Plate cutting energy efficiency control method based on one-control multi-drive architecture

Abstract

This invention provides an energy efficiency control method for sheet metal cutting based on a one-control, multi-drive architecture, belonging to the field of sheet metal cutting control technology. The method includes: collecting and preprocessing the structural parameters and cutting process parameters of the sheet metal cutting equipment; establishing an energy efficiency loss model, simulating energy distribution, and constructing a dynamic simulation model by combining the law of conservation of energy and material cutting theory; identifying energy efficiency risk areas; collecting operational data for these areas, constructing and optimizing an energy efficiency prediction model, and outputting predicted energy efficiency data for the next moment; setting adjustment and warning thresholds to trigger dynamic adjustments and alarms for process parameters, respectively. This invention achieves a refined description of energy distribution by constructing two types of models, proactively predicts energy efficiency problems through risk area identification, and combines a high-precision prediction model with a hierarchical control strategy, balancing energy efficiency optimization and risk response. It can be widely applied to sheet metal cutting processes in multiple industries, helping enterprises save energy, reduce consumption, and upgrade their intelligent systems.

Description

Technical Field

[0001] This invention belongs to the field of sheet metal cutting control technology, specifically a sheet metal cutting energy efficiency control method based on a one-control-multiple-drive architecture. Background Technology

[0002] Sheet metal cutting, as an indispensable basic processing step in the manufacturing industry chain, is widely used in various industries such as metal processing, furniture manufacturing, and building materials. Its production efficiency and energy consumption level are directly related to the overall production cost and green development capability of enterprises. With the continuous expansion of industrial production scale and the increasing market demand for processing precision, multi-unit collaborative cutting equipment has gradually become the industry mainstream. While improving processing capabilities, this type of equipment also brings more complex energy management issues.

[0003] The energy efficiency management of existing cutting equipment generally suffers from significant limitations. Traditional control methods struggle to achieve efficient coordination between different work units, easily leading to power imbalances and severe energy waste. Most equipment uses uniform operating parameter settings, failing to adapt to dynamic changes during processing. This results in substantial ineffective energy consumption at different processing stages and in different processing areas, leading to overall low energy utilization efficiency. Furthermore, existing energy efficiency monitoring and management are mostly based on post-processing statistics, only summarizing and analyzing overall energy consumption data after processing is complete. They cannot detect real-time changes in energy distribution during processing, nor can they proactively identify potential energy efficiency anomalies, making it difficult to effectively prevent increased equipment wear and reduced processing quality caused by energy efficiency abnormalities. Summary of the Invention

[0004] This invention provides a method for controlling the energy efficiency of sheet metal cutting based on a one-control, multi-drive architecture, in order to overcome the deficiencies in the existing technology.

[0005] This invention provides a method for controlling the energy efficiency of sheet metal cutting based on a one-control, multi-drive architecture, comprising: The structural parameters and cutting process parameters of the sheet metal cutting equipment are collected in real time, and the structural parameters and cutting process parameters are preprocessed to obtain preprocessed data.

[0006] Based on the preprocessed data, combined with the law of conservation of energy and the theory of material cutting, an energy efficiency loss model is established to simulate the energy distribution during the cutting process, and a dynamic simulation model for plate cutting is constructed.

[0007] Based on the energy distribution during the cutting process and the structural parameters, areas prone to abnormal energy loss are identified as energy efficiency risk areas.

[0008] Real-time operational data of energy efficiency risk areas are collected, an energy efficiency prediction model based on graph attention neural network is constructed, and the hyperparameters of the prediction model are optimized using particle swarm optimization algorithm. Input preprocessed data and real-time operational data of energy efficiency risk areas, and output the predicted energy efficiency data of energy efficiency risk areas at the next time step.

[0009] Based on the performance parameters of the sheet metal cutting equipment, adjustment thresholds and warning thresholds for energy efficiency risk areas are set. When the predicted energy efficiency data reaches the adjustment threshold, an adaptive adjustment mechanism is introduced to dynamically adjust the cutting process parameters. When the predicted energy efficiency reaches the warning threshold, an alarm is triggered.

[0010] The energy efficiency control method for sheet metal cutting based on a one-control, multi-drive architecture provided by this invention includes structural parameters such as geometric data of the cutting equipment and electromechanical characteristic parameters. The electromechanical characteristic parameters include the rated power of the motor, transmission efficiency coefficient, and cutting head loss coefficient. The cutting process parameters include cutting speed, cutting depth, feed rate, real-time power data of the multi-drive motors, and operating status data of each transmission mechanism.

[0011] The plate cutting energy efficiency control method based on a one-control, multi-drive architecture provided by the present invention includes the following preprocessing steps for structural parameters and cutting process parameters: A wavelet thresholding denoising algorithm is used to remove random noise from structural parameters and cutting process parameters, and the Z-score method is used to identify and remove outliers in structural parameters and cutting process parameters.

[0012] The moving average method is used to smooth the structural parameters and cutting process parameters, and data normalization is used to convert the structural parameters and cutting process parameters to a uniform range.

[0013] Interpolation was used to process the missing values ​​in the structural parameters and cutting process parameters to obtain a complete dataset, which was then used as preprocessing data.

[0014] The energy efficiency control method for plate cutting based on a one-control, multi-drive architecture provided by the present invention includes the following process for simulating energy distribution during the cutting process: Based on the material cutting theory, a material removal energy consumption model is established during the cutting process.

[0015] Based on the law of conservation of energy, the cutting process is decomposed into three components: cutting deformation energy, friction energy, and thermal energy, and an energy efficiency loss calculation model is established.

[0016] Set boundary conditions and initial conditions. Boundary conditions include power allocation at the cutting start point and energy recovery at the cutting end point.

[0017] The energy efficiency loss model is solved using the finite volume method to obtain the energy distribution during the cutting process.

[0018] According to the plate cutting energy efficiency control method based on a one-control-multi-drive architecture provided by the present invention, the process of constructing a dynamic simulation model for plate cutting includes: Set up the basic framework of the 3D simulation model based on the preprocessed data.

[0019] By combining the law of conservation of energy and the theory of multi-drive synergy, a set of equations describing the change of cutting energy efficiency over time is established. The set of equations includes power distribution equations and energy efficiency evolution equations.

[0020] The energy distribution during the cutting process is used as the initial standard for the three-dimensional simulation model, and the power distribution of each driving unit is used as the boundary standard.

[0021] Set the time step to perform dynamic simulation on the basic 3D simulation model.

[0022] By solving the equations using the finite volume method, time-series data of cutting energy efficiency are generated, resulting in a dynamic simulation model for plate cutting.

[0023] The energy efficiency control method for plate cutting based on a one-control, multi-drive architecture provided by the present invention includes the following process for identifying energy efficiency risk areas: The dynamic simulation model for cutting sheet metal is divided into regions.

[0024] Real-time energy efficiency data of each region in the dynamic simulation model of plate cutting is collected, and energy efficiency time-series curves of each region and spatial distribution map of the entire cutting surface are generated.

[0025] Based on the multi-drive synergy theory and spatial distribution map, power allocation status data in each region is obtained.

[0026] An energy efficiency risk coefficient calculation model is established, and the energy efficiency risk coefficient for each region is obtained based on the average energy efficiency loss, peak energy efficiency loss, power distribution status data and structural parameters of each region.

[0027] Areas where the energy efficiency risk coefficient reaches a preset threshold are designated as energy efficiency risk areas.

[0028] According to the plate cutting energy efficiency control method based on a one-control-multi-drive architecture provided by the present invention, the process of constructing an energy efficiency prediction model based on a graph attention neural network includes: Collect historical structural parameters, historical cutting process parameters, and historical energy efficiency data of energy efficiency risk areas of the sheet metal cutting equipment, as well as the energy efficiency data of the energy efficiency risk areas at the next moment corresponding to the historical energy efficiency data.

[0029] The collected historical structural parameters, historical cutting process parameters, historical energy efficiency data of energy efficiency risk areas, and energy efficiency data of energy efficiency risk areas at the next moment are preprocessed and divided into training set and test set.

[0030] A basic graph attention neural network model is constructed, comprising an input layer, a graph attention layer, a pooling layer, a fully connected layer, and an output layer. The input layer receives preprocessed input data, including historical structural parameters, historical cutting process parameters, and historical energy efficiency data for energy efficiency risk areas. The graph attention layer includes multiple attention heads for spatial feature extraction and multi-drive correlation modeling of the input data, generating feature maps. The pooling layer reduces the spatial dimensionality of the feature maps. The fully connected layer integrates the extracted features. The output layer outputs the prediction result based on the integrated result.

[0031] The basic model of the graph attention neural network is trained using the training set, and the model parameters that meet the prediction accuracy are retained to obtain the energy efficiency prediction model.

[0032] The energy efficiency control method for plate cutting based on a one-control, multi-drive architecture provided by the present invention includes the following process for optimizing the hyperparameters of the prediction model using a particle swarm optimization algorithm: Define a hyperparameter space, with the learning rate, number of graph attention heads, hidden layer dimension, and regularization parameter as the hyperparameters to be optimized, and set a range of values ​​for each hyperparameter.

[0033] Initialize the particle swarm, where each particle represents a set of hyperparameter combinations, and randomly assign an initial position and initial velocity to each particle.

[0034] The position of each particle is used as a hyperparameter of the prediction model, and the corresponding objective function value is calculated. The objective function value is the energy efficiency prediction error.

[0035] The individual optimal position and the global optimal position are updated based on the objective function value of each particle.

[0036] The velocity and position of each particle are adjusted according to the particle swarm update formula.

[0037] The process of calculating the objective function and updating the particle position is repeated until the preset number of iterations is reached. The hyperparameter combination represented by the global optimal position is then used as the hyperparameter combination of the prediction model.

[0038] The energy efficiency control method for plate cutting based on a one-control, multi-drive architecture provided by the present invention includes the following process for setting adjustment thresholds and early warning thresholds for energy efficiency risk areas: Collect the performance parameters of the sheet metal cutting equipment, including the maximum allowable energy loss, minimum energy efficiency, and expected normal energy efficiency range.

[0039] The expected normal energy efficiency range is used as the adjustment threshold, and the maximum allowable energy loss and minimum energy efficiency utilization rate are used as the warning threshold.

[0040] According to the plate cutting energy efficiency control method based on a one-control, multi-drive architecture provided by the present invention, the process of dynamically adjusting the cutting process parameters includes: Optimize the power distribution ratio of multi-drive motors to reduce ineffective losses in the transmission system.

[0041] Adjusting the matching relationship between cutting speed and feed rate improves the energy utilization rate of material removal.

[0042] Adjust the working sequence of each cutting head to achieve multi-drive collaborative operation and reduce overall energy consumption.

[0043] This invention provides a plate cutting energy efficiency control method based on a one-control, multi-drive architecture. Through the construction of an energy efficiency loss model and a dynamic simulation model, it achieves a refined description of the energy distribution during the cutting process. The energy efficiency loss model, established by combining the law of conservation of energy and material cutting theory, can accurately quantify the contribution ratio of each energy component during the cutting process. The plate cutting dynamic simulation model intuitively demonstrates the spatial and temporal distribution characteristics of energy efficiency through numerical simulation, providing a scientific basis for identifying energy efficiency risk areas. The energy efficiency risk area identification mechanism enables proactive prediction of potential energy efficiency problems. Through area division, energy efficiency data collection, and risk coefficient calculation, areas prone to abnormal energy efficiency losses can be accurately located, transforming energy efficiency management from passive response to proactive prevention, effectively avoiding equipment failures and energy waste caused by abnormal energy efficiency. A graph attention neural network-based energy efficiency prediction model combined with particle swarm optimization algorithm achieves high-precision energy efficiency prediction. The hierarchical control strategy ensures energy efficiency optimization during normal operation and effectively addresses energy efficiency risks under extreme conditions, enabling its widespread application in plate cutting processes across multiple industries such as metal processing, furniture manufacturing, and building materials, providing effective technical support for energy conservation, consumption reduction, and intelligent upgrading for enterprises. Attached Figure Description

[0044] The invention will now be further described with reference to the accompanying drawings.

[0045] Figure 1 This is a flowchart illustrating the plate cutting energy efficiency control method based on a one-control-multiple-drive architecture in this invention. Figure 2 This is a schematic diagram of the energy distribution during the simulated cutting process in this invention; Figure 3 This is a schematic diagram of the process for constructing a dynamic simulation model for plate cutting in this invention; Figure 4 This is a schematic diagram of the process for identifying energy efficiency risk areas in this invention. Detailed Implementation

[0046] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0047] like Figures 1 to 4 As shown in the embodiment of the present invention, the plate cutting energy efficiency control method based on a one-control, multi-drive architecture includes: The structural parameters and cutting process parameters of the sheet metal cutting equipment are collected in real time. Specifically, a communication connection with each drive unit of the sheet metal cutting equipment needs to be established through an industrial Ethernet bus. The communication protocol adopts the PROFINET or EtherCAT standard to ensure that the data transmission delay does not exceed 1 millisecond. At the same time, the static structural parameter document of the equipment is retrieved through the RESTful API interface of the equipment management system. The document format is XML or JSON, which contains the geometric structure data and electromechanical characteristic calibration values ​​of the equipment at the time of manufacture. The structural parameters and cutting process parameters are then preprocessed to obtain preprocessed data.

[0048] Structural parameters include geometric data and electromechanical characteristic parameters of the cutting equipment. The geometric data is acquired using a 3D laser scanning device and includes the dimensions of the cutting platform (length, width, and height accurate to 0.1 mm), the stroke range of each drive shaft, the installation coordinates of the cutting head, and the geometric transmission ratio parameters of each transmission mechanism. Electromechanical characteristic parameters include the motor's rated power, transmission efficiency coefficient, and cutting head loss coefficient. The motor's rated power is read from the motor's nameplate parameters. The transmission efficiency coefficient is determined through no-load testing. The cutting head loss coefficient is calibrated based on the cutting head's model and usage time. Cutting process parameters include cutting speed, cutting depth, feed rate, real-time power data of the multi-drive motors, and operating status data of each transmission mechanism. The real-time power data of the multi-drive motors is sampled at a frequency of no less than 100 Hz, with independent acquisition for each drive unit. The operating status data of each transmission mechanism includes discrete parameters such as running / stop status, fault codes, and temperature values.

[0049] The preprocessing of structural parameters and cutting process parameters includes: A wavelet thresholding denoising algorithm is used to remove random noise from structural parameters and cutting process parameters. The Daubechies4 wavelet is selected as the mother wavelet, and the decomposition level is set to 5 levels. The threshold is calculated using a general thresholding method, and the threshold formula is as follows:

[0050] In the formula, For the threshold, This is an estimate of the noise standard deviation. Given the signal length, wavelet coefficients are processed using a soft thresholding function; the Z-score method is used to identify and remove outliers in structural and cutting process parameters, and the Z-score value for each data point is calculated using the following formula:

[0051] In the formula, The standardized values ​​of the data points. For the original data points, The mean of the dataset. To determine the standard deviation of the dataset, data points with an absolute Z-score greater than 3 were marked as outliers and removed. A moving average method was used to smooth the structural and cutting process parameters. The window size was set to 5 to 15 data points, adaptively adjusted according to the degree of data fluctuation. A larger window was used for power data with large fluctuations, while a smaller window was used for relatively stable geometric parameters. Data normalization was used to transform the structural and cutting process parameters to a uniform range. A min-max normalization method was used to map all parameters to the [0,1] interval to ensure the comparability of parameters with different dimensions.

[0052] Interpolation methods were used to process missing values ​​in structural and cutting process parameters. Cubic spline interpolation was used for time series data, and Kriging interpolation was used for spatially distributed data. The time interval between the interpolated data points and the original data points was kept consistent to obtain a complete dataset. The complete dataset was used as preprocessed data and stored in HDF5 format. Each dataset contained metadata such as data acquisition timestamp, parameter name, value, and data quality label.

[0053] Based on the preprocessed data, combined with the law of conservation of energy and the theory of material cutting, an energy efficiency loss model is established to simulate the energy distribution during the cutting process, and a dynamic simulation model for plate cutting is constructed.

[0054] The process of simulating energy distribution during cutting includes: Based on the material cutting theory, a material removal energy consumption model is established during the cutting process, considering three main energy consumption components: shear deformation, plowing effect, and friction effect. Shear deformation energy is proportional to the shear strength and cutting cross-sectional area of ​​the material, plowing effect energy is related to the cutting edge radius of the tool and the hardness of the material, and friction energy is related to the tool-workpiece contact area and the friction coefficient.

[0055] Based on the law of conservation of energy, the cutting process is decomposed into three components: cutting deformation energy, friction energy, and heat energy. An energy efficiency loss calculation model is established, and the energy efficiency loss calculation formula is as follows:

[0056] In the formula, This represents the total energy loss at time t; This indicates the total number of drive units, which is determined based on the device configuration. This represents the input power of the i-th driving unit at time t; This represents the transmission efficiency coefficient of the i-th drive unit; This represents the material cutting energy of the i-th cutting region, which is determined based on the characteristics of the material being cut, such as approximately 2.2 for low-carbon steel and approximately 3.5 for stainless steel. This represents the cutting speed at time t; This represents the cutting depth at time t.

[0057] Set boundary conditions and initial conditions. Boundary conditions include power distribution at the cutting start point. The initial power distribution ratio of each drive unit is optimized according to the number and position of the cutting head to ensure power balance at the initial moment. Energy recovery at the cutting end point is achieved by converting the kinetic energy of the motor during deceleration into electrical energy and feeding it back to the grid through the braking resistor.

[0058] The energy efficiency loss model is solved using the finite volume method. The cutting region is discretized into several control volumes, each with dimensions of 1 mm × 1 mm × 1 mm and a time step of 0.01 seconds. The energy balance equation for each control volume is solved iteratively. The energy balance equation is as follows:

[0059] In the formula, Indicates the density of the material; This indicates the specific heat capacity of a material at constant pressure. This represents the rate of change of temperature over time. Indicates the control volume; This indicates that the heat flux density of the volume surface is controlled; Indicates the area of ​​the control volume surface; This indicates the energy generation rate within the control volume; This represents the energy loss rate within the control volume. The energy distribution during the cutting process is obtained, and the energy distribution data includes the energy density value at each spatial location and at each time point.

[0060] The process of constructing a dynamic simulation model for sheet metal cutting includes: Based on the preprocessed data, a three-dimensional simulation basic model framework is set up. A three-dimensional geometric model of the cutting equipment is established using finite element analysis software. The model includes main components such as frame, transmission system, cutting head, and worktable. The material properties of each component are selected from the material library, including parameters such as density, elastic modulus, and Poisson's ratio.

[0061] Combining the law of conservation of energy and the theory of multi-drive synergy, a set of equations describing the change of cutting energy efficiency over time is established. The set of equations includes a power distribution equation and an energy efficiency evolution equation. The power distribution equation is as follows:

[0062] In the formula, This represents the total input power at time t; Indicates the total number of drive units; This represents the input power of the i-th driving unit at time t.

[0063] The energy efficiency evolution equation is:

[0064] In the formula, This indicates the rate of change of energy efficiency loss over time; This represents the partial derivative of energy efficiency loss with respect to the power of the i-th drive unit; This represents the rate of change of the power of the i-th drive unit over time; This represents the partial derivative of energy loss with respect to cutting speed; This represents the rate of change of cutting speed over time. This represents the partial derivative of energy loss with respect to cutting depth; This represents the rate of change of cutting depth over time.

[0065] The power distribution equation describes the distribution of total input power among the driving units, satisfying that the sum of the power of each driving unit equals the total input power. The energy efficiency evolution equation describes the dynamic characteristics of energy loss as the cutting parameters change, and is in the form of a first-order differential equation. The energy distribution during the cutting process is used as the initial standard of the three-dimensional simulation model, and the power distribution of each driving unit is used as the boundary standard. Energy loads and power boundary conditions are applied to the corresponding nodes of the simulation model.

[0066] Set the time step to perform dynamic simulation on the 3D simulation base model. The time step is adaptively adjusted according to the cutting speed. The faster the cutting speed, the smaller the time step, ranging from 0.001 seconds to 0.01 seconds, to ensure the numerical stability of the simulation process.

[0067] The equations are solved using the finite volume method to generate time-series data on cutting energy efficiency. The data sampling interval is consistent with the time step, resulting in a dynamic simulation model for plate cutting. The model output includes key parameters such as the real-time power of each drive unit, energy efficiency loss in each region, and temperature distribution of the cutting head.

[0068] Based on the energy distribution during the cutting process and the structural parameters, areas prone to abnormal energy loss are identified as energy efficiency risk areas.

[0069] The process of identifying energy efficiency risk areas includes: The dynamic simulation model of plate cutting is divided into regions. The entire cutting surface is divided into several rectangular regions using a mesh division method. The size of each region is set to 5 mm × 5 mm to 20 mm × 20 mm according to the cutting accuracy requirements. When dividing the regions, it is ensured that each region contains the working range of at least one driving unit.

[0070] The system collects energy efficiency data from each region of the dynamic simulation model for cutting sheet metal in real time. It continuously collects the energy efficiency loss value of each region in units of time step, and generates energy efficiency time series curves for each region. The horizontal axis of the curve represents time and the vertical axis represents the energy efficiency loss value. At the same time, it generates a spatial distribution map of the entire cutting surface and uses a heat map to display the energy efficiency loss level of different regions. The color from blue to red indicates that the energy efficiency loss is from low to high.

[0071] Based on the multi-drive cooperative theory and spatial distribution map, power distribution state data in each region is obtained. The power superposition effect in the region where multiple drive units act simultaneously is analyzed, and the power overlap coefficient is calculated. The formula is as follows:

[0072] In the formula, The power overlap factor is... For the total power of the region, This is the sum of the power of each drive unit when it operates individually. A coefficient greater than 1 indicates the presence of a power superposition effect.

[0073] An energy efficiency risk coefficient calculation model is established. Based on the average energy efficiency loss, peak energy efficiency loss, power distribution status data, and structural parameters of each region, the energy efficiency risk coefficient for each region is obtained. The formula is as follows:

[0074] In the formula, Energy efficiency risk coefficient, This represents the regional average energy efficiency loss. This is the baseline value for normal energy efficiency loss. For regional peak energy efficiency loss, To the maximum permissible energy efficiency loss, , , For the weighting coefficients, satisfying The area where the energy efficiency risk coefficient reaches the preset risk coefficient threshold is designated as the energy efficiency risk area. The risk coefficient threshold is set to 0.7, which can be adjusted according to the operating status and maintenance needs of the equipment.

[0075] Real-time operational data of energy efficiency risk areas is collected through power sensors, position sensors, and temperature sensors installed on each drive unit. An energy efficiency prediction model based on graph attention neural network is constructed, and the hyperparameters of the prediction model are optimized using particle swarm optimization algorithm. The input is preprocessed data and real-time operational data of energy efficiency risk areas, and the output is the predicted energy efficiency data of the energy efficiency risk areas at the next moment.

[0076] The process of constructing an energy efficiency prediction model based on a graph attention neural network includes: Collect historical structural parameters, historical cutting process parameters, and historical energy efficiency data of energy efficiency risk areas of the sheet metal cutting equipment, as well as the energy efficiency data of the energy efficiency risk areas at the next moment corresponding to the historical energy efficiency data. The historical data should span no less than 3 months and contain no less than 100,000 data entries, including operating data under different cutting materials, different cutting parameters, and different equipment conditions.

[0077] The collected historical structural parameters, historical cutting process parameters, historical energy efficiency data of energy efficiency risk areas, and energy efficiency data of energy efficiency risk areas at the next moment are preprocessed. The preprocessing method is the same as the aforementioned data preprocessing method, and the data is divided into training set and test set with a ratio of 7:3.

[0078] A basic graph attention neural network model is constructed, including an input layer, a graph attention layer, a pooling layer, a fully connected layer, and an output layer. The input layer receives preprocessed input data, which includes historical structural parameters, historical cutting process parameters, and historical energy efficiency data of energy efficiency risk areas. The input feature dimension is set to 64, and historical data from 10 consecutive time steps are input at each time step. The graph attention layer includes multiple attention heads, with the number of attention heads set to 4 to 8, which are used to extract spatial features from the input data and model multi-drive associations. Each attention head calculates the association weights between different driving units through an attention mechanism to generate a feature map.

[0079] Pooling layers are used to reduce the spatial dimension of the feature maps. Average pooling is used, and the pooling window size is 2×2. Fully connected layers are used to integrate the extracted features. They contain 2 to 3 hidden layers, each with 128 to 512 neurons. The activation function used is ReLU.

[0080] The output layer is used to output prediction results based on the integration results. The output dimension is the number of energy efficiency risk areas, and the activation function is a linear function.

[0081] The graph attention neural network base model is trained using the training set. The Adam optimizer is used, the batch size is set to 32 to 128, and the number of training rounds is set to 100 to 500. An early stopping mechanism is used to prevent overfitting. Training is stopped when the validation set loss no longer decreases for 10 consecutive rounds. The model parameters that meet the prediction accuracy are retained. The prediction accuracy requires that the average relative error does not exceed 5%, thus obtaining the energy efficiency prediction model.

[0082] The process of optimizing the hyperparameters of a prediction model using the particle swarm optimization algorithm includes: Define a hyperparameter space, with the learning rate, number of graph attention heads, hidden layer dimension, and regularization parameter as the hyperparameters to be optimized, and set a value range for each hyperparameter. The learning rate ranges from 0.0001 to 0.01 and adopts a logarithmic uniform distribution.

[0083] The number of attention points in the graph ranges from 4 to 16, with a step size of 2; the dimensions of the hidden layers range from 64 to 512, with a step size of 32.

[0084] The regularization parameter ranges from 0.0001 to 0.01 and uses a logarithmic uniform distribution.

[0085] Initialize the particle swarm, with the number of particles set to 20 to 50. Each particle represents a set of hyperparameter combinations. Randomly assign an initial position and initial velocity to each particle, with the initial velocity range set to -0.1 to 0.1.

[0086] The position of each particle is used as a hyperparameter of the prediction model, and the corresponding objective function value is calculated. The objective function value is the energy efficiency prediction error, and the formula for calculating the energy efficiency prediction error is as follows:

[0087] in, Indicates the average relative prediction error. Indicates the number of test samples. This represents the predicted energy efficiency value of the j-th sample. This represents the actual energy efficiency value of the j-th sample.

[0088] The individual optimal position and the global optimal position are updated based on the objective function value of each particle. The individual optimal position is the position where the objective function value is the smallest in the history of each particle, and the global optimal position is the position where the objective function value is the smallest among all particles.

[0089] Based on the particle swarm optimization formula, the velocity and position of each particle are adjusted. The multi-drive power allocation optimization formula is as follows:

[0090] In the formula, This indicates the position of the i-th particle in the (k+1)-th iteration; This represents the position of the i-th particle in the k-th iteration; and The learning factor is set to 2 in this embodiment; and Represents a random number in the range [0,1]. This represents the optimal position of the i-th particle. This indicates the globally optimal position.

[0091] The process of calculating the objective function and updating the particle position is repeated until the preset number of iterations is reached. The number of iterations is set to 50 to 100 times. The hyperparameter combination represented by the global optimal position is used as the hyperparameter combination of the prediction model.

[0092] Based on the performance parameters of the sheet metal cutting equipment, adjustment thresholds and warning thresholds for energy efficiency risk areas are set. When the predicted energy efficiency data reaches the adjustment threshold, an adaptive adjustment mechanism is introduced to dynamically adjust the cutting process parameters; when the predicted energy efficiency reaches the warning threshold, an alarm is triggered.

[0093] The process of setting adjustment thresholds and early warning thresholds for energy efficiency risk zones includes: Collect the performance parameters of the sheet metal cutting equipment. The performance parameters include the maximum allowable energy efficiency loss, the minimum energy efficiency utilization rate, and the expected normal energy efficiency range. The maximum allowable energy efficiency loss is determined according to the equipment design specifications. In this embodiment, it is 1.5 times the normal energy efficiency loss. The minimum energy efficiency utilization rate is set according to the energy efficiency standard requirements and is not less than 60%. The expected normal energy efficiency range is determined by the 95% confidence interval of historical normal operation data.

[0094] The expected normal energy efficiency range is used as the adjustment threshold. When the predicted energy efficiency data exceeds the upper limit of this range, the adjustment mechanism is activated. The maximum allowable energy efficiency loss and the minimum energy efficiency utilization rate are used as the warning threshold. When the predicted energy efficiency loss exceeds the maximum allowable energy efficiency loss or the energy efficiency utilization rate is lower than the minimum energy efficiency utilization rate, an alarm is triggered.

[0095] The process of dynamically adjusting cutting process parameters includes: optimizing the power distribution ratio of the multi-drive motors, using model predictive control methods, with minimizing total energy efficiency loss as the objective function and the power limits of each drive unit as constraints. The objective function is:

[0096] In the formula, Represents the objective function value; This represents the energy loss at time τ; This indicates the predictive control time domain.

[0097] The constraints are:

[0098]

[0099]

[0100] In the formula, This represents the minimum allowable power of the i-th drive unit; This represents the maximum allowable power of the i-th drive unit; Indicates the minimum cutting speed; Indicates the maximum cutting speed; Indicates the minimum cutting depth; This indicates the maximum cutting depth.

[0101] The optimal power allocation scheme is solved and recalculated once in each control cycle, with the control cycle set to 10 milliseconds to 100 milliseconds to reduce the ineffective losses of the transmission system.

[0102] Adjust the matching relationship between cutting speed and feed rate, and establish a response surface model of cutting speed-feed rate-energy loss. The response surface model adopts a second-order polynomial form, and the formula is:

[0103] In the formula, Indicates energy loss; Indicates the cutting speed; Indicates the feed rate; Indicates the coefficient of the constant term; Denotes the coefficient of the linear term; Represents the coefficient of the quadratic term; Indicates the coefficient of the interaction term; This represents the random error term.

[0104] The gradient descent method is used to search for the parameter combination that minimizes energy loss, thereby maximizing the energy utilization rate of material removal. The energy utilization rate of material removal is defined as the ratio of effective material removal energy to total input energy.

[0105] Adjust the working sequence of each cutting head to achieve multi-drive collaborative operation. Optimize the start-up and stop times of the cutting heads through a genetic algorithm to avoid multiple cutting heads being in a high-power state at the same time, thereby reducing overall energy consumption.

[0106] In summary, this embodiment provides an energy efficiency control method for plate cutting based on a one-control, multi-drive architecture. By constructing an energy efficiency loss model and a dynamic simulation model, it achieves a refined description of the energy distribution during the cutting process. The energy efficiency loss model, established by combining the law of conservation of energy and material cutting theory, can accurately quantify the contribution ratio of each energy component during the cutting process. The plate cutting dynamic simulation model, through numerical simulation, intuitively demonstrates the distribution characteristics of energy efficiency in spatial and temporal dimensions, providing a scientific basis for identifying energy efficiency risk areas. The energy efficiency risk area identification mechanism enables proactive prediction of potential energy efficiency problems. Through area division, energy efficiency data collection, and risk coefficient calculation, areas prone to abnormal energy efficiency losses can be accurately located, transforming energy efficiency management from passive response to proactive prevention, effectively avoiding equipment failures and energy waste caused by abnormal energy efficiency. A graph attention neural network-based energy efficiency prediction model combined with particle swarm optimization algorithm achieves high-precision energy efficiency prediction. The hierarchical control strategy not only ensures energy efficiency optimization during normal operation, but also effectively addresses energy efficiency risks under extreme conditions. This enables it to be widely applied in sheet metal cutting processes across multiple industries, including metal processing, furniture manufacturing, and building materials, providing effective technical support for enterprises' energy conservation, consumption reduction, and intelligent upgrading.

[0107] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0108] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for energy efficiency control in sheet metal cutting based on a one-control, multi-drive architecture, characterized in that, include: The structural parameters and cutting process parameters of the sheet metal cutting equipment are collected in real time, and the structural parameters and cutting process parameters are preprocessed to obtain preprocessed data; Based on the preprocessed data, combined with the law of conservation of energy and material cutting theory, an energy efficiency loss model is established to simulate the energy distribution during the cutting process, and a dynamic simulation model for plate cutting is constructed. Based on the energy distribution during the cutting process and the structural parameters, areas prone to abnormal energy loss are identified as energy efficiency risk areas. Real-time operational data of the energy efficiency risk area is collected, an energy efficiency prediction model based on graph attention neural network is constructed, and the hyperparameters of the prediction model are optimized using particle swarm optimization algorithm. The preprocessed data and real-time operational data of the energy efficiency risk area are input, and the predicted energy efficiency data of the energy efficiency risk area at the next moment is output. Based on the performance parameters of the sheet metal cutting equipment, the adjustment threshold and warning threshold for the energy efficiency risk area are set. When the predicted energy efficiency data reaches the adjustment threshold, an adaptive adjustment mechanism is introduced to dynamically adjust the cutting process parameters. When the predicted energy efficiency reaches the warning threshold, an alarm is triggered.

2. The plate cutting energy efficiency control method based on a one-control, multi-drive architecture according to claim 1, characterized in that, The structural parameters include geometric data and electromechanical characteristic parameters of the cutting equipment; the electromechanical characteristic parameters include rated power of the motor, transmission efficiency coefficient, and cutting head loss coefficient; the cutting process parameters include cutting speed, cutting depth, feed rate, real-time power data of the multi-drive motor, and operating status data of each transmission mechanism.

3. The energy efficiency control method for plate cutting based on a one-control, multi-drive architecture according to claim 1, characterized in that, The preprocessing of the structural parameters and the cutting process parameters includes: The wavelet threshold denoising algorithm is used to remove random noise from the structural parameters and the cutting process parameters, and the Z-score method is used to identify and remove outliers from the structural parameters and the cutting process parameters. The structural parameters and the cutting process parameters are smoothed using a moving average method, and data normalization is used to convert the structural parameters and the cutting process parameters to a uniform range. The missing values ​​in the structural parameters and the cutting process parameters are processed by interpolation to obtain a complete dataset, which is then used as the preprocessed data.

4. The plate cutting energy efficiency control method based on a one-control, multi-drive architecture according to claim 1, characterized in that, The process of simulating energy distribution during cutting includes: Based on the material cutting theory, a material removal energy consumption model is established during the cutting process; Based on the law of conservation of energy, the cutting process is decomposed into three components: cutting deformation energy, friction energy, and thermal energy, and an energy efficiency loss calculation model is established. Set boundary conditions and initial conditions, wherein the boundary conditions include power allocation at the cutting start point and energy recovery at the cutting end point; The energy loss model is solved using the finite volume method to obtain the energy distribution during the cutting process.

5. The plate cutting energy efficiency control method based on a one-control, multi-drive architecture according to claim 1, characterized in that, The process of constructing a dynamic simulation model for sheet metal cutting includes: The framework of the three-dimensional simulation basic model is set based on the preprocessed data; Combining the law of conservation of energy and the theory of multi-drive synergy, a set of equations describing the change of cutting energy efficiency over time is established, which includes power distribution equations and energy efficiency evolution equations. The energy distribution during the cutting process is used as the initial standard of the three-dimensional simulation basic model, and the power distribution of each driving unit is used as the boundary standard. Set the time step and perform dynamic simulation on the three-dimensional simulation base model; The equations are solved using the finite volume method to generate time-series data on cutting energy efficiency, thus obtaining a dynamic simulation model for plate cutting.

6. The energy efficiency control method for plate cutting based on a one-control, multi-drive architecture according to claim 1, characterized in that, The process of identifying the energy efficiency risk areas includes: Divide the dynamic simulation model of sheet metal cutting into regions; Real-time collection of energy efficiency data from each region of the dynamic simulation model of plate cutting, generating energy efficiency time-series curves for each region and spatial distribution map of the entire cutting surface; Based on the multi-drive cooperative theory and the aforementioned spatial distribution map, power allocation status data for each region is obtained; An energy efficiency risk coefficient calculation model is established, and the energy efficiency risk coefficient of each region is obtained based on the average energy efficiency loss, peak energy efficiency loss, power distribution status data and structural parameters of each region. Areas where the energy efficiency risk coefficient reaches a preset threshold are designated as energy efficiency risk areas.

7. The energy efficiency control method for plate cutting based on a one-control, multi-drive architecture according to claim 1, characterized in that, The process of constructing an energy efficiency prediction model based on a graph attention neural network includes: Collect historical structural parameters, historical cutting process parameters, and historical energy efficiency data of energy efficiency risk areas of the plate cutting equipment, as well as the energy efficiency data of the energy efficiency risk areas at the next moment corresponding to the historical energy efficiency data; The collected historical structural parameters, historical cutting process parameters, historical energy efficiency data of energy efficiency risk areas, and energy efficiency data of energy efficiency risk areas at the next moment are preprocessed and divided into training set and test set; A basic graph attention neural network model is constructed, comprising an input layer, a graph attention layer, a pooling layer, a fully connected layer, and an output layer. The input layer receives preprocessed input data, including historical structural parameters, historical cutting process parameters, and historical energy efficiency data for energy efficiency risk areas. The graph attention layer includes multiple attention heads for spatial feature extraction and multi-drive correlation modeling of the input data, generating a feature map. The pooling layer reduces the spatial dimensionality of the feature map. The fully connected layer integrates the extracted features. The output layer outputs a prediction result based on the integration result. The graph attention neural network base model is trained using the training set, and the model parameters that meet the prediction accuracy are retained to obtain the energy efficiency prediction model.

8. The energy efficiency control method for plate cutting based on a one-control, multi-drive architecture according to claim 1, characterized in that, The process of optimizing the hyperparameters of the prediction model using the particle swarm optimization algorithm includes: Define a hyperparameter space, with the learning rate, number of graph attention heads, hidden layer dimension, and regularization parameter as the hyperparameters to be optimized, and set a range of values ​​for each hyperparameter; Initialize the particle swarm, where each particle represents a set of hyperparameter combinations, and randomly assign an initial position and initial velocity to each particle. The position of each particle is used as a hyperparameter of the prediction model, and the corresponding objective function value is calculated. The objective function value is the energy efficiency prediction error. Update the individual optimal position and the global optimal position based on the objective function value of each particle; Adjust the velocity and position of each particle according to the particle swarm update formula; The process of calculating the objective function and updating the particle position is repeated until a preset number of iterations is reached. The hyperparameter combination represented by the globally optimal position is then used as the hyperparameter combination of the prediction model.

9. The energy efficiency control method for plate cutting based on a one-control, multi-drive architecture according to claim 1, characterized in that, The process of setting the adjustment threshold and warning threshold for the energy efficiency risk zone includes: Collect performance parameters of the sheet metal cutting equipment, including maximum allowable energy loss, minimum energy efficiency rate, and expected normal energy efficiency range; The expected normal energy efficiency range is used as the adjustment threshold, and the maximum allowable energy efficiency loss and the minimum energy efficiency utilization rate are used as the warning threshold.

10. The energy efficiency control method for plate cutting based on a one-control, multi-drive architecture according to claim 1, characterized in that, The process of dynamically adjusting the cutting process parameters includes: Optimize the power distribution ratio of multi-drive motors to reduce ineffective losses in the transmission system; Adjust the matching relationship between cutting speed and feed rate to improve the energy utilization rate of material removal; Adjust the working sequence of each cutting head to achieve multi-drive collaborative operation and reduce overall energy consumption.

Images