve early warning of equipment failures, and automatically adjusts equipment operating parameters based on prediction results to reduce unplanned downtime.

[0035] It should be specifically noted that the predictive maintenance module constructs a "two-factor degradation model" for spindle, cutting tool, and ball screw components. Combining three years of historical fault data with real-time monitoring parameters, the degradation model calculates the real-time degradation level and sets a three-level early warning mechanism. The early warning level is determined based on the relationship between the real-time degradation level and the threshold.

[0036] It should be further explained that, for easily worn components such as spindles, cutting tools, and ball screws in CNC equipment, a "two-factor degradation model" is constructed through experimental testing and data fitting to achieve continuous quantitative assessment of the degradation degree of these components. Taking the cutting tool, which has the highest application frequency, as an example, its degradation process is affected by both cutting force and operating temperature. The model expression is set as follows: ,in, represent The degree of tool degradation at any given time (range 0-1, 0 for brand new, 1 for complete failure) is a cutting force influence coefficient, derived through fitting of tool life experiments under 500 sets of different materials and cutting parameters. The value used in practical applications is [value missing]. , for Real-time cutting force data directly reflects the cutting load on the tool. The temperature influence coefficient is obtained through experimental fitting. ; for The average operating temperature of the tool at any given time is used to quantify the impact of heat loss on tool life.

[0037] Collect component failure records within 3 years of equipment commissioning, and filter out the cutting force at the time of failure ( ), operating temperature ( ) and the degree of degradation at the corresponding time ( A fault feature database was established, and real-time data was collected using piezoelectric force sensors and PT100 temperature sensors installed on the equipment. and Due to the complex electromagnetic environment in the workshop, the raw data is easily interfered with and needs to be denoised using the Kalman filter algorithm. The formula is as follows: In the formula, for The filtered parameter values ​​at each time step. This is the state transition matrix (set to 1.02 based on the device's operational stability). The Kalman gain is calculated from the variance of historical data. for The original sampled value at time 10:00. For the observation matrix (set to 1 to ensure the dimensions of the observations and state values ​​are consistent), the filtered real-time... and The degradation model constructed by substituting the data is used to calculate the current time. The value is then compared with the preset "early warning baseline value" in the fault feature database. "Compare the components to determine whether they have entered the degradation warning range."

[0038] Based on data fusion The system is configured with a three-level early warning mechanism to achieve a tiered response from "alert" to "restriction" to "emergency intervention," thus preventing sudden malfunctions. Level 1 warning At this point, the component is in a state of mild degradation. The system automatically pushes a "Component Degradation Reminder" to the maintenance personnel's terminal via the industrial APP, and simultaneously calls the degradation model to calculate the remaining useful life (RUL). The formula is as follows: ,in, Set as the component failure threshold (set to 1). The current degradation rate is derived from the degradation model by taking the derivative. (Rate of temperature change). Level II warning ( When a component enters the moderate degradation stage, the system will link with the production scheduling module to mark the equipment as "to be maintained" and automatically limit high-load processing tasks (such as processes with cutting forces > 400N) during task allocation to prevent accelerated degradation. Level III warning ( When a component is close to failure, the system immediately triggers an audible and visual alarm in the workshop and automatically generates a maintenance work order. The work order clearly indicates the model of the faulty component (e.g., tool model CCMT09T304), the required maintenance tools (e.g., tool chuck wrench), and the replacement steps, guiding maintenance personnel to handle the issue quickly.

[0039] To extend component lifespan and ensure production continuity during early warning periods, when a Level 1 or higher early warning is triggered, the system will, based on the current... The automatic correction of equipment operating parameters, taking the adjustment of spindle and feed speed as an example, uses the following adjustment formula: In the formula, The adjusted spindle speed, The original rotational speed before adjustment is given, and 0.2 is the adjustment coefficient. The feed rate is... Adjust the speed proportionally to the spindle speed and reduce it accordingly. This avoids machining defects caused by a mismatch between the rotation speed and the feed rate.

[0040] Human-machine collaborative interaction module: Through visualization technology, the system displays the collaborative working status, processing progress and abnormal information of multiple devices in real time. Operators can intervene and adjust the system through gesture recognition and voice commands. The system provides intelligent operation suggestions and risk warnings based on the operator's behavior patterns and historical decision data.

[0041] It should be noted that the visualization system of the human-machine collaborative interaction module supports 2D / 3D view switching, uses color gradients to mark the equipment status (green for normal, yellow for warning, red for fault), and dynamically displays the workpiece flow trajectory and processing progress. The multimodal interaction uses a TOF depth camera and a noise-canceling microphone, and supports 5 types of customized gestures and 10 languages ​​of voice commands.

[0042] It should be further explained that operators can use AR glasses and visualization devices to view the real-time working status, processing progress, and fault prompts of each piece of equipment. It supports one-click access to operation manuals and maintenance guides, integrates a voice recognition engine, and supports natural language command control of equipment start / stop, parameter adjustment, program calls, and other operations. It also provides operation feedback and alarm prompts through voice synthesis technology. The system recognizes operator gestures through cameras or depth sensors to achieve contactless control, while monitoring the safe distance between personnel and equipment. When personnel enter a dangerous area, the system automatically slows down or stops the machine. It supports multi-person, multi-role collaborative operation, automatically assigns task permissions, and records operation logs to ensure the operation process is traceable and auditable. Based on the operator's role, experience level, and current task type, the system dynamically adjusts the complexity of the interactive interface and the information display method to improve operational efficiency and user experience.

[0043] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other. In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A collaborative control system for intelligent CNC mechanical manufacturing equipment, characterized in that, include: System architecture module: Through the central control unit, the communication, data processing, decision execution and feedback of multiple CNC machines are coordinated to achieve efficient collaborative control between the machines; Collaborative control algorithm module: Based on deep reinforcement learning, a collaborative control algorithm is constructed to build a multi-device state space model, analyze the workload, machining accuracy and energy consumption of each CNC device in real time, dynamically allocate machining tasks and optimize the collaboration timing between devices; Real-time communication optimization module: It adopts a hybrid communication architecture that combines the Time-Sensitive Networking (TSN) protocol with 5G communication technology to establish a deterministic low-latency communication channel. Through dynamic bandwidth allocation mechanism and priority queue management, it controls the transmission latency of instructions and transmitted data, while supporting redundant backup and fault self-recovery of communication links. Predictive maintenance module: Establishes degradation models for specified components of each CNC machine, combines historical operating data and real-time monitoring parameters to achieve early warning of equipment failure, and automatically adjusts equipment operating parameters based on prediction results to reduce unplanned downtime; Human-machine collaborative interaction module: Through visualization technology, the system displays the collaborative working status, processing progress and abnormal information of multiple devices in real time. Operators can intervene and adjust the system through gesture recognition and voice commands. The system provides intelligent operation suggestions and risk warnings based on the operator's behavior patterns and historical decision data.

2. The intelligent CNC mechanical manufacturing equipment collaborative control system according to claim 1, characterized in that: The data collected in the system architecture module includes basic equipment information, processing task information, and real-time operating status information. The central control unit adopts an industrial-grade dual-redundant server with a computing speed of no less than 200 TFLOPS and supports automatic switching between dual power supplies.

3. The intelligent CNC mechanical manufacturing equipment collaborative control system according to claim 1, characterized in that: The workload selection parameter is the current load rate of the device. For equipment number, , Total number of devices, number of tasks to be processed Remaining time for the current processing task of the equipment The machining accuracy selection parameter is the dimensional deviation of the workpieces processed by the equipment within 3 months. , Number the workpiece. , The number of workpieces processed within 3 months, surface roughness Machining accuracy pass rate The energy consumption status selection parameter is the power consumption of the equipment per unit time. Energy consumption per unit of workpiece processed Equipment standby power consumption .

4. The intelligent CNC mechanical manufacturing equipment collaborative control system according to claim 1, characterized in that: The cooperative control algorithm adopts the improved deep deterministic policy gradient (DDPG) algorithm, which introduces a priority experience replay mechanism and a dual Critic network structure. The algorithm training constructs a reward function based on "processing efficiency, processing accuracy pass rate, and total energy consumption".

5. The intelligent CNC mechanical manufacturing equipment collaborative control system according to claim 1, characterized in that: In the hybrid communication architecture, the TSN protocol is responsible for device-level communication within the workshop, which is achieved through a time synchronization mechanism. With precise time synchronization, 5G communication employs independent network slicing technology, allocating dedicated bandwidth for remote monitoring and cross-workshop collaboration.

6. The intelligent CNC mechanical manufacturing equipment collaborative control system according to claim 1, characterized in that: The dynamic bandwidth allocation mechanism divides data into three priority levels: control commands > precision detection data > log data. When high-priority data traffic exceeds the threshold, it preempts low-priority bandwidth. The communication link adopts a dual redundancy design of TSN main link and 5G backup link, and the switching response time is ≤100ms when the link fails.

7. The intelligent CNC mechanical manufacturing equipment collaborative control system according to claim 1, characterized in that: The predictive maintenance module constructs a "two-factor degradation model" for spindle, cutting tool, and ball screw components. Combining three years of historical fault data with real-time monitoring parameters, it calculates the real-time degradation level through the degradation model, sets a three-level early warning mechanism, and determines the early warning level based on the relationship between the real-time degradation level and the threshold.

8. The intelligent CNC mechanical manufacturing equipment collaborative control system according to claim 1, characterized in that: The visualization system of the human-machine collaborative interaction module supports 2D / 3D view switching, uses color gradients to mark the equipment status (green for normal, yellow for warning, red for fault), and dynamically displays the workpiece flow trajectory and processing progress. The multimodal interaction uses a TOF depth camera and noise-canceling microphone, and supports 5 types of customized gestures and 10 languages ​​of voice commands.

Images