Adaptive control method and system for a wire cutting apparatus

Abstract

This invention relates to the field of industrial automation control technology, specifically to an adaptive control method and system for wire cutting equipment. The method includes: acquiring the risk level of the wire cutting equipment at the current and multiple historical time points; inputting the risk level into an improved machine learning model; outputting the operating status of the wire cutting equipment at the next time point; and if the operating status indicates a fault, controlling the wire cutting equipment to stop operation and triggering an alarm. The improved machine learning model includes a loss function that is positively correlated with the absolute value of the difference between the risk level at the current and multiple historical time points and a set value; the risk level is positively correlated with the instantaneous discharge fluctuation and deterioration degree at each time point. This invention solves the problem of low accuracy in adaptive control.

Description

Technical Field

[0001] This invention relates to the field of industrial automation control technology. More specifically, this invention relates to an adaptive control method and system for wire cutting equipment. Background Technology

[0002] Wire electrical discharge machining (EDM), a key process in numerical control machining, is applied in metal mold manufacturing, precision parts machining, and complex structural component forming. During the process, the wire EDM equipment continuously generates pulsed discharges between the high-speed moving electrode wire and the workpiece, thereby achieving layer-by-layer material removal. Due to its advantages such as high machining accuracy, strong adaptability, and no need for cutting force, it has become an indispensable machining method in industrial manufacturing. However, EDM is inherently a highly unstable physical process, affected by various factors such as voltage, current, gap width, machining fluid flow state, chip removal efficiency, and workpiece material properties. In actual production, problems such as instantaneous short circuits, unstable discharge, wire fatigue, wire breakage, and deterioration of the machined surface roughness are prone to occur. Failure to identify and handle these anomalies in a timely manner will not only reduce machining efficiency but may also lead to workpiece scrap, equipment damage, and even safety risks.

[0003] To improve the stability of wire EDM equipment, existing technologies generally employ safety protection measures based on fixed thresholds, such as monitoring the instantaneous values ​​of discharge voltage and current, triggering shutdown or alarms when these values ​​exceed or fall below set ranges. However, these methods rely on human experience to determine the thresholds, which cannot adequately adapt to dynamic changes under different workpiece materials, processing paths, complex surfaces, or extreme processing parameters, often resulting in false alarms or missed alarms. Furthermore, fixed threshold methods cannot reflect trend changes in discharge quality; they can only identify extreme anomalies and cannot provide early warnings of discharge deterioration, leading to the continued risk of wire breakage during actual operation.

[0004] With the development of intelligent manufacturing, some studies have begun to try to use machine learning models to predict the operating status of wire cutting processes. However, traditional models often only input the original voltage or current signals and lack detailed descriptions of time series characteristics, cumulative effects of discharge degradation, and coupling relationships between parameters. Although some models introduce loss functions, they often use simple mean square error forms, which cannot highlight time slices with severe fault symptoms, nor can they adaptively weight the learning of the importance of different features, resulting in models that are not sensitive enough or have unstable training, thus leading to problems with low adaptive control accuracy. Summary of the Invention

[0005] To address the problem of low accuracy in adaptive control mentioned in the background art, the present invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides an adaptive control method for a wire cutting equipment, comprising: acquiring the risk level of the wire cutting equipment at the current and multiple historical time points; inputting the risk level into an improved machine learning model; outputting the operating state of the wire cutting equipment at the next time point; and if the operating state is a fault, controlling the wire cutting equipment to stop operating and triggering an alarm; wherein the improved machine learning model includes a loss function, the loss function being positively correlated with the absolute value of the difference between the risk level at the current and multiple historical time points and a set value; and the risk level being positively correlated with the instantaneous discharge fluctuation and deterioration degree at each time point.

[0007] The above technical solution inputs risk information from multiple historical time points of the wire EDM equipment into an improved machine learning model to achieve intelligent prediction of the operating status at the next time point. When a fault is predicted, the system can immediately trigger equipment shutdown and alarm, thus forming an active protection mechanism. This not only improves the accuracy and timeliness of predicting potential equipment faults, but also enhances the safety and reliability of the processing process, realizing intelligent adaptive control and dynamic risk management of the wire EDM equipment.

[0008] Furthermore, the loss function for: , For the first The level of risk at each point in time. This represents the total number of historical time points. To define the level of risk, This is the smoothing coefficient.

[0009] The aforementioned technical solution introduces adaptive normalized weighting based on risk values ​​into the loss function, enabling the model to pay more attention to time points with higher risk levels and significant deviations from the target risk during training, thereby enhancing its ability to learn key abnormal states. Weight normalization ensures the balance of the overall error distribution, preventing a single extreme value from excessively dominating training. Simultaneously, it allows the model to comprehensively consider the risk evolution trend of the global time series, exhibiting stronger sensitivity and discriminative ability towards continuous anomalies, thus improving the model's accuracy, stability, and practicality in predicting discharge process risks.

[0010] Furthermore, the first Risk level at each time point for: , For normalization function, For the first The degree of deterioration at each point in time, For the first Instantaneous discharge fluctuation at each time point, These are the preset hyperparameters.

[0011] The aforementioned technical solution combines the cumulative degree of deterioration with the current instantaneous fluctuations and introduces a normalization function to smoothly map the deterioration trend, enabling risk assessment to provide a continuous and sensitive response to changes in the discharge state. As the discharge state gradually deteriorates, the normalization function automatically amplifies the weight of risk indicators, thereby enhancing the ability to identify persistent anomalies; while when the state is normal or fluctuates slightly, the risk value remains stable, avoiding false alarms.

[0012] Furthermore, the first The degree of deterioration at each time point for: , For the first The degree of deterioration at each point in time, It is the hyperbolic tangent function. For the first Instantaneous discharge fluctuation at each time point.

[0013] The aforementioned technical solution combines current instantaneous fluctuations with historical cumulative deterioration levels through a recursive approach, and uses a nonlinear adjustment function to adaptively control the cumulative increase. This prevents minor fluctuations from being excessively amplified, while allowing sustained anomalies to accumulate and be reflected rapidly in the model. It can smoothly suppress the interference of random noise on deterioration indicators, while exhibiting a sensitive response to the continuous deterioration of the discharge state. This allows for more accurate capture of abnormal trends in the processing, providing a reliable basis for early warning of potential faults.

[0014] Furthermore, the first Instantaneous discharge fluctuation at each time point for: , For the first Voltage values ​​at each time point, For the first The first time node in the target voltage sequence Each voltage value The total number of voltage values ​​in the target voltage sequence. The median absolute deviation of the target voltage sequence. To preset the first hyperparameter, The Pearson correlation coefficient is used. , These represent the target voltage sequence and the target current sequence, respectively, which are obtained by using time nodes. It is constructed using the voltage and current values ​​within the historical time window set for the latest point.

[0015] The above technical solution achieves a more comprehensive and refined characterization of the discharge state by simultaneously considering the deviation between the current voltage and the historical voltage sequence, the overall fluctuation characteristics of the historical voltage sequence, and the dynamic correlation between voltage and current. This improves the sensitivity and accuracy of the discharge deterioration trend and provides a reliable basis for timely detection of potential faults in the processing.

[0016] Furthermore, it also includes standardizing the target voltage sequence and the target current sequence.

[0017] Furthermore, the operating status is divided into normal status and fault status, and the alarm is a buzzer alarm.

[0018] Furthermore, the machine learning model is a long short-term memory neural network model.

[0019] Furthermore, it also includes training the machine learning model, specifically: inputting the training set into the pre-built machine learning model for training; during the training process, calculating the loss between the output predicted value and the label; adjusting the model parameters using gradient descent to minimize the prediction error; iteratively adjusting the parameters of the machine learning model until the loss is less than a set value or the set number of training iterations is reached, and finally obtaining the trained machine learning model.

[0020] In a second aspect, the present invention provides an adaptive control system for a wire cutting device, including a memory and a processor, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, an adaptive control method for a wire cutting device as described in any one of the above claims is implemented.

[0021] The beneficial effects of this invention are as follows: This invention utilizes an improved machine learning model to comprehensively calculate risk indicators based on multi-dimensional information such as instantaneous discharge fluctuations and cumulative deterioration. By designing a reasonable loss function, it focuses on moments with high anomaly risk during training, enabling the model to sensitively and accurately capture abnormal trends in the discharge process. Once a potential fault is predicted, the system can immediately issue a stop command and trigger a buzzer alarm, effectively preventing equipment damage or workpiece scrap. Overall, this invention significantly improves the safety, reliability, and intelligence level of the wire EDM process, achieving real-time adaptive control and dynamic risk management of the equipment, providing strong technical support for ensuring processing quality and production safety. Attached Figure Description

[0022] Figure 1 This is a flowchart schematically illustrating an adaptive control method for a wire cutting device according to an embodiment of the present invention; Figure 2This is a schematic block diagram illustrating the structure of an adaptive control system for a wire cutting device according to an embodiment of the present invention. Detailed Implementation

[0023] An embodiment of an adaptive control method for a wire EDM machine.

[0024] like Figure 1 As shown in the flowchart, an adaptive control method for a wire EDM machine according to an embodiment of the present invention includes the following steps: S1: Obtain the risk level of the wire EDM equipment at the current and multiple historical time points.

[0025] In a preferred embodiment, the first Risk level at each time point for: , For normalization function, For the first The degree of deterioration at each point in time, For the first Instantaneous discharge fluctuation at each time point, These are the preset hyperparameters.

[0026] By normalizing and coupling the cumulative deterioration trend with the current fluctuation intensity, the risk output can exhibit a smooth and sensitive change according to the transition from normal to abnormal processing status. The normalization function significantly increases the weight when the deterioration reaches a critical range, thereby enhancing the amplification effect of the fluctuation signal on risk and making persistent anomalies easier to identify than random fluctuations; at the same time, it maintains a stable output without abrupt jumps when the status is normal. Therefore, the calculation of the risk level can more accurately reflect the true health status of the discharge process, improving the response speed and identification reliability of potential faults.

[0027] No. The degree of deterioration at each time point for: , For the first The degree of deterioration at each point in time, It is the hyperbolic tangent function. For the first Instantaneous discharge fluctuation at each time point.

[0028] By using a recursive structure to couple the current fluctuation status with the historical cumulative deterioration, and by adaptively adjusting the accumulation rate through a nonlinear function, minor fluctuations have only a limited impact, while continuous anomalies are rapidly amplified in the model. This enables sensitive capture of the process of discharge state evolving from normal to deterioration, which helps to accurately identify the cumulative characteristics of discharge anomalies and provide early warning of potential faults.

[0029] No. Instantaneous discharge fluctuation at each time point for: , For the first Voltage values ​​at each time point, For the first The first time node in the target voltage sequence Each voltage value The total number of voltage values ​​in the target voltage sequence. The median absolute deviation of the target voltage sequence. To preset the first hyperparameter, The Pearson correlation coefficient is used. , These represent the target voltage sequence and the target current sequence, respectively, which are obtained by using time nodes. It is constructed using the voltage and current values ​​within a historical time window set for the latest point. It also includes standardizing the target voltage and current sequences using Z-score standardization.

[0030] By simultaneously measuring the degree of discharge fluctuation at the current moment, the overall dispersion of the historical voltage sequence, and the changes in the correlation between voltage and current, a more comprehensive and sensitive characterization of the discharge state is achieved. It not only reflects the deviation between the current signal and historical signals, but also utilizes the steady-state characteristics of the historical sequence and the dynamic relationship between voltage and current to cross-validate anomalies. This prevents noise disturbances or short-term fluctuations from causing misjudgments, while significantly amplifying persistent discharge anomalies. This improves the ability to identify and predict discharge deterioration trends, helping to detect potential faults in the processing process earlier and more reliably.

[0031] S2: Input the risk level into the improved machine learning model and output the operating status of the wire cutting equipment at the next time node.

[0032] In a preferred embodiment, the improved machine learning model includes a loss function, the loss function for: , For the first The level of risk at each point in time. This represents the total number of historical time points. To define the level of risk, The normalized weights serve as a smoothing coefficient. By introducing an adaptive weighting mechanism based on risk level into the loss function, the model can automatically focus on key moments with higher risk values ​​and larger deviations during training, thereby enhancing its ability to learn from real abnormal states. The addition of normalized weights makes the overall error distribution smoother, preventing errors from excessively dominating training at certain moments. It also allows the model to balance the difference between risk trends and targets globally, making the training process more sensitive to abnormal changes and more robust to noise interference. Therefore, the model can more accurately depict the risk evolution of the discharge process, improving the reliability and practicality of anomaly prediction.

[0033] The machine learning model is a Long Short-Term Memory (LSTM) artificial neural network model. Further, the machine learning model is trained, exemplarily using the LTM artificial neural network model as an example. Specifically: First, a training set is constructed and obtained, containing the risk levels of multiple current and historical time points calculated above, with labels manually set. The training set is then input into the pre-constructed LTM artificial neural network model for training. During training, the loss between the output predicted value and the label is calculated; the model parameters are adjusted using gradient descent to minimize the prediction error; the parameters of the LTM artificial neural network model are iteratively adjusted until the loss is less than a set value or a set number of training iterations are reached, ultimately obtaining the trained LTM artificial neural network model.

[0034] S3: If the operating status is faulty, the wire cutting equipment will stop operating and an alarm will be triggered.

[0035] In a preferred embodiment, the operating status is divided into two categories: normal status and fault status. The normal status indicates that the electrical discharge machining process has not experienced abnormal fluctuations and can continue to operate stably. The fault status indicates that the machining process has abnormal signs such as discharge deterioration, short circuit risk, wire breakage tendency, or other abnormal signs that may damage the machining quality and equipment safety. Once a fault is identified, the control system immediately issues a stop command to the wire EDM equipment, causing the equipment to stop working in the shortest possible time to avoid further expansion of the fault or damage to the equipment, scrapping of the machined parts, etc.

[0036] In a preferred embodiment, to ensure that operators are aware of equipment malfunctions immediately, an alarm is triggered simultaneously with equipment shutdown. The alarm is a buzzer alarm, which emits a continuous or intermittent high-frequency warning sound to remind on-site personnel to check the equipment status promptly. Buzzer alarms offer advantages such as fast response, high recognition rate, and no need for additional display equipment, providing reliable safety alerts in complex industrial environments. The linkage design between equipment shutdown and audible alarm effectively improves the safety and stability of wire EDM processing, reduces accidental damage, and ensures the processing flow operates within a controllable range, thereby significantly improving the overall reliability and practicality of the system.

[0037] The present invention enables intelligent prediction of the operating status at the next time point and can immediately trigger equipment shutdown and alarm when a fault is predicted, thus forming an active protection mechanism. Through comprehensive analysis of the risk level, instantaneous discharge fluctuations, and cumulative deterioration trends, the model can accurately capture minute anomalies in the discharge process and their cumulative effects. The loss function design makes the training process more sensitive to key anomaly moments, thereby improving the model's ability to identify potential faults. Combining the temporal feature learning ability of the Long Short-Term Memory Neural Network and the real-time linkage of the system to equipment control, the present invention not only provides early warning of potential equipment anomalies, reducing the risk of wire breakage and processing defects, but also significantly improves the safety, stability, and reliability of the processing, realizing intelligent adaptive control and dynamic risk management of wire EDM equipment.

[0038] An embodiment of an adaptive control system for a wire EDM machine: like Figure 2 As shown in the figure, an adaptive control system of a wire cutting device according to an embodiment of the present invention includes a processor and a memory.

[0039] This invention also provides an adaptive control system for a wire cutting machine. For example... Figure 2 As shown, the system includes a processor and a memory, the memory storing computer program instructions, which, when executed by the processor, implement an adaptive control method for a wire cutting device according to the present invention.

[0040] The adaptive control system of the wire cutting equipment also includes other components well known to those skilled in the art, such as communication interfaces. The settings and functions of these components are known in the art and will not be described in detail here.

[0041] In this invention, the aforementioned memory can be any tangible medium containing or storing a program that can be used or combined with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium can be any suitable magnetic or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc., or any other medium that can be used to store desired information and can be accessed by an application, module, or both. Any such computer storage medium can be part of a device or accessible to or connected to a device. Any application or module described in this invention can be implemented using computer-readable / executable instructions stored or otherwise maintained by such a computer-readable medium.

[0042] In the description of this specification, "multiple" or "several" means at least two, such as two, three or more, unless otherwise explicitly specified.

[0043] While this specification has shown and described numerous embodiments of the invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in the practice of this invention.

Claims

1. An adaptive control method for a wire cutting machine, characterized in that, include: The risk level of the wire cutting equipment at the current and multiple historical time points is obtained. The risk level is input into an improved machine learning model, which outputs the operating status of the wire cutting equipment at the next time point. If the operating status is a fault, the wire cutting equipment is controlled to stop running and an alarm is triggered. The improved machine learning model includes a loss function that is positively correlated with the absolute value of the difference between the risk level at the current and multiple historical time points and a set value. The degree of risk is positively correlated with the instantaneous discharge fluctuation and the degree of deterioration at each time point.

2. The adaptive control method for a wire cutting machine according to claim 1, characterized in that, The loss function for: , For the first The level of risk at each point in time. This represents the total number of historical time points. To define the level of risk, This is the smoothing coefficient.

3. The adaptive control method for a wire cutting machine according to claim 1, characterized in that, No. Risk level at each time point for: , For normalization function, For the first The degree of deterioration at each point in time, For the first Instantaneous discharge fluctuation at each time point, These are the preset hyperparameters.

4. The adaptive control method for a wire cutting machine according to claim 1, characterized in that, No. The degree of deterioration at each time point for: , For the first The degree of deterioration at each point in time, It is the hyperbolic tangent function. For the first Instantaneous discharge fluctuation at each time point.

5. The adaptive control method for a wire cutting machine according to claim 1, characterized in that, No. Instantaneous discharge fluctuation at each time point for: , For the first Voltage values ​​at each time point, For the first The first time node in the target voltage sequence Each voltage value The total number of voltage values ​​in the target voltage sequence. The median absolute deviation of the target voltage sequence. To preset the first hyperparameter, The Pearson correlation coefficient is used. , These represent the target voltage sequence and the target current sequence, respectively, which are obtained by using time nodes. It is constructed using the voltage and current values ​​within the historical time window set for the latest point.

6. The adaptive control method for a wire cutting machine according to claim 5, characterized in that, It also includes standardizing the target voltage sequence and the target current sequence.

7. The adaptive control method for a wire cutting machine according to claim 1, characterized in that, The operating status is divided into normal status and fault status, and the alarm is a buzzer alarm.

8. The adaptive control method for a wire cutting machine according to claim 1, characterized in that, The machine learning model is a long short-term memory neural network model.

9. The adaptive control method for a wire cutting machine according to claim 1, characterized in that, This also includes training machine learning models, specifically: The training set is fed into a pre-built machine learning model for training. During the training process, the loss between the output prediction value and the label is calculated. Use gradient descent to adjust model parameters to minimize prediction error; The parameters of the machine learning model are iteratively adjusted until the loss is less than a set value or the set number of training iterations are reached, and finally a well-trained machine learning model is obtained.

10. An adaptive control system for a wire EDM machine, characterized in that, It includes a memory and a processor, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, an adaptive control method for a wire cutting device as described in any one of claims 1 to 9 is implemented.

Images