A flexible intelligent processing production line multi-serial machine cooperative scheduling method and system

Abstract

The present application relates to the technical field of workshop scheduling, in particular to a flexible intelligent processing production line multi-association machine cooperative scheduling method and system, comprising: establishing a numerical control parameter model for processing equipment, decomposing a processing program into basic process instruction units, analyzing the adaptability of the basic process instruction units and the equipment in combination with the numerical control parameter model and the real-time running state of the equipment, and generating a preliminary task allocation scheme; constructing a distributed control network among each device, generating a local task sequence for each device based on the preliminary task allocation scheme and the adaptability, and obtaining a final task allocation scheme and a task execution plan set based on a contract net protocol algorithm; establishing a multi-constraint cooperative framework, generating a cooperative production scheduling scheme under the multi-constraint cooperative framework; and establishing a heterogeneous device motion cooperative control model, and performing multi-level adjustment through event-driven feedback control. The present application can realize flexible cooperative scheduling of heterogeneous devices.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of workshop scheduling, in particular to a flexible intelligent machining production line multi-association machine collaborative scheduling method and system. BACKGROUND

[0002] With the rapid development of intelligent manufacturing, flexible intelligent machining production lines are increasingly widely used in high-end manufacturing fields such as automobile parts, aerospace, wind power equipment, and medical devices. Flexible intelligent machining production lines usually include different types of machining equipment such as lathes, grinding machines, and machining centers, as well as auxiliary equipment such as truss manipulators and joint robots. These devices have different working characteristics, processing capabilities, and control systems, forming a typical heterogeneous environment.

[0003] In the prior art, most control methods either optimize for a single type of device or simplify heterogeneous devices, which cannot fully consider the differences in characteristics of various devices and is difficult to achieve truly efficient collaborative work. In particular, there are obvious deficiencies in the following aspects: first, there is a lack of unified representation of heterogeneous device capabilities, which leads to an inability to accurately assess the matching degree of tasks and devices; second, in the traditional central control mode, the system has poor flexibility and limited adaptability to dynamically changing production environments; and finally, there is insufficient temporal and spatial coordination between devices, leading to action conflicts and resource waste.

[0004] Therefore, a flexible intelligent machining production line multi-association machine collaborative scheduling method and system are proposed. SUMMARY

[0005] The present application aims to provide a flexible intelligent machining production line multi-association machine collaborative scheduling method and system that achieves flexible collaborative control of heterogeneous devices. This includes establishing a numerical control parameter model for machining equipment, decomposing machining programs into basic process instruction units, analyzing the adaptability of basic process instruction units to equipment based on the numerical control parameter model and real-time running state of the equipment, and generating a preliminary task allocation scheme; building a distributed control network between each device, generating a local task sequence for each device based on the preliminary task allocation scheme and adaptability, and obtaining a final task allocation scheme and task execution plan set based on a contract net protocol algorithm; establishing a multi-constraint collaborative framework to generate a collaborative production scheduling scheme under the multi-constraint collaborative framework; establishing a heterogeneous device motion collaborative control model for time axis control and dynamic safety area management, and performing multi-level adjustment through event-driven feedback control.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0007] A flexible intelligent machining production line multi-association machine collaborative scheduling method, comprising:

[0008] A numerical control parameter model is established for the machining equipment, including geometric capability, precision parameter, dynamic performance, process capability and control characteristic, real-time running state and machining result data of the equipment are collected to update the numerical control parameter model;

[0009] The machining program is decomposed into basic process instruction units, the numerical control parameter model and the real-time running state of the equipment are combined to analyze the adaptability of the basic process instruction units and the equipment, and a preliminary task allocation scheme is generated;

[0010] A distributed control network is constructed among the various equipment, a task evaluation function is constructed for each equipment based on the preliminary task allocation scheme and the adaptability, a local task sequence of each equipment is generated according to global constraint conditions, and a final task allocation scheme and a task execution plan set are obtained based on a contract net protocol algorithm;

[0011] A collaborative production scheduling scheme is generated according to the final task allocation scheme and the task execution plan set under the multi-constraint collaborative framework;

[0012] A heterogeneous equipment motion collaborative control model is established, predictive time axis control and dynamic safety area management based on forward-looking motion envelope are performed, and multi-level adjustment is performed through event-driven feedback control, including bottom layer action parameter adjustment and upper layer model parameter correction.

[0013] Preferably, the numerical control parameter model is established by: obtaining equipment basic parameter data, historical machining data and process specification requirements, extracting equipment geometric capability, precision parameter, dynamic performance, process capability and control characteristic, and integrating them into a numerical control parameter model by using a multi-dimensional tensor representation method; the geometric capability includes workspace, travel range and maximum load; the dynamic performance includes maximum speed, acceleration, deceleration characteristic and emergency stop distance; the precision parameter includes positioning accuracy, repeat positioning accuracy and trajectory accuracy; the process capability includes executable process type, process quality index and production efficiency; the control characteristic includes instruction response delay, communication interface type and supported instruction set.

[0014] Preferably, the process of generating the preliminary task allocation scheme includes: obtaining and analyzing the machining program and product process requirements, decomposing the machining program into basic process instruction units and determining the dependency relationship between the basic process instruction units; based on the numerical control parameter model and the real-time running state of the equipment, evaluating the process matching degree, capability compliance and expected execution result of each basic process instruction unit and each machining equipment, calculating the adaptability; generating the preliminary task allocation scheme according to the adaptability and the preset allocation rule.

[0015] Preferably, the process of obtaining the final task allocation scheme and the task execution plan set specifically comprises: configuring a decision unit for each device in the distributed control network, each decision unit generating and optimizing a local task sequence under global constraints based on the preliminary task allocation scheme, the fitness and the respective task evaluation function; each decision unit declaring, bidding and negotiating according to the basic process instruction unit according to the contract net protocol algorithm; through multiple rounds of negotiation and evaluation, the final task allocation scheme of each device and the task execution plan set containing the task sequence and time arrangement are formed.

[0016] Preferably, the process of generating the collaborative production scheduling scheme specifically comprises: the multi-constraint collaborative framework is composed of process constraints, resource constraints, time constraints and space constraints; in the multi-constraint collaborative framework, the final task allocation scheme and the task execution plan set are used as inputs, and an optimization algorithm is used to solve a scheduling solution that satisfies all constraint conditions and optimizes the preset production target; the scheduling solution is converted into an executable collaborative production scheduling scheme, which clearly defines the start and end times of each task on each device and the required resources.

[0017] Preferably, the heterogeneous device motion collaborative control model specifically comprises: integrating the kinematic characteristics, action timing logic and spatial geometric information of each heterogeneous device; time axis control through the heterogeneous device motion collaborative control model includes: setting a globally unified virtual time reference, and planning the start and stop time of each device action and the synchronization node; dynamic safety area management includes: calculating and updating the safety working area and non-safety working area of each device according to the real-time motion state data of the device.

[0018] Preferably, the process of multi-level adjustment through event-driven feedback control specifically comprises: real-time monitoring of the production process, when the dynamic safety area management based on the prospective motion envelope predicts that there is a risk of motion interference in the future, triggering the bottom layer feedback control mechanism, the bottom layer feedback control mechanism adjusts the speed, acceleration and motion path of the related device in real time according to the risk level and interference type, dynamically avoiding potential collision; after task execution, the actual motion trajectory and completion time of the device collected through the sensor are compared with the predicted value of the heterogeneous device motion collaborative control model; when the deviation between the two exceeds the preset threshold, triggering the upper layer feedback control mechanism, the upper layer feedback control mechanism uses the deviation data to correct and optimize the dynamic performance parameters in the numerical control parameter model, improving the model accuracy and the accuracy of future decision-making.

[0019] A flexible intelligent processing production line multi-assembly machine collaborative scheduling system for executing a flexible intelligent processing production line multi-assembly machine collaborative scheduling method, comprising:

[0020] The device capability representation module establishes a numerical control parameter model for the processing device, including geometric capability, precision parameter, dynamic performance, process capability and control characteristic, collects device real-time running state and processing result data, and updates the numerical control parameter model.

[0021] The task decomposition matching module decomposes the processing program into basic process instruction units, analyzes the adaptability of the basic process instruction units and the device based on the numerical control parameter model and the device real-time running state, and generates a preliminary task allocation scheme.

[0022] The distributed negotiation module constructs a distributed control network among the devices, constructs a task evaluation function for each device based on the preliminary task allocation scheme and the adaptability, generates a local task sequence for each device according to global constraint conditions, and obtains a final task allocation scheme and a task execution plan set based on a contract net protocol algorithm.

[0023] The multi-constraint optimization module generates a collaborative production scheduling scheme according to the final task allocation scheme and the task execution plan set under the multi-constraint collaborative framework.

[0024] The action coordination module establishes a heterogeneous device motion coordination control model, performs time axis control and dynamic safety area management, and performs multi-level adjustment through event-driven feedback control.

[0025] Compared with the prior art, the beneficial effects of the present application are:

[0026] 1. The capability characteristics of different types of processing devices are uniformly represented as five dimensions of geometric capability, precision parameter, dynamic performance, process capability and control characteristic, and the actual processing capability of the device can also be predicted according to the real-time state of the device. The model introduces an adaptive updating mechanism, which dynamically adjusts the model parameters by collecting device running state and processing result data, compares the difference between the predicted value and the actual measured value, and forms a closed-loop optimization. This unified representation mechanism significantly improves the accuracy of heterogeneous device capability evaluation, provides a reliable basis for subsequent task allocation, and can more accurately match process requirements and device capabilities, which is the basis for efficient collaboration of heterogeneous devices.

[0027] 2. A distributed autonomous negotiation mechanism is used for task allocation, each device is given independent decision-making ability, and a negotiation system is formed by connecting the distributed control network. First, a preliminary task allocation scheme is generated and the fitness is calculated, then a decision unit is configured for each device, and a task evaluation function considering multiple objectives is constructed. Based on the state, ability and task demand of each device, task bidding and resource negotiation are carried out through the contract net protocol algorithm, and the optimal allocation of resources is realized. This distributed negotiation mode improves the adaptability and robustness of collaborative control, can flexibly adjust the task allocation according to the real-time state of the device, effectively cope with the dynamic changes of the production environment, and improve the overall efficiency of the production line, reduce the waiting time, and make the heterogeneous devices form an organic whole for collaborative work.

[0028] 3. A heterogeneous device action coordination scheduling method based on space-time coupling solves the problem of accurate coordination of heterogeneous devices at the action level. The method establishes a heterogeneous device motion coordination control model integrating kinematic characteristics, action timing logic and spatial geometric information, realizes global unified time reference through virtual time axis, and accurately controls the start and stop time of each device action and synchronization node. At the same time, based on the real-time motion state of the device, the safe working area and the non-safe working area are dynamically calculated and updated to avoid spatial interference and collision. Real-time monitoring of device state, task progress and abnormal events, when the preset conditions are triggered, feedback control is started, action parameters are adjusted and the results are fed back to the numerical control parameter model. The space-time coupled collaborative control significantly improves the accuracy and stability of the action coordination of heterogeneous devices, ensuring the continuity and safety of the production process.

[0029] 4. A multi-level, double-loop feedback control architecture is established, combining immediate physical risk avoidance with long-term model adaptive optimization. The architecture actively adjusts action parameters based on motion prediction to avoid potential collisions through the bottom-layer real-time safety loop; at the same time, through the upper-layer model cognition loop, the numerical control parameter model is corrected and optimized in reverse according to the actual execution deviation after the task is completed. This double-loop collaborative working mechanism not only ensures the physical safety during production execution, but also endows the entire system with the ability to learn from experience and continuously evolve, significantly improving the long-term stability and precision of decision-making of the system. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 A flexible intelligent processing production line multi-assembly machine collaborative scheduling method process schematic diagram of the application;

[0031] Figure 2 A process schematic diagram for obtaining the final task allocation scheme and task execution plan set of the application;

[0032] Figure 3 A dynamic safety area management process schematic diagram of the application;

[0033] Figure 4 Figure 1 is a structural schematic diagram of a flexible intelligent machining production line multi-connection machine cooperative scheduling system according to the present application. DETAILED DESCRIPTION

[0034] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0035] Referring to Figures 1 to 4 , the present application provides a flexible intelligent machining production line multi-connection machine cooperative scheduling method and system, and the technical solutions are as follows:

[0036] Embodiment one:

[0037] The present embodiment is applied to a cylinder cover production line of an automobile part manufacturing enterprise. The production line includes 5 numerical control machine tools (3 five-axis machining centers, 1 horizontal boring and milling machine and 1 precision grinding machine), 2 industrial robots and 1 automatic guided vehicle (AGV), and is mainly responsible for producing V6 engine cylinder covers. Since the cylinder cover product structure is complex, the machining process is multiple, and multiple devices need to be cooperated to complete, the traditional control method has problems such as unstable production rhythm, low equipment utilization rate, long waiting time between processes, etc. In order to realize efficient cooperative control of heterogeneous devices, a flexible intelligent machining production line multi-connection machine cooperative scheduling method is implemented, as shown in Figure 1 , which includes:

[0038] A numerical control parameter model is established for the machining device, including geometric capability, precision parameter, dynamic performance, process capability and control characteristics, real-time running state and machining result data of the device are collected, and the numerical control parameter model is updated;

[0039] The machining program is decomposed into a basic process instruction unit, the adaptability of the basic process instruction unit and the device is analyzed in combination with the numerical control parameter model and the real-time running state of the device, and a preliminary task allocation scheme is generated;

[0040] A distributed control network is constructed between each device, a task evaluation function is constructed for each device based on the preliminary task allocation scheme and the adaptability, a local task sequence of each device is generated according to the global constraint condition, and a final task allocation scheme and a task execution plan set are obtained based on the contract net protocol algorithm;

[0041] A cooperative production scheduling scheme is generated according to the final task allocation scheme and the task execution plan set under the multi-constraint cooperative framework;

[0042] Establish a motion cooperative control model for heterogeneous devices, perform predictive time-axis control and dynamic safety zone management based on forward-looking motion envelope, and conduct multi-level adjustment through event-driven feedback control, including adjustment of the underlying motion parameters and correction of the upper-level model parameters.

[0043] Furthermore, establishing a CNC parameter model specifically includes: acquiring basic equipment parameter data, historical machining data, and process specifications; extracting equipment geometric capabilities, accuracy parameters, dynamic performance, process capabilities, and control characteristics; and integrating them into a CNC parameter model using a multidimensional tensor representation method.

[0044] The numerical control parameter model is a third-order tensor M. ijk The first dimension i represents the equipment number on the production line, the second dimension j represents the parameter category (e.g., j=1 corresponds to geometric capability, j=2 corresponds to precision parameter, etc.), and the third dimension k corresponds to the specific parameter value. For non-numerical parameters, such as 'communication interface type', they are quantified using one-hot encoding and then stored in a tensor.

[0045] Geometric capabilities include workspace, travel range, and maximum load; dynamic performance includes maximum speed, acceleration, deceleration characteristics, and emergency stop distance; accuracy parameters include positioning accuracy, repeatability, and trajectory accuracy; process capabilities include executable process types, process quality indicators, and production efficiency; control characteristics include command response latency, communication interface type, and supported instruction sets.

[0046] By employing five-dimensional parameter representation and multi-dimensional tensor integration, a standardized description and unified assessment of the capabilities of different types of equipment are achieved. Detailed parameter classification ensures that the model covers all aspects of equipment characteristics, from spatial capabilities to control interfaces. This unified representation provides a scientific basis for comparing capabilities between equipment and matching tasks, eliminating the inconsistencies in equipment capability descriptions found in traditional methods.

[0047] Updating the CNC parameter model includes: setting an update cycle, triggering an update when the cycle is reached, updating the equipment's geometric capabilities, accuracy parameters, dynamic performance, and control characteristics based on the collected equipment operating status data, and updating the equipment's process capability-related parameters based on the latest collected machining result data.

[0048] Table 1 Examples of Basic Process Instruction Units

[0049]

[0050] For the machining of the engine cylinder head, the complete machining program needs to be first decomposed into a plurality of basic process instruction units. In this embodiment, the complete machining program of the cylinder head is obtained through the production management system, the machining program is usually provided in the form of G code or a process file in a specific format, a syntax analysis combined with semantic understanding is adopted to parse the G code or the process file, and the G code or the process file is decomposed into basic process instruction units. For example, as shown in Table 1, where Ra is an evaluation index of surface roughness, which refers to the profile arithmetic average deviation, and Ra3.2 indicates that the upper limit of Ra is 3.2.

[0051] Further, the process of generating the preliminary task allocation scheme specifically includes: obtaining and parsing the machining program and product process requirements, decomposing the machining program into basic process instruction units, and determining the dependency relationship between the basic process instruction units; based on the numerical control parameter model and the real-time running state of the equipment, evaluating the process matching degree, the capacity compliance degree and the expected execution result of each basic process instruction unit with each machining equipment, and calculating the adaptation degree; generating the preliminary task allocation scheme according to the adaptation degree and the preset allocation rule.

[0052] Through the process decomposition and adaptation degree evaluation process, an accurate mapping relationship between process requirements and equipment capacity is established, the process matching degree, the capacity compliance degree and the expected execution result are comprehensively considered, the preliminary task is optimally allocated, a solid foundation is provided for subsequent distributed negotiation, and resource waste is avoided.

[0053] Specifically, first, the current production target mode set by the upper-layer production management system is obtained, such as efficiency priority or quality priority. When in the efficiency priority mode, the method will aggregate a plurality of non-key processes that can be continuously executed on the same equipment into a basic process instruction unit with a relatively coarse granularity, so as to reduce the communication overhead and task switching time of subsequent distributed negotiation. On the contrary, when in the quality priority mode, each independent process, especially the process involving a critical dimension or precision, is decomposed in a fine-grained manner to ensure that each critical step can be independently evaluated and allocated to the optimal equipment. The decomposition logic further integrates the real-time health state of the equipment to achieve more fine-grained dynamic adjustment.

[0054] By adopting the process decomposition method of “dynamic granularity”, the process decomposition is changed from a static preprocessing step to the first link with intelligent perception and self-adaptive capability in the whole collaborative scheduling optimization chain. The generated task unit can better serve the macro production target, and the flexibility and purpose of production scheduling are significantly improved, whether pursuing maximum output or guaranteeing extreme quality, the task source can be adapted.

[0055] In this embodiment, the process matching evaluation device whether has the basic ability to execute the process; the ability compliance evaluation the matching degree of device performance and process requirements, such as whether the precision parameter of the device meets the requirements of the process on precision and efficiency; The expected execution result is to comprehensively evaluate the quality level, completion time and resource consumption that the basic process instruction unit may achieve under the current device state. Using the weighted scoring method, different weights are given to each evaluation dimension according to expert experience, and the comprehensive score is obtained by calculating the comprehensive score. In this embodiment, the weight is preferably 0.3, 0.3, 0.4. Specifically, the calculation method of the comprehensive score is: the normalized score of each factor (such as the adaptation degree, time cost, resource consumption) is multiplied by the corresponding weight coefficient, and then all the products are added. The expected execution result is obtained by receiving the numerical control parameter model and the real-time running state of the device through the built-in model; The built-in model can be an empirical formula, statistical regression or machine learning model obtained through a large number of experiments. In this embodiment, a random forest regression model is used for prediction, which takes device dynamic performance parameters and task characteristics as input and historical processing time and quality as output. The preset allocation rules include priority allocation rule, load balancing rule, process concentration rule and emergency task priority rule; The priority allocation rule is to preferentially allocate the basic process instruction unit to the device with the highest adaptation degree; The load balancing rule considers the device load situation to avoid overload of some devices; The process concentration rule is to preferentially allocate similar processes to the same device to reduce switching cost; The emergency task priority rule ensures that the emergency task is preferentially allocated.

[0056] Further, Figure 2 The flowchart for obtaining the final task allocation scheme and the task execution plan set of the present application, and the process of obtaining the final task allocation scheme and the task execution plan set specifically includes: in the distributed control network, each device is configured with a decision unit, and each decision unit generates and optimizes the local task sequence under the global constraint condition based on the preliminary task allocation scheme, the adaptation degree and the respective task evaluation function; Each decision unit declares, bids and negotiates according to the basic process instruction unit according to the contract net protocol algorithm; Through multiple rounds of negotiation and evaluation, the final task allocation scheme of each device and the task execution plan set containing the task order and time arrangement are formed.

[0057] The bidding information of each device decision unit is a data packet, containing [task ID, device ID, task evaluation value V ij , estimated start time, estimated completion time]. After the task manager receives all the bids, for the tasks without resource conflicts, the task with the highest V ijThe highest device wins the bid. When multiple winning tasks have time or resource conflicts on the same device or shared resources, a conflict resolution process is initiated. This process uses a priority-based iterative reallocation strategy: a) the highest priority conflict task allocation scheme is retained. b) the status of other conflict tasks is reset to 'pending declaration' and the priority score in the evaluation function of the original winning device is deducted, and then a new round of small-scale bidding-evaluation process is initiated. To prevent the negotiation process from falling into a dead loop, the system sets a maximum negotiation round, for example, the same conflict task group can be re-bid and evaluated for a maximum of 3 rounds, if it still cannot be resolved, the task will be transferred to the exception handling queue, and manual intervention or the execution of the default allocation strategy is performed.

[0058] The dynamic allocation of tasks is achieved through the decision unit and the contract net protocol. The multi-round negotiation and evaluation process can effectively solve resource conflicts, form an execution plan that meets both global constraints and optimizes local resource use, and improve the response speed and adaptability to production changes.

[0059] Table 2 shows the changes in the allocation of some tasks before and after negotiation. In the final scheme, most of the task allocation remains unchanged, but due to real-time state changes (such as T005 being reassigned to M001 due to high temperature of M003) and resource conflicts, the allocation of some tasks is adjusted.

[0060] Table 2 Comparison of task allocation schemes before and after negotiation

[0061]

[0062] In this embodiment, the global constraint conditions mainly include the deadline constraint (the overall production must be completed within the specified time window), the resource limit constraint (the use limit of shared resources among multiple devices), the process continuity constraint (a specific process must be completed continuously and not allowed to be interrupted), and the safety operation constraint (the spatial safety constraint when multiple devices work together). Based on the preliminary task allocation scheme, the fitness, and the task evaluation function of each device decision unit, the local task sequence is generated within the boundary of the global constraint condition. The generation process uses an optimization algorithm, with the goal of maximizing the score of the task evaluation function while satisfying each global constraint condition.

[0063] On the basis of the distributed control network, a task evaluation function is constructed for each device. The core of the task evaluation function is to provide a scientific and quantitative decision basis for the device in the "bidding" stage. The construction and calculation process of the function includes three key steps: First, the system will comprehensively evaluate six-dimensional factors, including: the adaptation degree factor reflecting the matching degree of the task and the device process; the time factor considering the expected execution time of the task and the current device load condition; the resource factor focusing on the availability of resources such as tools and fixtures required to execute the task; the quality factor evaluating the expected process quality level (such as tolerance, surface roughness); the energy consumption factor calculating the energy consumption required to execute the task; and the priority factor embodying the urgency and importance of the task. Second, before calculation, the function will normalize the original data of the above six dimensions to convert them into uniform and comparable score values. Most importantly, the weight coefficients of each factor are not fixed but can be dynamically adjusted according to real-time production targets. For example, when the production mode is set to "quality first", the system will automatically increase the weight coefficient of the quality factor; when set to "efficiency first", the weight of the time factor is increased; when set to "energy consumption first", the weight of the energy consumption factor is increased accordingly. Finally, the function generates a basic factor weighted score value that can comprehensively and objectively reflect the execution of the task under the current production target by multiplying the normalized score of all factors with their corresponding dynamic weight coefficients and then weighting and summing all the products.

[0064] The synergy benefit item is calculated, which is determined by prospectively analyzing the relevance between the task to be bid and the immediately preceding and following tasks in the local task sequence of the device. The relevance assessment includes at least: process continuity (e.g., the new task uses the same tool or fixture as the previous and next tasks, which can reduce the changeover time) and state inheritance (e.g., the processing parameters of the new task are similar to the previous task, which can reduce the device state adjustment time). If the new task has high synergy with the context tasks, the synergy benefit item is positive (reward); otherwise, if a large amount of changeover adjustment is required, it is negative (penalty). Finally, the comprehensive evaluation value of the task evaluation function is determined by the basic factor weighted score value and the synergy benefit item.

[0065] The "bid" decision of the device is more forward-looking and overall, which significantly improves the micro-timing and macro-efficiency of the production line. First, by introducing the "synergy benefit item", the evaluation function is no longer a short-sighted judgment of individual tasks, but rather a "process chain" perspective that intelligently evaluates the "smoothness" of tasks integrating into the existing production rhythm. This context-based evaluation mechanism can spontaneously guide tasks to the devices that best achieve continuous processing and reduce changeover waiting time, effectively reducing the "implicit" time cost caused by frequent tool changes, fixtures, or device state adjustments in the framework of distributed decision-making. Ultimately, this method not only optimizes the allocation of individual tasks, but also optimizes the local "rhythm" of the device task sequence, making the operation timing of the entire production line more stable and smooth, thereby improving overall production efficiency and device utilization.

[0066] The task negotiation and allocation among multiple devices is implemented by using the contract net protocol algorithm, including a task declaration stage, a bidding stage, an evaluation stage, an award stage, and a conflict resolution stage. In the task declaration stage, the basic process instruction units to be allocated are declared as tasks in the network. In the bidding stage, each device decision unit calculates the task evaluation function based on the preliminary task allocation scheme and the optimized local task sequence. In the evaluation stage, the calculation results of the task evaluation function based on all devices are evaluated uniformly to select the optimal bidding scheme. In the award stage, the task is formally allocated to the winning device, i.e., the device with the highest task evaluation function result, which incorporates the task into its local task sequence. In the conflict resolution stage, the problem of multiple devices bidding for the same task or resource conflict is solved.

[0067] In this embodiment, the local task sequence refers to the task execution plan optimized for the device within the decision boundary set by the global constraint condition based on the preliminary task allocation scheme and the adaptation degree data. The local task sequence mainly includes the priority order, estimated execution time, and resource requirements of the basic process instruction units that the device needs to process, and is an important basis for the device to make bidding decisions. The task execution plan set is a collection of detailed execution plans of each device formed after multiple rounds of negotiation under the contract net protocol, including the specific start and end times, required resources, and execution order of tasks on each device.

[0068] Further, the process of generating a collaborative production scheduling scheme specifically includes: the multi-constraint collaborative framework is composed of process constraints, resource constraints, time constraints, and space constraints; within the multi-constraint collaborative framework, the final task allocation scheme and the task execution plan set are used as inputs, and an optimization algorithm is used to solve a scheduling solution that satisfies all constraint conditions and optimizes the preset production target; the scheduling solution is converted into an executable collaborative production scheduling scheme, which clearly specifies the start and end times of each task on each device and the required resources.

[0069] A multi-constraint collaborative framework is constructed to include process, resource, time and space constraints, and an optimization algorithm is used to solve the optimal scheduling scheme that meets the multi-constraint conditions, so as to optimize the production target while ensuring feasibility, and the generated scheduling scheme has clear time arrangement and resource allocation, thereby improving the certainty of execution and the overall efficiency of the production line.

[0070] In this embodiment, the process constraint mainly focuses on the logical order and process requirements of the machining process, including the dependency relationship between processes, process parameter limitations and quality control requirements. The resource constraint focuses on the use limitations of various resources in the production process, including equipment resources, tooling resources, human resources and material resources. The time constraint focuses on the time arrangement of task execution, including deadline constraints, release time constraints, time window constraints and task duration constraints. The space constraint focuses on the spatial coordination of multiple devices in the shared workspace, including device workspace division, dynamic collision avoidance area, safety distance maintenance requirements and mutual exclusion access control of shared areas.

[0071] Based on the multi-constraint collaborative framework, the multi-objective optimization algorithm is used to solve the scheduling solution that meets all constraints and optimizes the preset production target: the optimization objective function is set according to production needs, which can comprehensively consider factors such as total time, total energy consumption, resource utilization balance and expected product quality index, and these factors are combined by weighting coefficients; the weight coefficient can be dynamically adjusted according to the production emphasis to meet the needs of different production scenarios; the solving algorithm adopts a hybrid optimization strategy, including using an improved genetic algorithm to generate an initial feasible solution set, using a simulated annealing algorithm for local search optimization, using a tabu search algorithm to jump out of the local optimum, and finally using Pareto analysis to screen the non-dominated solution set; the penalty function method is used for constraint processing, high penalty values are applied to hard constraint violations such as process order and resource capacity, and relatively low penalty values are applied to soft constraint violations such as balanced load.

[0072] In this embodiment, the flow of the optimized scheduling scheme is as follows:

[0073] Global search is performed using a genetic algorithm. The chromosome uses a process-based encoding method, and each individual represents a complete scheduling scheme. After 100 generations of evolution, the top 5 individuals with the highest fitness are selected as candidate solutions. The simulated annealing algorithm is started for local optimization on each of the 5 candidate solutions, and the neighborhood solution is generated by swapping the order of two tasks on the same device or moving a task to another available device. The initial temperature is set to 100, and the annealing coefficient is 0.98. Finally, the best solution among all local optimization results is selected as the scheduling solution.

[0074] F(x)=C max +λ1·∑V proc+ λ2·∑V res ,

[0075] The objective function is to minimize the maximum completion time and add penalty terms, including V proc , the penalty for violating the process dependency constraint, which is 1 if violated, otherwise 0; V res , the penalty for exceeding resource capacity. λ1, λ2 are large enough penalty coefficients, 10000 is adopted in the embodiment.

[0076] The scheduling solution obtained by optimization is converted into an executable collaborative production scheduling scheme, which clearly defines the specific start and end times of each task on each device and the required resources. The data structure of the collaborative production scheduling scheme includes global information, device control information, and shared resource information. The global information includes the overall start time, end time, expected output, and quality target; the device control information details the task list, resource allocation, and state change point on each device; and the shared resource information records the type, capacity, and allocation timeline of each shared resource. The conversion process of the collaborative production scheduling scheme includes time point conversion, task refinement, and resource allocation. Time point conversion converts relative time into absolute timestamp; task refinement refines abstract tasks into specific device operation instructions; and resource allocation converts resource demand into specific resource allocation plan. Finally, the scheduling scheme is distributed to each device through a distributed control network.

[0077] Further, the heterogeneous device motion collaborative control model specifically includes:

[0078] Integrate the kinematics characteristics, action timing logic, and spatial geometry information of each heterogeneous device; specifically, a global world coordinate system is established within the system. The spatial geometry information of all devices is loaded through a unified data structure (triangular mesh model is adopted in the embodiment) and registered in the world coordinate system. The kinematics model of each device is encapsulated as a standard interface, which receives motion instructions (target pose, speed, timestamp) in a unified format and can calculate its envelope in the global coordinate system in real time. Time axis control through the heterogeneous device motion collaborative control model includes: setting a global unified virtual time reference, and planning the start and stop time and synchronization nodes of each device action;

[0079] The establishment and synchronization process of the global unified virtual time reference is as follows: in the production line control system, a central dispatch controller is preset as a time master station, when the system is initialized or a new device accesses the network, each device controller executes an internal time calibration program, which is based on the principle of network time protocol (NTP) or precise time protocol (PTP), communicates with the time master station, and realizes precise synchronization of the respective local clock; during system operation, all motion instructions issued by the controller and state information reported by the device are forcibly attached with a synchronization timestamp based on the unified time, providing a unified time reference for all heterogeneous devices, so that the timing relationship (such as starting, synchronizing, and delaying) of cross-device cooperative action can be defined and executed to the millisecond level.

[0080] The dynamic safety area management includes: calculating and updating the safety working area and the non-safety working area of each device according to the real-time motion state data of the device. This management is based on the forward-looking motion envelope. The forward-looking motion envelope refers to predicting the spatial range that all parts of the device may occupy according to the current state of the device and the motion instructions within t seconds (for example, t = 0.5s) in the future. The calculation method is: taking the geometric model of the device at the current time, and interpolating to calculate the pose of the device at N key time points within t seconds in the future according to the motion instructions, and performing set operation on the geometric models at the N+1 poses to obtain an envelope. The system periodically (such as every 100ms) calculates the forward-looking motion envelope of each device, and uses the GJK algorithm to detect whether there is interference between these future envelopes, and calls the GJK algorithm to calculate the shortest distance between any two non-fixed links. When the distance is less than a preset safety threshold (for example, 200mm), the system immediately suspends the motion of the related device through the event-driven feedback control mechanism, thereby realizing predictive obstacle avoidance.

[0081] The cooperative control model integrating the motion characteristics, timing logic and spatial information of the devices realizes precise action control and dynamic safety area management under the unified time reference, significantly reduces the risk of interference between devices, ensures the safety of multi-device simultaneous operation, and improves stability and reliability.

[0082] Further, the process of multi-level adjustment through event-driven feedback control specifically includes: real-time monitoring of device state changes, task execution progress and abnormal events in the production process; when the monitoring information triggers a preset condition, starting the feedback control mechanism; the feedback control mechanism adjusts the action parameters of the related devices, including speed, acceleration and motion path, according to the specific event type and deviation data through the heterogeneous device motion cooperative control model, and feeds back the adjustment results and actual execution results for updating the numerical control parameter model.

[0083] When a task is completed, its actual machining precision P actual is collected and compared with the precision parameter P model in the model. The parameter is updated using an exponential moving average method:

[0084] P new = α·P actual +(1-α)·P model ,

[0085] wherein α is a learning rate, ranging from 0 to 1, and is set to 0.1 in this embodiment. This update formula feeds back the latest machining result to the model, realizing adaptive adjustment of the parameter.

[0086] The event-driven feedback control mechanism can respond to production abnormalities in real time and dynamically adjust action parameters, not only processing immediate abnormalities, but also improving adaptive ability and ensuring long-term stable operation.

[0087] Embodiment Two:

[0088] A flexible intelligent machining production line multi-assembly machine collaborative scheduling system, as shown in Figure 3 , includes a device capability representation module, a task decomposition matching module, a distributed negotiation module, a multi-constraint optimization module, and an action collaboration module. Specifically, it includes:

[0089] The device capability representation module establishes a numerical control parameter model for the machining device, including geometric capability, precision parameter, dynamic performance, process capability, and control characteristics. It collects real-time running state and machining result data of the device, and updates the numerical control parameter model.

[0090] Further, establishing a numerical control parameter model specifically includes: obtaining device basic parameter data, historical machining data, and process specification requirements, extracting device geometric capability, precision parameter, dynamic performance, process capability, and control characteristics, and integrating them into a numerical control parameter model using a multi-dimensional tensor representation method. Geometric capability includes workspace, travel range, and maximum load. Dynamic performance includes maximum speed, acceleration, deceleration characteristics, and emergency stop distance. Precision parameters include positioning accuracy, repeat positioning accuracy, and trajectory accuracy. Process capabilities include executable process types, process quality indicators, and production efficiency. Control characteristics include instruction response delay, communication interface type, and supported instruction set.

[0091] The task decomposition matching module decomposes the machining program into basic process instruction units, analyzes the adaptability of the basic process instruction units to the device based on the numerical control parameter model and the real-time running state of the device, and generates a preliminary task allocation scheme.

[0092] Further, the process of generating the preliminary task allocation scheme specifically comprises: obtaining and parsing the machining program and product process requirements, decomposing the machining program into basic process instruction units, and determining the dependency relationship between the basic process instruction units; based on the numerical control parameter model and the real-time running state of the equipment, evaluating the process matching degree, capacity compliance and expected execution result of each basic process instruction unit with each equipment, and calculating the adaptation degree; generating the preliminary task allocation scheme according to the adaptation degree and the preset allocation rule.

[0093] The distributed negotiation module builds a distributed control network among the devices, builds a task evaluation function for each device based on the preliminary task allocation scheme and the adaptation degree, generates a local task sequence for each device according to the global constraint condition, and obtains the final task allocation scheme and the task execution plan set based on the contract net protocol algorithm;

[0094] Further, the process of obtaining the final task allocation scheme and the task execution plan set specifically comprises: configuring a decision unit for each device in the distributed control network, and each decision unit generates and optimizes a local task sequence under the global constraint condition based on the preliminary task allocation scheme, the adaptation degree and the respective task evaluation function; each decision unit declares, bids and negotiates according to the basic process instruction unit according to the contract net protocol algorithm; through multiple rounds of negotiation and evaluation, the final task allocation scheme of each device and the task execution plan set containing the task order and time arrangement are formed.

[0095] The multi-constraint optimization module generates a collaborative production scheduling scheme according to the final task allocation scheme and the task execution plan set under a multi-constraint collaborative framework;

[0096] Further, the process of generating the collaborative production scheduling scheme specifically comprises: the multi-constraint collaborative framework is composed of process constraints, resource constraints, time constraints and space constraints; in the multi-constraint collaborative framework, the final task allocation scheme and the task execution plan set are used as inputs, and an optimization algorithm is used to solve a scheduling solution that satisfies all constraint conditions and optimizes the preset production target; the scheduling solution is converted into an executable collaborative production scheduling scheme, and the collaborative production scheduling scheme clearly specifies the start and end times of each task on each device and the required resources.

[0097] The action coordination module establishes a heterogeneous device motion coordination control model, performs time axis control and dynamic safety area management, and performs multi-level adjustment through event-driven feedback control.

[0098] Further, the motion coordination control model of the heterogeneous devices specifically comprises: integrating kinematics characteristics, action timing logic and spatial geometry information of each heterogeneous device; time axis control through the motion coordination control model of the heterogeneous devices comprises: setting a globally unified virtual time reference, and planning start-stop time and synchronization nodes of actions of each device; dynamic safety area management comprises: calculating and updating safety working areas and non-safety working areas of each device according to real-time motion state data of the device;

[0099] The process of multi-level adjustment through event-driven feedback control specifically comprises: monitoring device state changes, task execution progress and abnormal events in real time in the production process; when monitoring information triggers a preset condition, starting a feedback control mechanism; the feedback control mechanism adjusts action parameters of related devices, including speed, acceleration and motion path, through the motion coordination control model of the heterogeneous devices according to specific event types and deviation data, and feeds back adjustment results and actual execution results for updating the numerical control parameter model.

[0100] Although the embodiments of the present application have been shown and described, it can be understood by those skilled in the art that various changes, modifications, replacements and variations can be made to the embodiments without departing from the principles and spirits of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A method for collaborative scheduling of multiple units in a flexible intelligent manufacturing production line, characterized in that, include: Establish a CNC parameter model for the machining equipment, including geometric capabilities, accuracy parameters, dynamic performance, process capabilities, and control characteristics; collect real-time operating status and machining result data of the equipment; and update the CNC parameter model. The machining program is decomposed into basic process instruction units. The compatibility between the basic process instruction units and the equipment is analyzed by combining the CNC parameter model and the real-time operating status of the equipment, and a preliminary task allocation scheme is generated. A distributed control network is constructed among various devices. Based on the initial task allocation scheme and adaptability, a task evaluation function is constructed for each device. According to global constraints, a local task sequence for each device is generated. The final task allocation scheme and task execution plan set are obtained based on the contract network protocol algorithm. Under a multi-constraint collaborative framework, a collaborative production scheduling scheme is generated based on the final task allocation scheme and the task execution plan set. Establish a motion cooperative control model for heterogeneous devices, perform predictive time-axis control and dynamic safety zone management based on forward-looking motion envelope, and conduct multi-level adjustment through event-driven feedback control, including adjustment of low-level motion parameters and correction of high-level model parameters; The heterogeneous device motion collaborative control model specifically includes: integrating the kinematic characteristics, action timing logic, and spatial geometric information of various heterogeneous devices; time axis control through the heterogeneous device motion collaborative control model includes: setting a globally unified virtual time reference and planning the start and stop times and synchronization nodes of each device's actions; dynamic safety area management includes: calculating and updating the safe working area and unsafe working area of ​​each device based on the real-time motion status data of the devices. The process of multi-level adjustment through event-driven feedback control specifically includes: real-time monitoring of the production process; when the dynamic safety zone management based on forward-looking motion envelope predicts a future risk of motion interference, a low-level feedback control mechanism is triggered. This mechanism adjusts the speed, acceleration, and motion path of relevant equipment in real time according to the risk level and interference type to dynamically avoid potential collisions. After task execution, the actual motion trajectory and completion time of the equipment collected by sensors are compared with the predicted values ​​of the heterogeneous equipment motion collaborative control model. When the deviation exceeds a preset threshold, an upper-level feedback control mechanism is triggered. This mechanism uses the deviation data to reverse-correct and optimize the dynamic performance parameters in the CNC parameter model, improving model accuracy and the accuracy of future decisions.

2. The method for coordinated scheduling of multiple units in a flexible intelligent manufacturing production line according to claim 1, characterized in that, Establishing a CNC parameter model specifically includes: acquiring basic equipment parameter data, historical machining data, and process specifications; extracting the equipment's geometric capabilities, accuracy parameters, dynamic performance, process capabilities, and control characteristics; and integrating these into a CNC parameter model using a multidimensional tensor representation method. Geometric capabilities include workspace, travel range, and maximum load; dynamic performance includes maximum speed, acceleration, deceleration characteristics, and emergency stop distance; accuracy parameters include positioning accuracy, repeatability, and trajectory accuracy; process capabilities include executable process types, process quality indicators, and production efficiency; and control characteristics include command response latency, communication interface type, and supported instruction sets.

3. The method for coordinated scheduling of multiple units in a flexible intelligent manufacturing production line according to claim 1, characterized in that, The process of generating a preliminary task allocation scheme specifically includes: acquiring and parsing the machining program and product process requirements, decomposing the machining program into basic process instruction units, and determining the dependencies between each basic process instruction unit; based on the CNC parameter model and the real-time operating status of the equipment, evaluating the process matching degree, capability compliance degree and expected execution result of each basic process instruction unit with each piece of equipment, and calculating the fit degree; and generating a preliminary task allocation scheme based on the fit degree and preset allocation rules.

4. The method for coordinated scheduling of multiple units in a flexible intelligent manufacturing production line according to claim 1, characterized in that, The process of obtaining the final task allocation scheme and task execution plan set specifically includes: in the distributed control network, configuring decision units for each device; each decision unit generating and optimizing a local task sequence under global constraints based on the preliminary task allocation scheme, suitability, and its own task evaluation function; each decision unit making declarations, bidding, and negotiations according to the contract network protocol algorithm and the basic process instruction unit; and through multiple rounds of negotiation and evaluation, forming the final task allocation scheme for each device and a task execution plan set including task sequence and time arrangement.

5. The method for multi-unit collaborative scheduling of a flexible intelligent processing production line according to claim 1, characterized in that, The process of generating a collaborative production scheduling scheme specifically includes: the multi-constraint collaborative framework consists of process constraints, resource constraints, time constraints, and space constraints; within the multi-constraint collaborative framework, taking the final task allocation scheme and task execution plan set as input, an optimization algorithm is used to solve for a scheduling solution that satisfies all constraints and optimizes the preset production target; the scheduling solution is transformed into an executable collaborative production scheduling scheme, which specifies the start and end times of each task on each device and the required resources.

6. A flexible intelligent processing production line multi-machine collaborative scheduling system, executing the flexible intelligent processing production line multi-machine collaborative scheduling method as described in claim 1, characterized in that, include: The equipment capability characterization module establishes a CNC parameter model for the processing equipment, including geometric capability, accuracy parameters, dynamic performance, process capability and control characteristics, collects real-time operating status and processing result data of the equipment, and updates the CNC parameter model. The task decomposition and matching module decomposes the machining program into basic process instruction units, analyzes the compatibility between the basic process instruction units and the equipment by combining the CNC parameter model and the real-time operating status of the equipment, and generates a preliminary task allocation scheme. The distributed negotiation module constructs a distributed control network among various devices. Based on the initial task allocation scheme and adaptability, it constructs a task evaluation function for each device, generates a local task sequence for each device according to global constraints, and obtains the final task allocation scheme and task execution plan set based on the contract network protocol algorithm. The multi-constraint optimization module generates a collaborative production scheduling scheme based on the final task allocation scheme and task execution plan set under the multi-constraint collaborative framework. The motion coordination module establishes a motion coordination control model for heterogeneous devices, performs predictive time-axis control and dynamic safety zone management based on forward-looking motion envelopes, and conducts multi-level adjustment through event-driven feedback control. The heterogeneous device motion collaborative control model specifically includes: integrating the kinematic characteristics, action timing logic, and spatial geometric information of various heterogeneous devices; time axis control through the heterogeneous device motion collaborative control model includes: setting a globally unified virtual time reference and planning the start and stop times and synchronization nodes of each device's actions; dynamic safety area management includes: calculating and updating the safe working area and unsafe working area of ​​each device based on the real-time motion status data of the devices. The process of multi-level adjustment through event-driven feedback control specifically includes: real-time monitoring of the production process; when the dynamic safety zone management based on forward-looking motion envelope predicts a future risk of motion interference, a low-level feedback control mechanism is triggered. This mechanism adjusts the speed, acceleration, and motion path of relevant equipment in real time according to the risk level and interference type to dynamically avoid potential collisions. After task execution, the actual motion trajectory and completion time of the equipment collected by sensors are compared with the predicted values ​​of the heterogeneous equipment motion collaborative control model. When the deviation exceeds a preset threshold, an upper-level feedback control mechanism is triggered. This mechanism uses the deviation data to reverse-correct and optimize the dynamic performance parameters in the CNC parameter model, improving model accuracy and the accuracy of future decisions.

Images