Automatic scheduling and monitoring management system for three-axis plane milling machine

Abstract

The invention relates to the technical field of automatic manufacturing, and discloses a triaxial plane milling machine automatic scheduling and monitoring management system which comprises a resource vector module, a process entropy real-time monitoring module, a scheduling decision module and a self-adaptive correction module. Selecting an optimal sequence according to the evaluation sequence cost; the process entropy real-time monitoring module calculates an actual process entropy track in real time during processing; and the adaptive correction module obtains a trajectory deviation degree by comparing a predicted trajectory and an actual trajectory, and corrects the resource vector when a condition is met to form closed-loop learning. According to the method, the process entropy is used as a quantitative index for connecting scheduling and a physical process, active scheduling is performed through prospective trajectory prediction and evaluation, and self-evolution of the model is realized by using a self-adaptive correction mechanism based on trajectory deviation, so that deep fusion of scheduling and monitoring and dynamic optimization of the system are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of automation manufacturing technology, in particular to a three-axis planer milling machine automation scheduling and monitoring management system. BACKGROUND

[0002] In modern manufacturing industry, the automation production management of numerical control machine tools such as three-axis planer milling machines usually relies on manufacturing execution system for task scheduling, and is assisted by state monitoring system to ensure the reliability of equipment operation. However, the existing technical solutions still have inherent technical defects in realizing the deep integration of scheduling and monitoring.

[0003] The current production scheduling logic mainly plans according to upper management information such as order priority, delivery period and material preparation situation, and the processing sequence generated by it is often rigid, lacking real-time physical state perception of the machine tool. At the same time, although the state monitoring system of the machine tool can monitor physical quantities such as vibration and temperature through sensors, its role is mostly passive response. Usually, only when a monitoring index exceeds the preset fixed safety threshold, the system will trigger an alarm or forced shutdown.

[0004] This mode of separation between scheduling and monitoring and passive response leads to a serious disconnection between decision-making and physical process. When the scheduling system dispatches a high-load processing task, it cannot pre-assess the cumulative damage that the task may cause to the current machine tool state; while the monitoring system can only respond when the damage accumulates to a certain extent and manifests as a significant abnormal signal, at which time irreversible damage to the workpiece quality or machine tool health has often already occurred. The simple signal linkage in the existing technology, such as suspending the task when the vibration exceeds the limit, cannot fundamentally solve the problem, because it does not take the dynamic health status of the machine tool as an endogenous variable affecting scheduling decisions. Therefore, the system cannot realize proactive self-regulation, such as actively choosing a low-load rest task to restore the stability of the machine tool after it has undergone high-intensity processing, so as to effectively cope with the processing process uncertainty caused by dynamic factors such as material batch differences and tool gradual wear, limiting the overall stability and optimization level of the production process. SUMMARY

[0005] In view of the deficiencies of the prior art, the present application provides a three-axis planer milling machine automation scheduling and monitoring management system, which solves the technical problems of scheduling rigidity, monitoring lag and disconnection between decision-making and physical process state in the prior art three-axis planer milling machine automation management system.

[0006] To achieve the above-mentioned purpose, the first aspect of the present application provides a three-axis planer milling machine automation scheduling and monitoring management system, which comprises: A resource vector module is configured to define and store a resource vector for each task to be processed, which contains physical attributes of the task. Specifically, the resource vector includes a tool wear factor and a quality risk coefficient. The tool wear factor is used to represent the intensity of tool wear caused by the task, and the quality risk coefficient is used to represent the sensitivity of the task to the stability of the machine tool.

[0007] A process entropy real-time monitoring module is connected to sensors of the machine tool. The sensors include, but are not limited to, vibration sensors, acoustic emission sensors, and motor current sensors. The module is configured to collect signals of the sensors during the processing, and calculate a real-time actual process entropy trajectory based on the signals. The calculation includes wavelet packet transform of the collected signals to obtain energy proportions of each frequency band, and calculation of the process entropy at time t according to the following formula:

[0008] A scheduling decision module is connected to the resource vector module and the process entropy real-time monitoring module. The module is configured to generate predicted entropy trajectories for multiple candidate sequences based on the resource vectors stored in the resource vector module and the current process entropy, and calculate sequence costs of each candidate sequence by integrating the predicted entropy trajectories. Finally, the candidate sequence with the lowest sequence cost is selected as the optimal sequence for processing. The calculation formula of the sequence cost is:

[0009] ​​​​​​​​​​​​​​​​​​​​​​and an adaptive correction module connected with the scheduling decision module and the process entropy real-time monitoring module. The module is configured to: calculate a trajectory deviation by comparing the predicted entropy trajectory of the optimal sequence with the actual process entropy trajectory; and correct the resource vector stored in the resource vector module when the trajectory deviation exceeds a preset error threshold. The correction is to automatically adjust the tool wear factor and / or the quality risk coefficient in the resource vector according to the size of the trajectory deviation and by introducing a learning rate parameter associated with the material information of the task. After the correction, the updated resource vector is stored in the resource vector module for subsequent scheduling decisions.

[0010] The second aspect of the present application provides a three-axis planar milling machine automatic scheduling and monitoring management method, which is realized by the above-mentioned system and includes the following steps: S1, defining and storing a resource vector containing the physical properties of each task to be processed; S2, collecting machine tool sensor signals during processing and calculating an actual process entropy trajectory in real time; S3, generating predicted entropy trajectories for multiple candidate sequences based on the resource vector and the current process entropy, and selecting the optimal sequence for processing according to the sequence cost evaluation; S4, calculating a trajectory deviation by comparing the predicted entropy trajectory of the optimal sequence with the actual process entropy trajectory, and correcting the resource vector when the trajectory deviation meets a preset condition.

[0011] The present application provides a three-axis planar milling machine automatic scheduling and monitoring management system. It has the following beneficial effects: 1. The present application sets a scheduling decision module, which can generate and evaluate the predicted entropy trajectories of different processing sequences in advance based on the resource vector of the task to be processed, and select the optimal sequence according to the sequence cost, realizing the foresight of scheduling decision. This way changes the scheduling from the traditional passive, rule-based dispatching to an active, future physical state prediction-based optimization process, so that the system can actively avoid risks by optimizing the task sequence before potential unstable states occur.

[0012] 2. The present application introduces process entropy as a unified state quantization index, establishing a direct and quantitative correlation between scheduling decision and physical process. The process entropy real-time monitoring module translates the complex physical state of the machine tool during processing into a continuous process entropy trajectory, and the scheduling decision module takes this trajectory and its predicted value as the core decision variable, overcoming the technical defect that the upper scheduling information and the lower physical state are disconnected in the prior art, and realizing the deep integration of the two.

[0013] 3、The adaptive correction module is set, the system model is endowed with self-learning and optimization capability, the module calculates the trajectory deviation degree between the predicted entropy trajectory and the actual process entropy trajectory, and automatically corrects the basic data in the resource vector module according to the trajectory deviation degree, the closed-loop learning mechanism enables the internal model of the system to continuously learn from the prediction error, and converges to the real physical process law, so that the decision accuracy can be maintained when facing dynamic uncertain factors such as material batch change and tool state evolution. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 The system architecture of the present application is shown in the figure. Figure 2 The method flowchart of the present application is shown in the figure. DETAILED DESCRIPTION

[0015] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0016] Embodiment: Please refer to the accompanying Figure 1 The embodiment of the present application provides a three-axis planar milling machine automatic scheduling and monitoring management system, which comprises: A resource vector module is used to define and store a resource vector containing physical properties for each to-be-processed task. In the embodiment, the resource vector module is included. The core function of the module is to provide a standardized and calculable data basis for scheduling decision and adaptive learning. In order to realize this function, the module converts each abstract to-be-processed task or process into a multi-dimensional resource vector containing its key physical properties.

[0017] Specifically, the resource vector module is configured to establish and maintain a dynamic knowledge base. In the knowledge base, each to-be-processed task is mapped to a unique resource vector The structured definition of the resource vector is as follows: ; The components of the resource vector are described as follows: It is a unique identifier for the task, which is used to index, track and data associate the processing task in the system, and ensures the accuracy of data processing.

[0018] A workpiece material information identifier, which is associated with the material grade, batch or other physical characteristics of the workpiece to be machined. The identifier is not only used for record, but also as the basis for selecting the corresponding learning rate parameter in the subsequent adaptive correction step, so that the system can exhibit differentiated learning and adaptive ability for materials of different machining difficulty.

[0019] A machining time estimator, whose initial value can be derived from the analysis results of CAD / CAM software in the process planning stage, or obtained based on historical machining data of similar tasks. This parameter provides basic data for the sequence planning and time evaluation of the scheduling decision module.

[0020] Further, the resource vector includes two core physical property parameters: a tool wear factor and a quality risk coefficient These two parameters aim to quantify the inherent physical impact of the machining task.

[0021] The tool wear factor is a normalized dimensionless parameter, which is used to represent the intensity of the task on the tool life. In a preferred embodiment, its initial value is obtained by comprehensive evaluation of the task process parameters, including but not limited to cutting depth, feed rate, spindle speed and workpiece material hardness. By abstracting these complex process effects into a single factor, the scheduling decision module can directly evaluate the potential impact of the task on the tool state when making predictions.

[0022] The quality risk coefficient is also a normalized dimensionless parameter, which is used to represent the sensitivity of the task to the stability of the machine tool state. For example, a finishing process with high dimensional accuracy or surface finish requirements is more sensitive to state fluctuations such as thermal deformation and slight vibration of the machine tool, so it will be assigned a higher value. This coefficient enables the scheduling decision module to identify and prioritize tasks that have special requirements for the machine tool state, or actively avoid executing such high-risk tasks when the machine tool state is not good.

[0023] The resource vector module is not only a static data storage unit, but also closely cooperates with other modules of the system to form the core of data flow and evolution.

[0024] On the one hand, the module provides decision input for the scheduling decision module. Specifically, the tool wear factor and the quality risk coefficient in the resource vector are key input variables in the entropy increase trajectory prediction model. For example, when predicting the entropy increase caused by a task, the following entropy increase function can be used: ; wherein, and are preset weight coefficients. In this way, the scheduling decision is no longer based only on time or priority, but can foresee the potential impact of different tasks on the physical state of the machine tool, so as to make a more reasonable sequence arrangement.

[0025] On the other hand, the module is the object of the adaptive correction module, so as to realize the closed-loop learning of the system. When the adaptive correction module calculates the trajectory deviation degree between the actual machining process and the prediction result , the module will update the corresponding resource vector in the knowledge base. Its correction process can be described by the following formula: ; ; wherein, and are new values after correction; and are old values before correction; and and are learning rate parameters associated with material information .

[0026] In this way, if the system's prediction of the processing difficulty of a certain batch of materials is too low, the and values in the resource vector will be automatically increased. In this way, when the same or similar material is encountered again in the next task, the prediction model of the scheduling decision module will be based on the updated resource vector that is more in line with the actual situation, so that the prediction result is more accurate.

[0027] In summary, the resource vector module provides a basic data model and implementation approach for the entire system to realize predictive scheduling and adaptive learning based on physical state, by quantitatively modeling the multi-dimensional physical attributes of the machining task and establishing a dynamic update closed loop linked with prediction, feedback and correction mechanisms.

[0028] The process entropy real-time monitoring module is used to collect sensor signals of the machine tool during the machining process, and a real-time actual process entropy trajectory is calculated accordingly; In this embodiment, it includes a process entropy real-time monitoring module. The purpose of this module is to objectively characterize such a complex physical process as milling through a unified and quantifiable index, so as to provide real-time and reliable physical state input for subsequent predictive scheduling and adaptive learning.

[0029] Specifically, the process entropy real-time monitoring module is configured to output an actual process entropy trajectory continuously and in real time throughout the whole cycle of the machining task execution. The trajectory is generated following a rigorous processing procedure from multi-source physical signal acquisition to single information theory index output.

[0030] In a preferred embodiment, the data input of the module is derived from one or more sets of sensors deployed at key locations of the machine tool. In order to capture the state changes in the machining process comprehensively, the sensors can include acceleration sensors for monitoring mechanical vibration, acoustic emission sensors for detecting events such as internal micro-cracks in the material, and spindle motor load current sensors for reflecting the cutting load energy consumption. The module synchronously acquires these sensor signals of different sources and different physical meanings to ensure accurate alignment of the obtained data in the time dimension, forming a multi-dimensional signal vector .

[0031] After obtaining the original signals, in order to extract the core features that can effectively represent the process stability from the complex and noisy signals, the module performs the following processing steps: First, the multi-dimensional signal vector is decomposed into a set of energy features . In a rolling short time window, the feature extraction is performed. Preferably, the wavelet packet transform technique is adopted. The reason for choosing this technique is that the milling machining signals usually have non-stationary and transient characteristics, and the wavelet packet transform can provide good resolution in both time and frequency domains, suitable for capturing the subtle changes of such signals. By decomposing the signal into multiple layers of wavelet packets, the energy of the signal can be mapped to non-overlapping frequency bands, thereby obtaining a feature vector that can describe the energy distribution at the current time. wherein represents the signal energy in the

[0032] th frequency band. Secondly, in order to convert the energy features in the physical domain into analysis objects in the information theory domain, the module normalizes the energy feature vector and constructs a probability distribution model. Specifically, by calculating the proportion of the energy of each frequency band in the total energy, a probability distribution vector is obtained. ; Here, indicates that near the time , the energy of the machining process has a probability distribution of in the th frequency band.a few specific frequency bands, while a process that tends to be unstable or abnormal will spread its energy to a wider frequency band.

[0033] Finally, the module calculates the Shannon entropy of the process at this moment based on the probability distribution constructed above, and defines it as the process entropy . The calculation follows the formula: ; The process entropy calculated by this formula is a single, dimensionless scalar value. Its physical meaning is to quantify the degree of uncertainty or disorder of the process state. When the process is stable, the energy distribution is concentrated, the uncertainty is low, and the process entropy value is also low; on the contrary, when unstable phenomena such as chatter and tool wear intensify, the energy distribution becomes dispersed, the uncertainty increases, and the process entropy value will rise accordingly.

[0034] The process entropy real-time monitoring module generates a continuous curve that changes over time, i.e., the actual process entropy trajectory, by repeatedly executing the above calculation steps in a continuous, rolling window.

[0035] This actual process entropy trajectory plays a dual key role in the entire system: On the one hand, it is transmitted to the adaptive correction module as a benchmark or ground truth for comparison with the predicted entropy trajectory generated by the scheduling decision module. The difference between the two, i.e., the trajectory deviation, is the fundamental basis for driving the system model to learn and optimize itself.

[0036] On the other hand, at the end of a task processing, the endpoint value of this trajectory, i.e., the actual process entropy at this moment, is passed to the scheduling decision module as the initial state of the next scheduling and prediction cycle. This ensures that each prediction decision is based on the latest and most real physical state of the machine tool, forming the basis for the closed loop of the system's rapid response and decision-making.

[0037] In summary, the process entropy real-time monitoring module translates the invisible and complex process state into an intuitive and quantifiable process entropy trajectory through the conversion from multiple physical domains to a single information domain, providing the core state perception ability for the system to achieve accurate prediction, effective comparison, and continuous adaptive learning.

[0038] The scheduling decision module is connected with the resource vector module and the process entropy real-time monitoring module, and is configured to generate predicted entropy trajectories for multiple candidate sequences based on the resource vector and the current process entropy, and select an optimal sequence for processing according to the evaluation sequence cost; In this embodiment, it contains a scheduling decision module. This module serves as the decision core of the system, and its purpose is to abandon the traditional scheduling method based on static rules, and instead adopt a forward-looking decision mechanism based on the prediction of future physical states, in order to select the processing path that is most beneficial to the long-term stable operation of the system.

[0039] The scheduling decision module is connected with the resource vector module and the process entropy real-time monitoring module, forming the data path required for its decision. On the one hand, it obtains the standardized resource vector of each task to be processed from the resource vector module; on the other hand, it receives real-time feedback from the process entropy real-time monitoring module, i.e. the final actual process entropy of the machine tool when the previous task is completed, and takes this value as the initial state of this decision.

[0040] In a complete decision-making cycle, the scheduling decision module performs the following series of precise and interlocking steps: Firstly, the module does not evaluate individual tasks in isolation, but selects several tasks from the queue of tasks to be processed and arranges them into multiple different, short-range candidate sequences, denoted as The significance of this approach is to expand the scope of decision-making from single optimal point selection to optimal path planning, thereby enabling the evaluation of the continuous impact between tasks.

[0041] Secondly, for each generated candidate sequence , the module calls its internal entropy increase trajectory predictor to generate a predicted entropy trajectory for the sequence. The prediction process is based on a state transition model. Specifically, for any task in the sequence, the prediction of how the execution of the task will change the process entropy of the system is made. Assuming that the initial process entropy before executing the task is , the terminal process entropy after executing the task can be predicted by the following formula: ; Here, the function is an entropy increase function that is directly related to the physical properties of the task, i.e. its resource vector. In a preferred embodiment, the entropy increase function is defined as the weighted sum of the tool wear factor and the quality risk coefficient of the task: ; where and are the system's preset weight coefficients. Through this model, a task with high or high The value of the task will be predicted to cause a greater increase in process entropy. For a sequence containing multiple tasks, the state transition model is iteratively applied, i.e. the predicted termination entropy of the previous task becomes the predicted initial entropy of the next task, thus constructing a continuous predicted entropy trajectory throughout the entire sequence.

[0042] Furthermore, after the predicted entropy trajectories of all candidate sequences are generated, the module needs to quantitatively evaluate these trajectories representing different future possibilities. For this purpose, the module introduces the concept of sequence cost, which is used to represent the cumulative instability or health loss that will be caused to the machine tool by executing a certain sequence. Sequence cost It is calculated by integrating the predicted entropy trajectory over the total estimated time of the sequence, with the specific formula as follows: ; where, is the sequence cost of the candidate sequence ; is the time; is the total estimated time of executing the candidate sequence ; is the predicted entropy trajectory of the sequence at time . The physical meaning of this integral value is that it not only considers the peak value of entropy, but also considers the duration of high-entropy state, thus being able to more comprehensively evaluate the overall impact of the sequence.

[0043] Finally, the scheduling decision module performs the selection operation. It selects the one with the lowest sequence cost from all candidate sequences as the optimal sequence .

[0044] ; After determining the optimal sequence, the module dispatches the first task in the sequence to the controller of the machine tool for execution. At the same time, the module transmits the complete predicted entropy trajectory generated for the optimal sequence to the adaptive correction module as the basis for subsequent comparison and learning.

[0045] In summary, the scheduling decision module, through the construction of a decision-making process that generates candidate predicted trajectories, evaluates costs and selects the optimal one, changes the scheduling decision from a passive, rule-based dispatch based on history to an active, optimized process based on future state prediction. It quantifies and compares the potential impact of different processing paths on the physical state of the system, enabling the system to autonomously select a processing sequence that can both complete production tasks and maximize the stability of the system's health.

[0046] And an adaptive correction module, connected to the scheduling decision module and the process entropy real-time monitoring module, is configured to: calculate the trajectory deviation by comparing the predicted entropy trajectory of the optimal sequence with the actual process entropy trajectory, and correct the resource vector stored in the resource vector module when the trajectory deviation meets the preset conditions; In this embodiment, an automated scheduling and monitoring management system for a three-axis planar milling machine includes an adaptive correction module. This module plays a crucial role in achieving closed-loop learning and model self-evolution within the entire system. Its purpose is to systematically quantify and correct the deviation between predictions and reality, enabling the digital model within the system to continuously and automatically converge towards the true laws of the physical world.

[0047] The adaptive correction module maintains data connections with both the scheduling decision module and the real-time process entropy monitoring module, forming the information pathway required for its learning function. After a processing task is completed, this module receives the predicted entropy trajectory generated for that task from the scheduling decision module and the actual process entropy trajectory measured during processing from the real-time process entropy monitoring module.

[0048] After acquiring the two trajectories mentioned above, the adaptive correction module is configured to execute a precise, step-by-step correction process: First, the core task of this module is to quantify the prediction deviation. To do this, it calculates a scalar value, the trajectory deviation, by comparing the two trajectories point by point. This calculation aims to accumulate the absolute difference between the predicted and actual values ​​throughout the entire processing task. In a preferred embodiment, the trajectory deviation is calculated according to the following integral formula: ; in, This refers to the actual processing time of the task. The actual process entropy trajectory output by the real-time process entropy monitoring module; and This is the predicted entropy trajectory output by the scheduling decision module. The value, in its physical sense, represents the overall error in the system's internal model's understanding of the physical processes of the task. A larger one... The value indicates that the system significantly underestimates or overestimates the processing difficulty of the task or its impact on the machine tool status.

[0049] Secondly, the module includes a correction trigger mechanism. Not every minute deviation will trigger the correction procedure, to avoid the system overreacting to random noise. Instead, the module will calculate the trajectory deviation... Compare with a preset, dynamically adjustable error threshold. Only when When the error threshold is exceeded, the correction procedure is formally activated.

[0050] When the correction is triggered, the module performs its core parameter adjustment operation. Instead of the system's control algorithm itself, the module targets the underlying data stored in the resource vector module, i.e. the resource vector associated with the completed task. Specifically, the module will automatically fine-tune the tool wear factor and the quality risk coefficient in the resource vector according to the magnitude of the trajectory deviation. Its correction logic can be described by the following formula: ; ; In this correction formula: and are the new values of the corrected tool wear factor and quality risk coefficient, respectively; and are the old values before correction.

[0051] It is worth noting that the correction process introduces learning rate parameters and . The values of these two parameters are not fixed but are associated with the material information identifier in the task resource vector. The purpose of this design is to endow the system with differentiated learning capabilities. For example, when the system processes a new, attribute-unknown difficult-to-machine material, a higher learning rate can be configured for it, so that the system can quickly and significantly adjust the risk parameters in its resource vector according to the larger trajectory deviation through the processing of a few tasks, thereby quickly adapting to the processing characteristics of the new material. Conversely, for regular materials, a smaller learning rate can be used to achieve smooth and fine tuning.

[0052] Finally, after calculating the new resource vector parameters, the adaptive correction module writes these updated values back to the knowledge base of the resource vector module, overwriting the original parameters. Through this step, the slow learning loop of the entire system is closed. This means that the next time the scheduling decision module needs to evaluate a task containing the same or similar material, it will base its prediction of the entropy-increasing trajectory on this already corrected resource vector that is closer to the physical reality.

[0053] In summary, the adaptive correction module triggers the automatic process of adjusting the parameter update model through a quantitative deviation, enabling the entire system to learn from its own prediction errors. It acts as an internal calibration mechanism, ensuring that the system's decision-making foundation does not remain unchanged and become disconnected from the dynamically changing physical world, thereby achieving self-improvement and continuous evolution of the entire automated scheduling and monitoring management system.

[0054] Please refer to the attached drawings Figure 2 Another embodiment of the application provides a three-axis planer milling machine automation scheduling and monitoring management method, comprising the following steps: S1, defining and storing a resource vector containing physical attributes for each task to be processed; S2, collecting sensor signals of the machine tool during processing, and calculating an actual process entropy trajectory in real time based on the sensor signals; S3, generating a predicted entropy trajectory for multiple candidate sequences based on the resource vector and the current process entropy, and selecting an optimal sequence for processing according to the evaluation sequence cost; S4, and calculating the trajectory deviation by comparing the predicted entropy trajectory of the optimal sequence with the actual process entropy trajectory, and modifying the resource vector when the trajectory deviation meets the preset condition.

[0055] The embodiment method can be used to execute the above-mentioned system embodiment, and the principles and technical effects are similar, which will not be repeated here.

[0056] Although embodiments of the application have been shown and described, it is to be understood that the application is not limited to these embodiments. Since modifications, equivalents, replacements, and variations of these embodiments can be understood by those of ordinary skill in the art without departing from the principles and spirit of the application, the scope of the application is defined by the appended claims and their equivalents.

Claims

1. An automated scheduling and monitoring management system for a three-axis planar milling machine, characterized in that, include: The resource vector module is used to define and store a resource vector containing the physical attributes of each task to be processed. The real-time process entropy monitoring module is used to collect sensor signals from the machine tool during the processing and calculate an actual process entropy trajectory in real time based on these signals. The scheduling decision module, connected to the resource vector module and the real-time process entropy monitoring module, is configured to: generate predicted entropy trajectories for multiple candidate sequences based on the resource vector and the current process entropy, and select an optimal sequence for processing based on the evaluated sequence cost; An adaptive correction module, connected to the scheduling decision module and the process entropy real-time monitoring module, is configured to: calculate the trajectory deviation by comparing the predicted entropy trajectory of the optimal sequence with the actual process entropy trajectory, and correct the resource vector stored in the resource vector module when the trajectory deviation meets a preset condition.

2. The automated scheduling and monitoring management system for a three-axis planar milling machine according to claim 1, characterized in that, The resource vector includes a tool wear factor and a quality risk coefficient. The tool wear factor is used to characterize the intensity of the task's wear on tool life, and the quality risk coefficient is used to characterize the sensitivity of the task to the stability of the machine tool condition.

3. The automated scheduling and monitoring management system for a three-axis planar milling machine according to claim 1, characterized in that, By performing wavelet packet transform on the acquired sensor signals, the energy proportion of each frequency band is obtained, and the process entropy is calculated according to the following formula. : ； in, For time The process entropy; For frequency band index; Total number of frequency bands; For the first Each frequency band in time The proportion of energy.

4. The automated scheduling and monitoring management system for a three-axis planar milling machine according to claim 1, characterized in that, The scheduling decision module is configured to use the final prediction process entropy of the optimal sequence as the initial process entropy when generating the next candidate sequence, so as to realize the continuous state transmission of scheduling decisions.

5. The automated scheduling and monitoring management system for a three-axis planar milling machine according to claim 1, characterized in that, The scheduling decision module is configured to calculate the sequence cost of the candidate sequence by integrating the predicted entropy trajectory. The calculation formula is as follows: ； in, Candidate sequences The sequence cost; For time; For the candidate sequence Total estimated time; For the candidate sequence In time The predicted entropy trajectory.

6. The automated scheduling and monitoring management system for a three-axis planar milling machine according to claim 1, characterized in that, The adaptive correction module is configured to trigger a correction of the resource vector when the trajectory deviation exceeds a preset error threshold.

7. The automated scheduling and monitoring management system for a three-axis planar milling machine according to claim 2, characterized in that, The adaptive correction module corrects the resource vector by automatically adjusting the tool wear factor and / or the quality risk coefficient in the resource vector according to the magnitude of the trajectory deviation.

8. The automated scheduling and monitoring management system for a three-axis planar milling machine according to claim 7, characterized in that, When adjusting the tool wear factor and / or the quality risk coefficient, the adaptive correction module also introduces a learning rate parameter associated with the material information of the task to achieve differentiated learning for different material processing difficulties.

9. The automated scheduling and monitoring management system for a three-axis planar milling machine according to claim 1, characterized in that, After completing the correction of the resource vector, the adaptive correction module stores the updated resource vector in the resource vector module for use in subsequent scheduling decisions, thereby forming a closed loop of learning and optimization.

10. An automated scheduling and monitoring management method for a three-axis planar milling machine, comprising the following steps, according to any one of claims 1-9: S1. Define and store a resource vector containing the physical attributes for each task to be processed; S2. Collect sensor signals from the machine tool during the processing and calculate an actual process entropy trajectory in real time based on these signals. S3. Based on the resource vector and the current process entropy, generate predicted entropy trajectories for multiple candidate sequences, and select the optimal sequence for processing based on the evaluated sequence cost; S4. By comparing the predicted entropy trajectory of the optimal sequence with the actual process entropy trajectory, the trajectory deviation is calculated, and the resource vector is corrected when the trajectory deviation meets the preset conditions.

Images