Industrial production line dynamic scheduling method and system based on artificial intelligence

Abstract

The invention discloses an artificial intelligence-based industrial production line dynamic scheduling method and system, and relates to the technical field of intelligent manufacturing and industrial intelligent scheduling, and the method comprises the steps: collecting real-time state data, carrying out the preprocessing, taking the maximization of productivity, the shortest delivery time and the lowest energy consumption as optimization objectives, constructing a multi-objective reinforcement learning model, and carrying out the optimization of the multi-objective reinforcement learning model; scheduling priority data is generated, and operation distribution of each process node is adjusted in combination with a current equipment load threshold value and production bottleneck node information; when an abnormal condition is detected, triggering a rescheduling mechanism according to scheduling priority data, and updating an operation sequence and a resource allocation result; and synchronously feeding back the updated job allocation result and execution effect to the multi-target reinforcement learning model, and carrying out iterative optimization on the multi-target reinforcement learning model through a priority experience playback mechanism to realize continuous optimization of a scheduling strategy. According to the method, through a mode of combining multi-target reinforcement learning and dynamic scheduling, the learning efficiency and the optimization effect are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent manufacturing and industrial intelligent scheduling, and in particular to an artificial intelligence-based dynamic scheduling method and system for industrial production lines. Background Art

[0002] Traditional scheduling methods based on fixed rules are no longer able to meet the comprehensive requirements of high efficiency, low energy consumption, and rapid delivery in dynamic environments. To improve overall production capacity and resource utilization, the industry is gradually introducing artificial intelligence, data-driven optimization, and adaptive learning mechanisms. These methods aim to achieve intelligent scheduling and dynamic resource coordination for industrial production lines through real-time perception, intelligent decision-making, and feedback optimization. However, most current scheduling systems still rely primarily on static rule bases or local optimization strategies, lacking comprehensive consideration of multi-objective performance indicators. These systems are unable to adapt to complex scenarios such as equipment status fluctuations, evolving production bottlenecks, and unexpected events. Significant room for improvement is particularly needed in dynamically reconfiguring scheduling strategies and improving the algorithm's responsiveness and adaptability to real-time changes. Furthermore, current systems generally fail to continuously evolve scheduling optimization strategies and are unable to autonomously iterate and optimize decision models based on feedback, resulting in limited scheduling effectiveness and insufficient robustness. To address these issues, there is an urgent need for an artificial intelligence-based dynamic scheduling method and system for industrial production lines that can adaptively and efficiently allocate tasks at multiple process nodes, thereby improving the operational performance and intelligence of production systems in complex dynamic environments.

[0003] CN114548840B discloses an artificial intelligence-based production scheduling method and system. This method sorts order requests by obtaining user location information and target time, and introduces user levels to perform secondary sorting of the regional table to generate a task queue. While this solution establishes a customer-responsive scheduling model, improving the personalization and speed of order processing, it primarily focuses on determining the priority of external orders and does not fully consider the real-time equipment status, resource bottlenecks, and energy consumption targets within the production process. Therefore, it is still insufficient in achieving dynamic coordination and optimal resource allocation within the production process.

[0004] CN115469612B discloses a dynamic scheduling method for a job shop. This method uses normalized priority rules to dynamically schedule process switching and resource allocation between machine tools, enhancing process processing accuracy. However, this method relies on a preset rule base and lacks adaptive model learning capabilities. This method is unable to achieve real-time optimization and rapid response to dynamic changes in the production environment (such as sudden equipment failures or bottleneck process migration). Furthermore, scheduling strategy updates require human intervention and lack self-evolution capabilities. Summary of the Invention

[0005] In view of the following common problems in existing industrial production scheduling technologies: first, there is a lack of a comprehensive optimization mechanism for multiple objectives such as production capacity, delivery cycle and energy consumption, which limits the scheduling effect; second, the scheduling algorithm is generally static and lacks adaptability, making it difficult to cope with real-time changing working conditions; third, there is a lack of effective scheduling feedback mechanism and model iteration capability, which makes it impossible to achieve continuous optimization of the scheduling strategy. This paper proposes the present invention.

[0006] Therefore, the problem to be solved by the present invention is how to construct an intelligent scheduling method that can dynamically perceive equipment status and order progress, has multi-objective adaptive optimization capabilities, can quickly reschedule under abnormal conditions, and can continuously iterate and optimize scheduling strategies, so as to achieve a dynamic balance between maximizing production capacity, shortest delivery time and lowest energy consumption, and comprehensively improve the operating efficiency and intelligent decision-making level of industrial production lines.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] In a first aspect, the present invention provides an artificial intelligence-based dynamic scheduling method for industrial production lines.

[0009] It includes,

[0010] Collecting real-time status data of each process node in the industrial production line and preprocessing the real-time status data;

[0011] Based on pre-processed real-time status data, a multi-objective reinforcement learning model is constructed to generate scheduling priority data with the optimization goals of maximizing production capacity, minimizing delivery time, and minimizing energy consumption.

[0012] Based on the scheduling priority data, combined with the current equipment load threshold and production bottleneck node information, adjust the job allocation of each process node;

[0013] When an abnormal situation is detected, a rescheduling mechanism is triggered according to the scheduling priority data to update the job sequence and resource allocation results;

[0014] The updated job allocation results and execution effects are synchronously fed back to the multi-objective reinforcement learning model, and the multi-objective reinforcement learning model is iteratively optimized through a priority experience replay mechanism to achieve continuous optimization of the scheduling strategy.

[0015] As a preferred solution of the artificial intelligence-based dynamic scheduling method for industrial production lines described in the present invention, the updated job allocation results and execution effects are synchronously fed back to the multi-objective reinforcement learning model, and the multi-objective reinforcement learning model is iteratively optimized through a priority experience replay mechanism to achieve continuous optimization of the scheduling strategy, including:

[0016] Record the capacity indicator data, delivery time indicator data, and energy consumption indicator data of each process node during the execution of the updated scheduling plan, and establish an execution effect evaluation data set;

[0017] Comparing the execution effect evaluation data set with the expected indicator data of the job allocation plan, calculating the deviation value of each indicator, and generating scheduling effect deviation data;

[0018] Based on the scheduling effect deviation data, labeling priority scores for state transition samples in the multi-objective reinforcement learning model;

[0019] The labeled state transition samples are stored in the priority experience pool, and the samples are sorted according to the priority scores;

[0020] Extracting the top 60% of the priority scores from the priority experience pool to form a training batch, inputting them into the value evaluation network, calculating the temporal difference error, and updating the network weight parameters;

[0021] Based on the temporal difference error, gradient adjustment is performed on the decision probability distribution of the strategy generation network to generate optimized network model parameters;

[0022] The optimized network model parameters are updated to the multi-objective reinforcement learning model to complete the strategy iteration.

[0023] As a preferred solution of the artificial intelligence-based dynamic scheduling method for industrial production lines of the present invention, when an abnormal situation is detected, a rescheduling mechanism is triggered according to the scheduling priority data to update the job sequence and resource allocation results, including:

[0024] Continuously monitor the equipment operating status data of each process node. When the temperature value exceeds the preset upper limit or the vibration frequency value exceeds the normal range, it is marked as an abnormal process node;

[0025] Calculate the remaining workload and estimated recovery time of the abnormal process node to generate abnormal impact assessment data;

[0026] Based on the abnormal impact assessment data, extract the job allocation plan of the upstream three-level nodes and the downstream three-level nodes of the abnormal process node, and establish an emergency response buffer zone;

[0027] Reading the scheduling priority data, and transferring the pending orders of the abnormal process node to the available nodes in the emergency response buffer in descending order of priority;

[0028] Counting the remaining production capacity and material storage capacity of the available nodes to generate a node resource status table;

[0029] Based on the node resource status table, the to-be-processed orders are allocated to the available nodes according to a preset ratio, and a new job sequence table is generated;

[0030] Calculate the estimated completion time for each order in the new job sequence table. If the estimated completion time exceeds the delivery deadline, mark the corresponding order as an expedited order.

[0031] The expedited order is inserted into the front position of the current job queue, while the processing order of other non-expedited orders is postponed, and an updated scheduling plan is output.

[0032] As a preferred solution of the artificial intelligence-based dynamic scheduling method for industrial production lines of the present invention, the method for generating the job allocation plan is as follows:

[0033] Read the scheduling priority data and obtain the load factor value of each process node, where the load factor value is equal to the ratio of the current workload to the maximum capacity of the equipment;

[0034] Set a load threshold parameter. When the load rate value is greater than the first threshold, the corresponding process node is marked as a high-load node.

[0035] Calculate the predecessor task completion rate and subsequent node idle rate of each process node, and mark the process node whose predecessor task completion rate is less than a second threshold or whose subsequent node idle rate is greater than a third threshold as a production bottleneck node;

[0036] For the job tasks of the high-load node, select a node of the same type with a load rate value less than a fourth threshold as a backup node according to the order of the scheduling priority data;

[0037] Calculating the material transmission distance from the standby node to the high-load node, and selecting the standby node with the shortest material transmission distance as the task transfer target node;

[0038] For the production bottleneck node, extract the operation storage capacity of the upstream node and the production capacity margin of the downstream node to generate the node balance parameter;

[0039] Allocate the tasks of the production bottleneck node to three adjacent process nodes according to a preset ratio based on the node balance parameters;

[0040] A job allocation plan is generated, wherein the job allocation plan includes task transfer paths and allocation ratio data.

[0041] As a preferred solution of the artificial intelligence-based dynamic scheduling method for industrial production lines of the present invention, the method for generating the scheduling priority data is as follows:

[0042] Read the normalized state matrix from the production state database, extract the equipment operation state data, order progress information data and operation load data in the standardized data matrix, and construct a state vector;

[0043] Establishing a multi-objective reinforcement learning model, wherein the multi-objective reinforcement learning model adopts a dual network structure, including a value evaluation network and a strategy generation network;

[0044] Based on the state vector, calculating the scheduling priority score of each process node through the value assessment network, wherein the comprehensive score includes a production capacity score, a delivery time score, and an energy efficiency score;

[0045] Inputting the scheduling priority score into the strategy generation network to generate a scheduling priority value for each process node;

[0046] Arrange the scheduling priority values in descending order to generate scheduling priority data.

[0047] As a preferred solution of the artificial intelligence-based dynamic scheduling method for industrial production lines of the present invention, the method for establishing the multi-objective reinforcement learning model is as follows:

[0048] Constructing a state space matrix and an action space matrix, wherein the state space matrix includes process node numbers, equipment operating status data, order progress information data, and workload data; and the action space matrix includes scheduling priority options for process nodes;

[0049] Constructing a reward function based on the state space matrix and the action space matrix, wherein the reward function includes a production capacity reward item, a delivery reward item, and an energy consumption reward item;

[0050] Build a value evaluation network and a strategy generation network based on the reward function;

[0051] Setting an experience replay buffer, storing the state transition sequence, action selection results and obtained reward values in the experience replay buffer;

[0052] Updating the parameters of the value assessment network using a temporal difference algorithm, wherein the temporal difference algorithm calculates the error between the target value and the predicted value based on the Bellman equation;

[0053] Updating the parameters of the policy generation network using a policy gradient algorithm, wherein the policy gradient algorithm optimizes the action selection strategy by maximizing the expectation of long-term cumulative reward;

[0054] Use soft updates to synchronize the parameters of the value evaluation network and the strategy generation network to reduce the shock of the training process;

[0055] An experience priority sampling mechanism is set up to give priority to samples with high temporal difference errors for training the multi-objective reinforcement learning model.

[0056] As a preferred solution of the artificial intelligence-based dynamic scheduling method for industrial production lines described in the present invention, the value assessment network adopts a three-layer structure, including an input layer, a hidden layer and an output layer; the input layer receives the state space matrix; the output layer generates a state value score; the strategy generation network adopts a four-layer structure, including an input layer, two hidden layers and an output layer, the input layer receives the state value score; the output layer generates the action selection probability.

[0057] In a second aspect, the present invention provides an artificial intelligence-based industrial production line dynamic scheduling system, comprising:

[0058] The acquisition module is used to collect the real-time status data of each process node in the industrial production line and transmit it to the central data processing unit for pre-processing;

[0059] The construction module builds a multi-objective reinforcement learning model based on pre-processed real-time status data, with the optimization goals of maximizing production capacity, minimizing delivery time, and minimizing energy consumption, and generates scheduling priority data;

[0060] An adjustment module, which adjusts the job allocation of each process node based on the scheduling priority data, combined with the current equipment load threshold and production bottleneck node information;

[0061] An updating module, configured to trigger a rescheduling mechanism according to the scheduling priority data and update the job sequence and resource allocation results when an abnormal condition is detected;

[0062] The optimization module is used to synchronously feed back the updated job allocation results and execution effects to the multi-objective reinforcement learning model, and iteratively optimize the multi-objective reinforcement learning model through a priority experience replay mechanism to achieve continuous optimization of the scheduling strategy.

[0063] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the processor executes the computer program, it implements any step of the above-mentioned artificial intelligence-based industrial production line dynamic scheduling method.

[0064] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the above-mentioned artificial intelligence-based industrial production line dynamic scheduling method is implemented.

[0065] Compared with the existing technology, the beneficial effects of the present invention are as follows: by constructing a multi-objective reinforcement learning model with a dual network structure, the three objectives of production capacity, delivery time and energy consumption are unified into the same framework for optimization, overcoming the problem that traditional methods are difficult to balance multiple competing objectives. At the same time, the separation structure of the value evaluation network and the strategy generation network adopted significantly improves the training stability and convergence speed of the model, and avoids the oscillation in the strategy optimization process; by combining the dynamic job allocation mechanism of load threshold and bottleneck node information, on the basis of ensuring scheduling priority, the balanced distribution of production load is achieved, the material turnover loss is reduced, and the production bottleneck is effectively alleviated through the calculation of node balance parameters. The overall production efficiency has been improved; by establishing a multi-level emergency response mechanism and combining the buffering capacity of upstream and downstream nodes, an efficient exception handling process has been formed. An order redistribution method based on a preset ratio is adopted, and a priority insertion mechanism for expedited orders is introduced, which not only ensures production continuity under abnormal conditions, but also minimizes the impact of abnormal events on delivery time; through the priority experience replay mechanism, the execution effect is evaluated and fed back in real time, and a closed-loop continuous optimization system is established. The priority score labeling method based on scheduling effect deviation data improves the utilization efficiency of samples with important reference value, accelerates the optimization speed of the model, and enables the scheduling strategy to continuously adapt to the dynamic changes of the production environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort. Among them:

[0067] Figure 1 Flowchart of the dynamic scheduling method for industrial production lines based on artificial intelligence.

[0068] Figure 2 Flowchart for generating scheduling priority data for an artificial intelligence-based dynamic scheduling method for industrial production lines. DETAILED DESCRIPTION

[0069] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0070] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0071] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0072] The present invention is described in detail with reference to schematic diagrams. For ease of illustration, cross-sectional views of device structures may be partially enlarged and not to scale when describing embodiments of the present invention. Furthermore, the schematic diagrams are merely illustrative and should not limit the scope of the present invention. Furthermore, in actual production, the three-dimensional dimensions of length, width, and depth should be included.

[0073] In the description of the present invention, it should be noted that the terms "upper, lower, inner, and outer" and other references to orientations or positional relationships are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate and simplify the description of the present invention. They are not intended to indicate or imply that the devices or components referred to must have, be constructed, or operate in a specific orientation, and therefore should not be construed as limitations on the present invention. Furthermore, the terms "first, second, or third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0074] In this disclosure, unless otherwise specified or limited, the terms "mounted," "connected," and "connected" should be interpreted broadly. For example, they may refer to fixed, removable, or integral connections. They may also refer to mechanical, electrical, or direct connections, indirect connections through an intermediary, or internal communication between two components. Those skilled in the art will understand the specific meanings of these terms in this disclosure.

[0075] Example 1

[0076] Reference Figures 1 and 2 , which is the first embodiment of the present invention, provides an artificial intelligence-based dynamic scheduling method for industrial production lines, comprising:

[0077] S1: Collect the real-time status data of each process node in the industrial production line and pre-process the real-time status data.

[0078] In an optional embodiment, the real-time status data includes equipment operation status data, order progress information data, and operation load status data.

[0079] S1.1: Install data acquisition sensors at each process node in the industrial production line.

[0080] It should be noted that data acquisition sensors include equipment operation sensors, workpiece tracking sensors and load monitoring sensors.

[0081] In an optional embodiment, the temperature value, speed value, current value and vibration frequency value of each process node are collected by the equipment operation sensor to obtain the equipment operation status data; the number of completed workpieces, workpiece processing time and workpiece waiting time of each process node are collected by the workpiece tracking sensor to obtain the order progress information data; the material accumulation amount, workpiece in production quantity and equipment idle time of each process node are collected by the load monitoring sensor to obtain the operation load situation data.

[0082] S1.2: Integrate the equipment operation status data, order progress information data and operation load status data into a status data packet and transmit it to the data preprocessing unit via industrial Ethernet.

[0083] In an optional implementation manner, the data pre-processing unit performs abnormal data detection and elimination processing on the status data packet to generate a normalized status matrix.

[0084] In an optional embodiment, the temperature, current, and vibration frequency values in the status data packet are detected for anomalies based on the 3σ criterion. If the current value exceeds ±15% of the rated value, the bearing temperature value exceeds the threshold, or the vibration frequency value exceeds the safe range for three consecutive samplings, the data is marked as abnormal and eliminated, generating a post-cleaning status data set. For the post-cleaning status data set, the number of completed workpieces, workpiece processing time, and workpiece waiting time are linearly interpolated based on the acquisition timestamp to generate a synchronized status data sequence. A fast Fourier transform is performed on the equipment vibration amplitude in the synchronized status data sequence, and the energy proportion in the 0-100Hz frequency band is extracted as the vibration feature. The PCA algorithm is used to reduce the dimensionality of the temperature, current, and vibration frequency values to generate a low-dimensional feature vector. The low-dimensional feature vector, order progress information data, and workload status data are mapped to the [0,1] interval using Min-Max normalization to generate a normalized state matrix.

[0085] S1.3: Store the normalized state matrix into the production state database according to the preset sampling time interval.

[0086] S2: Based on the preprocessed real-time status data, with the optimization goals of maximizing production capacity, minimizing delivery time, and minimizing energy consumption, a multi-objective reinforcement learning model is constructed to generate scheduling priority data.

[0087] S2.1: Read the normalized state matrix from the production status database, extract the equipment operation status data, order progress information data and operation load data in the standardized data matrix, and construct the state vector.

[0088] S2.2: Establish a multi-objective reinforcement learning model, which adopts a dual network structure, including a value evaluation network and a strategy generation network.

[0089] In an optional embodiment, the value assessment network adopts a three-layer structure, including an input layer, a hidden layer and an output layer; the input layer receives the state space matrix; the output layer generates a state value score; the strategy generation network adopts a four-layer structure, including an input layer, two hidden layers and an output layer, the input layer receives the state value score; the output layer generates the action selection probability.

[0090] Specifically include:

[0091] S2.2.1: Construct a state space matrix and an action space matrix. The state space matrix includes the process node number, equipment operating status data, order progress information data, and workload data; the action space matrix includes the scheduling priority options for the process node;

[0092] S2.2.2: Based on the state space matrix and the action space matrix, construct a reward function, where the reward function includes a production capacity reward item, a delivery reward item, and an energy consumption reward item;

[0093] Preferably, the specific formula of the reward function is as follows:

[0094]

[0095] Among them, R(s,a) is the reward function value of taking action a in state s, α is the capacity reward weight coefficient, β is the delivery time reward weight coefficient, γ is the energy consumption reward weight coefficient, n is the total number of process nodes, P i is the actual capacity of the i-th node, P max is the maximum capacity limit, m is the total number of orders, T j is the actual completion time of the jth order, is the delivery deadline of the jth order, E cur is the current energy consumption value, E max The maximum energy consumption limit.

[0096] It should be noted that the value range of the reward function R is [0,1], where 0 represents the worst scheduling state and 1 represents the best scheduling state; the capacity reward item adopts simple normalization and the value range is [0,1]; the delivery reward item adopts exponential function mapping and the value range is [0,1]; the energy consumption reward item adopts linear normalization and the value range is [0,1].

[0097] S2.2.3: Build a value evaluation network and a strategy generation network based on the reward function;

[0098] S2.2.4: Set up an experience replay buffer and store the state transition sequence, action selection results, and reward values in the experience replay buffer;

[0099] S2.2.5: Update the parameters of the value assessment network using a temporal difference algorithm, which calculates the error between the target value and the predicted value based on the Bellman equation;

[0100] S2.2.6: Update the parameters of the policy generation network using a policy gradient algorithm, which optimizes the action selection strategy by maximizing the expected long-term cumulative reward.

[0101] S2.2.7: Use soft updates to synchronize the parameters of the value evaluation network and the policy generation network to reduce the oscillation of the training process;

[0102] S2.2.8: Set up an experience priority sampling mechanism to prioritize samples with high temporal difference errors for training multi-objective reinforcement learning models.

[0103] S2.3: Based on the state vector, the scheduling priority score of each process node is calculated through the value assessment network, where the comprehensive score includes the production capacity score, delivery time score and energy efficiency score.

[0104] S2.4: Input the scheduling priority score into the strategy generation network to generate the scheduling priority value of each process node.

[0105] Preferably, the specific formula for the scheduling priority value is as follows:

[0106]

[0107] Among them, D(s,a) is the scheduling priority value of taking action a in state s, ω1 is the weight coefficient of state action value, ω2 is the weight coefficient of load balancing, ω3 is the weight coefficient of waiting time, V(s,a) is the value evaluation network output value of taking action a in state s, max α′ is the maximum value operation among all optional actions a', L a is the current workload of the process node corresponding to action a, L max The maximum load threshold set for the system, W a is the current waiting time of the process corresponding to action a, W max is the maximum waiting time allowed by the system, and exp(*) is the natural exponential function.

[0108] For example, taking an electronic product assembly line as an example, when the value assessment network outputs state s, the value scores of the three optional actions are: a1 (prioritize process A) = 0.87, a2 (prioritize process B) = 0.92, a3 (prioritize process C) = 0.78; the strategy generation network will combine real-time load data when calculating: assuming that the current load of process B is 45 pieces (threshold 50 pieces) and the waiting time is 120 minutes (threshold 180 minutes), substituting it into the numerical formula of scheduling priority can obtain 0.903; at the same time, it is calculated that a1 = 0.812 and a3 = 0.704, so the system sets the scheduling priority of process B to the highest (0.903), and the generated processing order is B→A→C.

[0109] S2.5: Arrange the scheduling priority values in descending order to generate scheduling priority data.

[0110] S3: Based on the scheduling priority data, combined with the current equipment load threshold and production bottleneck node information, adjust the job allocation of each process node.

[0111] S3.1: Read the scheduling priority data and obtain the load rate value of each process node, where the load rate value is equal to the ratio of the current workload to the maximum production capacity of the equipment.

[0112] Preferably, the specific formula for the load rate value is as follows:

[0113]

[0114] Where L(b) is the load rate value of the b-th process node, μ is the time load weight coefficient, η is the queue load weight coefficient, ξ is the cache load weight coefficient, k is the number of jobs currently being processed, and t i is the processing time of the ith job, c i is the complexity coefficient of the i-th job, T b is the standard time period of node b, C b is the processing capacity coefficient of node b, q b is the current waiting queue length of node b, Q b is the maximum queue capacity of node b, τ b is the current available cache capacity of node b, B b is the total cache capacity of node n.

[0115] S3.2: Set a load threshold parameter. When the load rate value is greater than a first threshold, the corresponding process node is marked as a high-load node.

[0116] It should be noted that the load threshold parameters include a first threshold, a second threshold, a third threshold and a fourth threshold; the first threshold is set based on the statistical value of the historical maximum load rate of the equipment; the second threshold is set based on the average completion rate of the previous tasks; the third threshold is set based on the idle rate distribution of subsequent nodes; and the fourth threshold is set based on the average capacity utilization rate of similar nodes.

[0117] S3.3: Calculate the predecessor task completion rate and subsequent node idle rate of each process node, and mark the process node whose predecessor task completion rate is less than the second threshold or whose subsequent node idle rate is greater than the third threshold as a production bottleneck node.

[0118] S3.4: For the job tasks of the high-load node, select a node of the same type with a load rate value less than a fourth threshold as a backup node according to the order of the scheduling priority data.

[0119] S3.5: Calculate the material transmission distance from the standby node to the high-load node, and select the standby node with the shortest material transmission distance as the target node for task transfer.

[0120] Preferably, the specific formula for material transmission distance is as follows:

[0121]

[0122] Where D(i,j) is the material transmission distance from station i to station j, x i and y i is the coordinate position of station i, x j and y j is the coordinate position of workstation j, ∈ t is the path coefficient, σ is the obstacle detour coefficient, M is the number of obstacles on the transmission path, δ k is the detour indicator of the kth obstacle (0 or 1), h k is the additional detour distance caused by the kth obstacle.

[0123] S3.6: For the production bottleneck node, extract the operation storage capacity of the upstream node and the production capacity margin of the downstream node to generate the node balance parameters.

[0124] S3.7: Based on the node balance parameters, the work tasks of the production bottleneck node are distributed to the three adjacent process nodes according to the preset ratio.

[0125] For example, a typical process for handling a bottleneck node in an injection molding workshop is as follows: The upstream module backlog of an injection molding machine (node X) reaches 200 (normal value <50), and the downstream spraying process is idle at 40%. The system calculates the node balance parameters and finds: upstream storage capacity / downstream margin = 200 / (500 × 0.4) = 1.0 (warning value > 0.8). Based on the preset "30%-40%-30%" allocation rule, the current injection molding task is split into three parts: 30% is transferred to the closest backup injection molding machine Y in the workshop (transmission distance 15 meters, path coefficient 1.2), 40% is allocated to the original node X for further processing, and the remaining 30% is dispatched to injection molding machine Z across the workshop (subject to the obstacle coefficient of 1.5). Real-time monitoring of the adjusted node load shows that node X's load has dropped from 95% to 58%, node Y's load has increased from 30% to 65%, and node Z's load has increased from 45% to 72%. This improves overall production capacity by 32% while avoiding overloading of any single node.

[0126] S3.8: Generate a job allocation plan, where the job allocation plan includes task transfer paths and allocation ratio data.

[0127] S4: When an abnormal situation is detected, the rescheduling mechanism is triggered according to the scheduling priority data to update the job sequence and resource allocation results.

[0128] S4.1: Continuously monitor the equipment operating status data of each process node. When the temperature value exceeds the preset temperature upper limit or the vibration frequency value exceeds the normal range, it is marked as an abnormal process node.

[0129] S4.2: Calculate the remaining workload and estimated recovery time of the abnormal process node to generate abnormal impact assessment data.

[0130] Preferably, the correlation formula between the remaining workload and the estimated recovery time is as follows:

[0131]

[0132] Among them, T r is the expected recovery time, N is the number of remaining job types, and ε i is the unit workload coefficient of the i-th type of work, q i is the remaining quantity of the i-th type of job, P c is the current processing capacity, t d is the continuous working time, t max It is the maximum continuous working time limit.

[0133] S4.3: Based on the abnormal impact assessment data, extract the work allocation plan of the upstream three-level nodes and downstream three-level nodes of the abnormal process node, and establish an emergency response buffer zone.

[0134] S4.4: Read the scheduling priority data and transfer the pending orders of the abnormal process nodes to the available nodes in the emergency response buffer in descending order of priority.

[0135] S4.5: Count the remaining production capacity and material storage capacity of available nodes and generate a node resource status table.

[0136] S4.6: Based on the node resource status table, the orders to be processed are allocated to the available nodes according to the preset ratio, and a new operation sequence table is generated.

[0137] For example, when a CNC milling machine suddenly issues a spindle overheat alarm, the system executes the following emergency procedures:

[0138] 1) Assess the remaining processing volume, including three types of workpieces (20 pieces of type A × 1.2 coefficient, 15 pieces of type B × 0.8 coefficient, and 8 pieces of type C × 1.5 coefficient). Considering that the equipment requires 90 minutes to cool down, the total recovery time is calculated as ((20 × 1.2) + (15 × 0.8) + (8 × 1.5)) / (60 pieces / hour) + 90 / 60 = 2.33 hours.

[0139] 2) Activate the emergency buffer zone and select three upstream turning centers and two downstream grinding machines as available nodes;

[0140] 3) Based on the remaining capacity of each node (turning center 1 has 35 remaining pieces, turning center 2 has 28 remaining pieces, etc.), class A workpieces are allocated to turning center 1 in proportion (35 / (35+28)=55.6%);

[0141] 4) When recalculating the delivery time, it was found that the Class B workpiece would be delayed by 2 hours, so it was marked as an expedited order and inserted at the top of the queue for Grinding Machine 1, ultimately reducing the overall order delay by 63%.

[0142] S4.7: Calculate the estimated completion time for each order in the new job sequence table. When the estimated completion time exceeds the delivery deadline, mark the corresponding order as an expedited order.

[0143] S4.8: Insert the expedited order to the front of the current job queue, postpone the processing order of other non-expedited orders, and output the updated scheduling plan.

[0144] S5: The updated job allocation results and execution effects are synchronously fed back to the multi-objective reinforcement learning model, and the multi-objective reinforcement learning model is iteratively optimized through the priority experience replay mechanism to achieve continuous optimization of the scheduling strategy.

[0145] S5.1: Record the capacity indicator data, delivery time indicator data, and energy consumption indicator data of each process node during the execution of the updated scheduling plan, and establish an execution effect evaluation data set.

[0146] S5.2: Compare the execution effect evaluation data set with the expected indicator data of the job allocation plan, calculate the deviation value of each indicator, and generate scheduling effect deviation data.

[0147] S5.3: Based on the scheduling effect deviation data, label the priority scores of state transition samples in the multi-objective reinforcement learning model.

[0148] S5.4: Store the labeled state transition samples into the priority experience pool and sort the samples according to the priority scores.

[0149] S5.5: Extract the top 60% of priority score samples from the priority experience pool to form a training batch, input them into the value evaluation network, calculate the temporal difference error, and update the network weight parameters.

[0150] S5.5: Based on the temporal difference error, gradient adjustment is performed on the decision probability distribution of the strategy generation network to generate optimized network model parameters.

[0151] S5.6: Update the optimized network model parameters to the multi-objective reinforcement learning model to complete the policy iteration.

[0152] In summary, the present invention unifies the three objectives of production capacity, delivery time and energy consumption into the same framework for optimization by constructing a multi-objective reinforcement learning model with a dual network structure, overcoming the problem that traditional methods are difficult to balance multiple competing objectives. At the same time, the separation structure of the value evaluation network and the strategy generation network adopted significantly improves the training stability and convergence speed of the model, and avoids the oscillation in the strategy optimization process; through the dynamic job allocation mechanism combining load threshold and bottleneck node information, on the basis of ensuring scheduling priority, the balanced distribution of production load is achieved, the material turnover loss is reduced, and the production bottleneck is effectively alleviated through the calculation of node balance parameters, which improves the overall production efficiency. production efficiency; by establishing a multi-level linkage emergency response mechanism and combining the buffering capacity of upstream and downstream nodes, an efficient exception handling process is formed. An order redistribution method based on a preset ratio is adopted, and a priority insertion mechanism for expedited orders is introduced, which not only ensures production continuity under abnormal conditions, but also minimizes the impact of abnormal events on delivery time; through the priority experience replay mechanism, the execution effect is evaluated and fed back in real time, and a closed-loop continuous optimization system is established. The priority score labeling method based on scheduling effect deviation data improves the utilization efficiency of samples with important reference value, accelerates the optimization speed of the model, and enables the scheduling strategy to continuously adapt to the dynamic changes of the production environment.

[0153] Example 2

[0154] This embodiment also provides an artificial intelligence-based industrial production line dynamic scheduling system, including:

[0155] The acquisition module is used to collect the real-time status data of each process node in the industrial production line and transmit it to the central data processing unit for pre-processing;

[0156] The construction module builds a multi-objective reinforcement learning model based on pre-processed real-time status data, with the optimization goals of maximizing production capacity, minimizing delivery time, and minimizing energy consumption, and generates scheduling priority data;

[0157] The adjustment module adjusts the job allocation of each process node based on the scheduling priority data, combined with the current equipment load threshold and production bottleneck node information;

[0158] An update module is used to trigger a rescheduling mechanism based on the scheduling priority data when an abnormal situation is detected, and to update the job sequence and resource allocation results;

[0159] The optimization module is used to synchronously feed back the updated job allocation results and execution effects to the multi-objective reinforcement learning model, and iteratively optimize the multi-objective reinforcement learning model through the priority experience replay mechanism to achieve continuous optimization of the scheduling strategy.

[0160] This embodiment also provides an electronic device, which includes a processor, memory, a communication interface, a display screen, and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a multi-task edge computing resource scheduling method is implemented. The display screen of the computer device can be a liquid crystal display or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad provided on the computer device housing, or an external keyboard, touchpad or mouse.

[0161] This embodiment further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the method proposed in the above embodiment is implemented.

[0162] The storage medium proposed in this embodiment and the method proposed in the above embodiment belong to the same inventive concept. For technical details not fully described in this embodiment, please refer to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0163] Through the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented with the help of software and necessary general-purpose hardware, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute the method of the embodiment of the present invention.

[0164] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

[0165] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code. The scheme in the embodiment of the present application can be implemented in various computer languages.

[0166] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0167] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0168] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0169] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0170] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.

Claims

1. A dynamic scheduling method for industrial production lines based on artificial intelligence, characterized by: include, Collecting real-time status data of each process node in the industrial production line and preprocessing the real-time status data; Based on pre-processed real-time status data, a multi-objective reinforcement learning model is constructed to generate scheduling priority data with the optimization goals of maximizing production capacity, minimizing delivery time, and minimizing energy consumption. Based on the scheduling priority data, combined with the current equipment load threshold and production bottleneck node information, adjust the job allocation of each process node; When an abnormal situation is detected, a rescheduling mechanism is triggered according to the scheduling priority data to update the job sequence and resource allocation results; The updated job allocation results and execution effects are synchronously fed back to the multi-objective reinforcement learning model, and the multi-objective reinforcement learning model is iteratively optimized through a priority experience replay mechanism to achieve continuous optimization of the scheduling strategy.

2. The method for dynamic scheduling of industrial production lines based on artificial intelligence according to claim 1, wherein: The updated job allocation results and execution effects are synchronously fed back to the multi-objective reinforcement learning model, and the multi-objective reinforcement learning model is iteratively optimized through a priority experience replay mechanism to achieve continuous optimization of the scheduling strategy, including: Record the capacity indicator data, delivery time indicator data, and energy consumption indicator data of each process node during the execution of the updated scheduling plan, and establish an execution effect evaluation data set; Comparing the execution effect evaluation data set with the expected indicator data of the job allocation plan, calculating the deviation value of each indicator, and generating scheduling effect deviation data; Based on the scheduling effect deviation data, labeling priority scores for state transition samples in the multi-objective reinforcement learning model; The labeled state transition samples are stored in the priority experience pool, and the samples are sorted according to the priority scores; Extracting the top 60% of the priority scores from the priority experience pool to form a training batch, inputting them into the value evaluation network, calculating the temporal difference error, and updating the network weight parameters; Based on the temporal difference error, gradient adjustment is performed on the decision probability distribution of the strategy generation network to generate optimized network model parameters; The optimized network model parameters are updated to the multi-objective reinforcement learning model to complete the strategy iteration.

3. The method for dynamic scheduling of industrial production lines based on artificial intelligence according to claim 2, characterized in that: When an abnormal situation is detected, a rescheduling mechanism is triggered according to the scheduling priority data to update the job order and resource allocation results, including: Continuously monitor the equipment operating status data of each process node. When the temperature value exceeds the preset upper limit or the vibration frequency value exceeds the normal range, it is marked as an abnormal process node; Calculate the remaining workload and estimated recovery time of the abnormal process node to generate abnormal impact assessment data; Based on the abnormal impact assessment data, extract the job allocation plan of the upstream three-level nodes and the downstream three-level nodes of the abnormal process node, and establish an emergency response buffer zone; Reading the scheduling priority data, and transferring the pending orders of the abnormal process node to the available nodes in the emergency response buffer in descending order of priority; Counting the remaining production capacity and material storage capacity of the available nodes to generate a node resource status table; Based on the node resource status table, the to-be-processed orders are allocated to the available nodes according to a preset ratio, and a new job sequence table is generated; Calculate the estimated completion time for each order in the new job sequence table. If the estimated completion time exceeds the delivery deadline, mark the corresponding order as an expedited order. The expedited order is inserted into the front position of the current job queue, while the processing order of other non-expedited orders is postponed, and an updated scheduling plan is output.

4. The method for dynamic scheduling of industrial production lines based on artificial intelligence according to claim 3, characterized in that: The method for generating the job allocation scheme is: Read the scheduling priority data and obtain the load rate value of each process node, where the load rate value is equal to the ratio of the current workload to the maximum capacity of the equipment; Set a load threshold parameter. When the load rate value is greater than the first threshold, the corresponding process node is marked as a high-load node. Calculate the predecessor task completion rate and subsequent node idle rate of each process node, and mark the process node whose predecessor task completion rate is less than a second threshold or whose subsequent node idle rate is greater than a third threshold as a production bottleneck node; For the job tasks of the high-load node, select a node of the same type with a load rate value less than a fourth threshold as a backup node according to the order of the scheduling priority data; Calculating the material transmission distance from the standby node to the high-load node, and selecting the standby node with the shortest material transmission distance as the task transfer target node; For the production bottleneck node, extract the operation storage capacity of the upstream node and the production capacity margin of the downstream node to generate the node balance parameter; Allocate the tasks of the production bottleneck node to three adjacent process nodes according to a preset ratio based on the node balance parameters; A job allocation plan is generated, wherein the job allocation plan includes task transfer paths and allocation ratio data.

5. The method for dynamic scheduling of industrial production lines based on artificial intelligence according to claim 4, characterized in that: The method for generating the scheduling priority data is: Read the normalized state matrix from the production state database, extract the equipment operation state data, order progress information data and operation load data in the standardized data matrix, and construct a state vector; Establishing a multi-objective reinforcement learning model, wherein the multi-objective reinforcement learning model adopts a dual network structure, including a value assessment network and a strategy generation network; Based on the state vector, calculating the scheduling priority score of each process node through the value assessment network, wherein the comprehensive score includes a production capacity score, a delivery time score, and an energy efficiency score; Inputting the scheduling priority score into the strategy generation network to generate a scheduling priority value for each process node; Arrange the scheduling priority values in descending order to generate scheduling priority data.

6. The method for dynamic scheduling of industrial production lines based on artificial intelligence according to claim 5, characterized in that: The method for establishing the multi-objective reinforcement learning model is: Constructing a state space matrix and an action space matrix, wherein the state space matrix includes process node numbers, equipment operating status data, order progress information data, and workload data; and the action space matrix includes scheduling priority options for process nodes; Constructing a reward function based on the state space matrix and the action space matrix, wherein the reward function includes a production capacity reward item, a delivery reward item, and an energy consumption reward item; Build a value evaluation network and a strategy generation network based on the reward function; Setting an experience replay buffer, storing the state transition sequence, action selection results and obtained reward values in the experience replay buffer; Updating the parameters of the value assessment network using a temporal difference algorithm, wherein the temporal difference algorithm calculates the error between the target value and the predicted value based on the Bellman equation; Updating the parameters of the policy generation network using a policy gradient algorithm, wherein the policy gradient algorithm optimizes the action selection strategy by maximizing the expectation of long-term cumulative reward; Use soft updates to synchronize the parameters of the value evaluation network and the strategy generation network to reduce the shock of the training process; An experience priority sampling mechanism is set up to give priority to samples with high temporal difference errors for training the multi-objective reinforcement learning model.

7. The method for dynamic scheduling of industrial production lines based on artificial intelligence according to claim 6, characterized in that: The value assessment network adopts a three-layer structure, including an input layer, a hidden layer and an output layer; the input layer receives the state space matrix; the output layer generates the state value score; the strategy generation network adopts a four-layer structure, including an input layer, two hidden layers and an output layer, the input layer receives the state value score; the output layer generates the action selection probability.

8. An artificial intelligence-based dynamic scheduling system for industrial production lines, based on the artificial intelligence-based dynamic scheduling method for industrial production lines according to any one of claims 1 to 7, characterized in that: include, The acquisition module is used to collect the real-time status data of each process node in the industrial production line and transmit it to the central data processing unit for pre-processing; The construction module builds a multi-objective reinforcement learning model based on pre-processed real-time status data, with the optimization goals of maximizing production capacity, minimizing delivery time, and minimizing energy consumption, and generates scheduling priority data; An adjustment module, which adjusts the job allocation of each process node based on the scheduling priority data, combined with the current equipment load threshold and production bottleneck node information; An updating module, configured to trigger a rescheduling mechanism according to the scheduling priority data and update the job sequence and resource allocation results when an abnormal condition is detected; The optimization module is used to synchronously feed back the updated job allocation results and execution effects to the multi-objective reinforcement learning model, and iteratively optimize the multi-objective reinforcement learning model through a priority experience replay mechanism to achieve continuous optimization of the scheduling strategy.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the artificial intelligence-based industrial production line dynamic scheduling method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the artificial intelligence-based dynamic scheduling method for industrial production lines according to any one of claims 1 to 7 are implemented.

Images