Industrial robot flexible machining process for new energy automobile parts

Abstract

The invention discloses an industrial robot flexible machining technology for new energy automobile parts, and relates to the technical field of industrial robot flexible machining, and the technology comprises the steps: constructing an energy consumption-electromagnetic predictor through a historical load curve and virtual simulation, and predicting a power peak time domain sequence; a servo switch strategy is switched in real time according to a prediction result, and a distributed noise suppression unit is started to suppress electromagnetic interference; multi-mode signals are synchronously collected in the noise suppression window, a thermal noise coupling coefficient is generated through thermal-signal quantization, and credible data are screened; the digital twinborn compensation model receives the confirmation data and outputs a cutter spacing trace correction amount and a pulse cooling instruction, and linkage adjustment of a heat source, power and cooling is achieved; and after machining is completed, residual size errors and energy consumption indexes are summarized, the weight of the self-adaptive update predictor is fed back, and the prediction precision of the next workpiece is improved. Through the synergistic effect of prediction, noise suppression, quantization, compensation and self-adaption, the machining precision and energy efficiency are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial robot flexible processing, and in particular to a flexible processing technology of industrial robots for new energy vehicle parts. Background Art

[0002] In flexible machining cells for new energy vehicle parts, industrial robots are typically equipped with high-speed electric spindles, performing complex operations such as milling and drilling on battery trays and high-voltage busbars. The continuous heating of the spindle and workpiece during high-speed cutting is believed to be a key factor in dimensional accuracy drift. To monitor machining status in real time, the system employs multimodal sensors, including vision, force, torque, and vibration. These highly sensitive electronic devices must operate alongside high-power drive components within a confined space. When toolpaths enter peak load periods, transient power spikes generated by the electric spindle and servo drive increase overall energy consumption and induce broadband electromagnetic noise. Electromagnetic interference is commonly observed in manufacturing sites, disrupting signal acquisition and causing equipment malfunctions. Complex metal structures further exacerbate the coupling between wireless and wired channels. To mitigate thermal deformation and signal drift, researchers are beginning to introduce digital twin models. These virtual environments optimize sensor layout and implement thermal error compensation, synchronizing the virtual and physical machining chains.

[0003] However, due to the temporal and spatial overlap of high-speed cutting and multi-sensor collaboration, heat sources, power spikes, and electromagnetic fields intertwine, forming a complex coupling phenomenon that amplifies each other. The gradual heating of the spindle and workpiece causes minute displacements of the geometric datum, which in turn decouples the real-time measurements that tool position compensation relies on. Subsequently, radiated and conducted interference from the drive system and high-frequency cables enter the sensor chain, causing jitter in the force signal, visual coordinates, and vibration spectrum, affecting the controller's error assessment. When cooling flow and energy consumption scheduling fail to keep pace with load changes, thermal distortion and signal drift jointly compress the trustworthy boundaries of the digital twin model, reducing the effectiveness of predictive compensation. As a result, hole positions and slots gradually deviate from design requirements, busbar assembly becomes misaligned, rework rates rise, and the cycle time and quality of the entire flexible unit are repeatedly strained, ultimately undermining the high-precision and high-flexibility machining advantages that the present invention aims to achieve.

[0004] To this end, the present invention provides a flexible processing technology for industrial robots of new energy vehicle parts. Summary of the Invention

[0005] (1) Technical problems solved In response to the shortcomings of the existing technology, the present invention provides a flexible processing technology for industrial robots of new energy vehicle parts. By utilizing historical load curves and virtual simulation to construct an energy consumption-electromagnetic predictor, the power peak time domain sequence is predicted; the servo switch strategy is instantly switched according to the prediction results and the distributed noise suppression unit is started to suppress electromagnetic interference; multimodal signals are synchronously collected within the noise suppression window, and the thermal noise coupling coefficient is generated through heat-signal quantization to screen reliable data; the digital twin compensation model receives confirmation data, outputs the tool position micro-correction amount and pulse cooling instruction, and realizes the linkage adjustment of heat source, power and cooling; after processing is completed, the residual dimensional error and energy consumption index are summarized, and the predictor weight is adaptively updated based on feedback to improve the prediction accuracy of the next workpiece; the technical problems recorded in the background technology are solved.

[0006] (2) Technical solution To achieve the above objectives, the present invention is implemented through the following technical solutions: a flexible processing technology for industrial robots of new energy vehicle parts, comprising: constructing an energy consumption-electromagnetic predictor using historical load curves and virtual simulation results to output a time-domain sequence of power peaks in the next tool path section; Instantly switch the servo switch strategy and activate the distributed noise suppression unit according to the power peak time domain sequence to perform shielding and harmonic suppression; Multimodal signals are collected synchronously within the noise suppression window, and the thermal-noise coupling coefficient is generated by calculating the displacement potential of the coupling between cutting contact heat energy and structural stiffness and the short-time Fourier entropy of the signal to determine the data availability. The digital twin compensation model receives data confirmed by the thermal noise coupling coefficient, outputs a slight correction value for the tool position and a pulse cooling instruction, and injects it into the next tool path according to a unified timestamp, realizing the coordinated adjustment of heat source, power and cooling; After machining is completed, the residual dimensional error and energy consumption index are summarized and fed back to the energy consumption-electromagnetic predictor to adaptively update the weights to improve the prediction accuracy of the next workpiece.

[0007] Furthermore, the power consumption data of the electric spindle, the power consumption data of the servo drive, and the power data generated by the virtual simulation are collected and preprocessed. The preprocessing steps include removing outliers, filling missing values ​​through linear interpolation, and performing normalization. The power peak, peak duration and peak interval features are extracted from the historical load curve data, and the simulated power peak, simulated peak duration and simulated peak interval features are extracted from the virtual simulation data.

[0008] Furthermore, a long short-term memory network is used to construct a prediction model. The input features include the power peak, peak duration, and peak interval of the historical load curve data, the simulated power peak, simulated peak duration, and simulated peak interval of the virtual simulation data, and the current tool path parameters. The output is the power peak time domain sequence of the next tool path segment.

[0009] Furthermore, the model was trained using the training set, the mean square error between the predicted power sequence and the actual sequence was used as the loss function, the gradient descent method was used to optimize the parameters, and the model performance was evaluated through five-fold cross validation; The current machining conditions and preprocessed features are input into the trained model to generate a time domain sequence of power spikes in the next tool path section, and this sequence is used as the time reference.

[0010] Furthermore, the controller receives and parses the power spike time domain sequence, extracts the start time, peak amplitude and duration of each power spike, and decides whether to switch the servo drive to low power mode based on whether the peak amplitude exceeds a preset threshold within a preset time window before the power spike arrives.

[0011] Furthermore, the low power mode is achieved by reducing the feed speed and adjusting the electric spindle speed and returning to normal mode after the power spike ends, while the distributed noise suppression unit is activated within the power spike time window; The electromagnetic shielding device is activated and the active filter is enabled to attenuate electromagnetic noise and offset harmonic currents. The signal-to-noise ratio of the sensor signal is monitored within the noise suppression window. When the signal-to-noise ratio is lower than a preset threshold, the active filter gain is adjusted until the signal-to-noise ratio reaches the preset threshold. Finally, the power peak time window and the corresponding noise suppression status are passed to the subsequent steps.

[0012] Furthermore, multimodal signals from visual sensors, force and torque sensors, and vibration sensors are collected synchronously. Cutting contact heat energy is calculated based on the contact area between the tool and the workpiece, cutting speed, cutting force, and heat conversion coefficient, and displacement potential is calculated in combination with structural stiffness. Short-time Fourier transform is performed on the force signal, torque signal and vibration signal, and short-time Fourier entropy is calculated to measure the signal purity; the displacement potential and short-time Fourier entropy are input into the fuzzy decision tree model to generate the thermal-noise coupling coefficient and the data is filtered according to the thermal-noise coupling coefficient.

[0013] Furthermore, the digital twin compensation model receives available data confirmed by the thermal noise coupling coefficient and extracts real-time status information during the machining process, including tool position and workpiece temperature; Use multi-physics field coupling simulation to predict machining errors, calculate tool position corrections to offset the predicted errors, and adjust tool positions in real time through the industrial robot control system; The cooling requirements are calculated based on the workpiece temperature and cutting contact heat energy, and pulse cooling instructions are generated to control the coolant flow rate. The tool position micro-correction amount and pulse cooling instructions are injected into the next tool path planning according to a unified timestamp to achieve coordinated adjustment of heat source, power and cooling.

[0014] Furthermore, the actual dimensions of the workpiece are measured using high-precision measuring equipment and compared with the designed dimensions to calculate the residual dimensional error. The energy consumption of the electric spindle, servo drive, and cooling system during machining is also collected to calculate the total energy consumption. The power peak time domain sequence output by the energy consumption-electromagnetic predictor is compared with the actual collected power peak time domain sequence to calculate the power prediction error. The processing error predicted by the digital twin compensation model is compared with the actual measured residual size error to calculate the processing error prediction deviation.

[0015] Furthermore, a neural network-based machine learning model is used to adaptively adjust the weight parameters by optimizing the objective function, where the objective function is composed of a weighted combination of the square of the integral of the power prediction error during the processing time period, the square of the modulus of the processing error prediction deviation, and the total energy consumption; The updated weight parameters are applied to the energy consumption-electromagnetic predictor to generate the time domain sequence of power spikes for the next workpiece. Based on this, the servo switching strategy and the control parameters of the distributed noise suppression unit, as well as the tool position fine-tuning and pulse cooling instructions of the digital twin compensation model are optimized.

[0016] (3) Beneficial effects The present invention provides a flexible processing technology for industrial robots of new energy vehicle parts, which has the following beneficial effects: An energy consumption-electromagnetic predictor is constructed using historical load curves and virtual simulation results to output the time domain sequence of power spikes in the next tool path section, providing an accurate time benchmark for subsequent noise suppression scheduling. This enables the process to predict the occurrence of power spikes before processing and provides forward-looking guidance for noise suppression measures.

[0017] The controller instantly switches the servo switching strategy and activates the distributed noise suppression unit based on the power spike time domain sequence, achieving targeted shielding and harmonic suppression at the source. Compared with traditional noise suppression methods, this solution can accurately intervene in high-risk sections, avoid redundant filtering, significantly improve the noise suppression effect, reduce unnecessary energy consumption, and enhance the stability and reliability of the processing process.

[0018] By calculating the displacement potential of the coupling of cutting contact heat energy and structural stiffness, as well as the short-time Fourier entropy difference of the signal, a thermal-noise coupling coefficient is generated to determine the data availability. Compared with traditional single signal processing, the heat-signal dual quantization can significantly improve the accuracy of data evaluation, combining thermal effects with signal purity to achieve all-round protection of data quality.

[0019] By adjusting the tool position in real time and dynamically controlling the coolant flow, it is possible to effectively compensate for machining errors, control the workpiece temperature, significantly improve machining accuracy and quality, ensure the coordination and real-time nature of process parameters, overcome the adverse effects of thermal distortion and signal drift on machining accuracy in traditional methods, achieve coordinated control of heat source, power and cooling, and give full play to the advantages of digital twin technology in flexible machining.

[0020] By summarizing residual dimensional errors and energy consumption indicators, these are fed back to the energy consumption-electromagnetic predictor to adaptively update weights, thereby improving the prediction accuracy of the next workpiece. This closed-loop feedback mechanism not only optimizes the predictor's performance but also achieves a gradual improvement in machining accuracy and energy efficiency by adjusting weight parameters, ensuring continuous improvement in the cycle time and quality of the flexible machining unit. Compared to traditional static processes, dynamic optimization significantly enhances the process's adaptability and long-term stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 The figure is a schematic diagram of the flexible processing process of the industrial robot for automobile parts according to the present invention. DETAILED DESCRIPTION

[0022] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0023] See also Figure 1 The present invention provides a flexible processing technology for industrial robots of new energy vehicle parts, comprising: Step 1: Collect and preprocess the power consumption data of the electric spindle, the servo drive, and the power data generated by the virtual simulation. The preprocessing steps include removing outliers, filling missing values ​​through linear interpolation, and performing normalization. Extract power peak, peak duration and peak interval features from historical load curve data, and extract simulated power peak, simulated peak duration and simulated peak interval features from virtual simulation data; A prediction model is constructed using a long short-term memory network. The input features include the power peak, peak duration, and peak interval of historical load curve data, the simulated power peak, simulated peak duration, and simulated peak interval of virtual simulation data, and the current tool path parameters. The output is a time-domain sequence of power peaks for the next tool path segment. The model is trained using the training set. The mean square error between the predicted power sequence and the actual sequence is used as the loss function. The gradient descent method is used to optimize the parameters, and the model performance is evaluated through five-fold cross-validation. The current processing conditions and preprocessed features are input into the trained model to generate the power peak time domain sequence of the next tool path segment, and this sequence is passed to the subsequent steps as the time reference.

[0024] The step 1 includes the following: Step 101: Data collection and preprocessing The power consumption data of the electric spindle and the servo drive are collected. The unit of these data is watt, and the time resolution is set to the millisecond level to ensure that the details of the power changes are captured. At the same time, the tool path parameters, including cutting speed, feed rate and tool path, are input through the virtual simulation software to simulate the processing process of the next tool path section and generate the corresponding simulated power data. In the preprocessing stage, the historical load curve data and the simulated power data are firstly eliminated for outliers. The specific method is to identify and remove data points whose power values ​​exceed the physical reasonable range, such as the part where the power mutation exceeds the rated value of the equipment; then, for the missing values ​​in the data, linear interpolation is performed by calculating the average value of the two adjacent data points; finally, all the data are normalized to the range of 0 to 1 through linear transformation of the maximum and minimum values. The purpose of normalization is to eliminate the influence of the power data due to unit or magnitude differences and ensure the consistency of subsequent processing.

[0025] Historical load curve data reflects the energy consumption patterns in actual machining, while simulation data provides the basis for predicting the next toolpath segment. Preprocessing ensures the integrity and consistency of the data, which can improve the accuracy of subsequent feature extraction and model prediction, effectively reduce noise interference, and provide high-quality input data for subsequent analysis.

[0026] Step 102: Feature extraction After completing data preprocessing, key features are extracted from the historical load curve data, including the power peak, peak duration, and peak interval of each tool path; The power peak refers to the maximum power value recorded during a toolpath machining cycle, measured in watts. The peak duration refers to the duration of time, measured in seconds, that the power value exceeds twice the average power value for that toolpath. The peak interval refers to the time difference between two adjacent power peaks, measured in seconds. Similarly, the corresponding simulated power peak, simulated peak duration, and simulated peak interval are extracted from the power data generated by virtual simulation, using the same extraction method as historical data. During feature extraction, the time window is set to the complete machining cycle of each toolpath, that is, the entire time period from the tool entering the workpiece to leaving the workpiece, to ensure that the features can fully reflect the power variation characteristics during the machining process. The power peak, peak duration, and peak interval can characterize the key timing behaviors of energy consumption and electromagnetic interference during machining, directly affecting the accuracy of subsequent predictions.

[0027] Step 103: Model construction Based on feature extraction, an energy consumption-electromagnetic predictor is constructed, using long short-term memory network as the core algorithm; The Long Short-Term Memory (LSTM) network is an artificial neural network suitable for processing time series data and can capture the time-dependent nature of power data. The model's input features include power peaks, peak durations, and peak intervals extracted from historical load curve data, as well as simulated power peaks, peak durations, and peak intervals extracted from virtual simulation data. It also includes current toolpath parameters, such as cutting depth and feed rate, which are input in numerical form. The output of the model is the time-domain sequence of power spikes in the next toolpath segment. Specifically, it refers to the sequence of power values ​​predicted at discrete time points, including the occurrence time and amplitude of each power peak. The time series characteristics of power data are complex, making it difficult for traditional static models to accurately predict them. However, the long-short-term memory network can effectively fit the dynamic changes of power spikes by memorizing historical information and current inputs. It can then combine historical data and simulation data to generate high-precision prediction results, providing a reliable basis for subsequent process optimization.

[0028] Step 104: Model training and optimization Model training uses historical load curve data and the corresponding actual power sequence of the next tool path section as the training set; The input features are paired with the actual power sequence and learned using a long short-term memory network. The training goal is to make the predicted power sequence as close as possible to the actual sequence. To this end, the loss function is defined as the mean squared difference between the predicted and actual power sequences. This is calculated by summing the squares of the differences between the predicted and actual values ​​at each time point and taking the average.

[0029] By repeatedly adjusting model parameters and using gradient descent to gradually reduce the loss function, the team simultaneously optimized the learning rate and the number of hidden layer nodes. The learning rate determines the parameter adjustment step size, while the number of hidden layer nodes influences the model's complexity, until the loss function value stabilizes. To evaluate the model's performance, a five-fold cross-validation method was employed. This method involves dividing the training set into five equal parts, using four of the five parts to train the model and the remaining part to verify the prediction error, ultimately requiring the prediction error to be within 5%. Through extensive data training and parameter adjustments, the model is able to learn the underlying patterns of power variations and demonstrate generalization capabilities. The trained model achieves high prediction accuracy and is adaptable to power spike prediction requirements under various processing conditions.

[0030] Step 105: Output of power peak time domain sequence After model training is complete, the current machining conditions and pre-processed features are fed into the LSTM network to predict the next toolpath segment. The current machining conditions include toolpath parameters (such as cutting speed and feed rate) and workpiece material properties (such as hardness). The pre-processed features include power peaks, peak durations, and peak intervals from historical and simulated data. The model calculates the time-domain sequence of output power spikes, specifically power values ​​at a series of discrete time points, including the predicted amplitude and occurrence time of each power peak. The output power spike time-domain sequence is represented with time as the horizontal axis and power value as the vertical axis, which can intuitively reflect the power change trend of the next tool path section. The power peak of the next tool path section directly affects the timing and intensity of electromagnetic interference. Accurately predicting its time-domain distribution is the key to subsequent noise suppression. This sequence provides detailed power change information, facilitating targeted intervention in subsequent process steps.

[0031] Step 106: Time base provision The power spike time domain sequence is used as a time reference and passed to the noise suppression scheduling link in step 2. The peak occurrence time and predicted peak amplitude in the power spike time domain sequence clearly identify the occurrence time and intensity of the power spike, which is specifically manifested in the time point and power value corresponding to each peak. This information provides an accurate timing reference for the noise suppression unit in step 2, so that noise suppression measures can be accurately intervened during high-risk periods when power spikes occur, such as by adjusting processing parameters or enabling noise suppression equipment to reduce electromagnetic interference. The timing characteristics of power spikes are highly correlated with the occurrence of electromagnetic interference, and understanding its distribution in advance can optimize the noise suppression strategy. Based on this time reference, step 2 can achieve efficient interference suppression and improve the stability and accuracy of the processing process.

[0032] When used, a complete power spike prediction process is formed, from data collection and preprocessing to providing a time base. The combination of historical load curve data and virtual simulation results ensures the comprehensiveness and foresight of the prediction. The application of long-short-term memory networks enables accurate modeling of time series data. The output of power spike time domain sequences provides clear guidance for subsequent process steps. The impact of electromagnetic interference on machining accuracy in high-speed cutting mainly comes from power spikes. By predicting its time domain distribution, interference issues can be effectively addressed, significantly improving the stability of flexible machining units, reducing machining errors, and providing technical support for high-precision manufacturing.

[0033] Step 2: The controller receives and parses the power spike time domain sequence, extracts the start time, peak amplitude and duration of each power spike, and decides whether to switch the servo drive to low power mode according to whether the peak amplitude exceeds a preset threshold within a preset time window before the power spike arrives. The low power mode is achieved by reducing the feed speed and adjusting the electric spindle speed, and the normal mode is restored after the power spike ends. At the same time, the distributed noise suppression unit is started within the power spike time window, the electromagnetic shielding device is activated, and the active filter is enabled to attenuate electromagnetic noise and offset harmonic currents. The signal-to-noise ratio of the sensor signal is monitored within the noise suppression window. When the signal-to-noise ratio is lower than the preset threshold, the active filter gain is adjusted until the signal-to-noise ratio reaches the preset threshold. Finally, the power spike time window and the corresponding noise suppression state are passed to the subsequent steps. The second step includes the following: Step 201: Receiving and analyzing power peak time domain sequence Step 2 receives the power spike time domain sequence from step 1. This sequence contains a series of power values ​​at discrete time points, specifically including the occurrence time of each power spike and the predicted power peak amplitude; the occurrence time is recorded in seconds, and the power peak amplitude is expressed in watts.

[0034] When parsing this sequence, the start time, peak amplitude, and duration of each power spike are extracted. The duration is determined by calculating the time difference between the power values ​​at adjacent time points dropping to the average level. Specifically, this is the time interval from the peak time point until the power value returns to the average power value during system operation. The timing information of the power spike directly determines the activation timing of the servo drive and noise suppression unit. Accurately extracting this information ensures the accuracy of subsequent treatment measures, providing a precise timing basis for the servo switching strategy and the activation of the noise suppression unit, thereby improving the timeliness and relevance of intervention measures.

[0035] The controller adjusts the working mode of the servo drive within a preset time window before the arrival of each power spike according to the occurrence time and peak amplitude in the power spike time domain sequence.

[0036] The preset time window is set to 0.05 seconds, determined based on the machining unit's response speed, to ensure the servo drive has sufficient time to complete the mode switch. When the predicted power peak amplitude exceeds a preset threshold, the controller switches the servo drive to low-power mode. The preset threshold is set at 80% of the motorized spindle's rated power. For example, if the motorized spindle's rated power is 2000 watts, the threshold is 1600 watts. In low-power mode, the servo drive's feed rate is reduced to 70% of its normal value to reduce transient current surges. The motorized spindle's speed is also reduced by 10% in revolutions per minute to smooth the power curve. After the power spike expires, the controller returns the servo drive to normal mode to maintain machining efficiency. Mode switching is achieved using pulse-width modulated signals, with the controller issuing commands with millisecond-level resolution. By adjusting the servo drive's operating mode, the amplitude of the power spike can be effectively reduced, thereby minimizing electromagnetic interference (EMI). This reduces energy consumption and electromagnetic noise while maintaining machining stability.

[0037] Step 202: Distributed noise suppression unit starts During the time window of each power spike, a distributed noise suppression unit is activated, which includes an electromagnetic shielding device and an active filter. The time window is defined as starting from a preset time window before the power spike occurs and ending at the end of the power spike duration. The electromagnetic shielding device uses a local shielding cover made of high-permeability material, such as Permalloy, which covers the periphery of the electric spindle and servo drive to attenuate broadband electromagnetic noise. The switch of the shielding cover is controlled by a relay with a response time of less than 10 milliseconds. The active filter detects the harmonic current in the system in real time and injects a reverse compensation current to offset the harmonics; the harmonic current is obtained through high-speed sampling with a sampling rate of 10 kHz, and the compensation current is generated by the inverter to make the total current of the system close to a sine wave, thereby reducing electromagnetic interference. The noise suppression unit is turned off after the power spike duration ends to reduce unnecessary energy consumption; electromagnetic shielding and active filtering can suppress the propagation and generation of electromagnetic noise from the source, protect the sensor from interference, improve the purity of the sensor signal, and ensure the reliability of subsequent data acquisition; During each power spike time window, the sensor signal's signal-to-noise ratio (SNR) is monitored to assess noise suppression effectiveness. The SNR is calculated as the logarithm of the ratio of effective signal power to noise power (10 times the base-10 logarithm of the ratio, expressed in decibels). If the SNR falls below a preset threshold of 20 decibels, the active filter gain is adjusted in increments of 0.5 from 1 to 5 until the required SNR is achieved. The verification results, including the signal-to-noise ratio sequence within the time window, are passed to step three as a reference for multimodal signal acquisition. Real-time monitoring of the signal-to-noise ratio allows for timely detection of insufficient noise suppression and optimization through active filter gain adjustment. This ensures sensor signal quality and provides high-quality data support for subsequent thermal-noise coupling coefficient calculations.

[0038] The power spike time window and corresponding noise suppression status, including servo mode switching records and signal-to-noise ratio sequences, are passed to step three. This data provides the noise suppression window for the simultaneous acquisition of multimodal signals in step three, ensuring signal purity. The noise suppression status record provides a reference for subsequent data analysis and processing, helping to accurately assess signal usability. The noise suppression window and status information are crucial for data acquisition and analysis in step three, ensuring data validity and reliability. This enhances the overall coordination of the process and improves machining accuracy and quality.

[0039] During use, step two achieves precise intervention on power spikes and effective suppression of electromagnetic interference. The instant switching of the servo switching strategy and the activation of the distributed noise suppression unit work together to reduce the impact of power spikes and electromagnetic noise. Verification of the noise suppression effect and its connection to subsequent steps ensure high sensor signal quality and process continuity. By combining prediction and intervention to address the impact of electromagnetic interference on machining accuracy during high-speed cutting, the generation and propagation of interference are controlled at the source. This significantly improves the stability and machining quality of the flexible machining unit, providing technical support for high-precision manufacturing.

[0040] Step 3: Synchronously collect multimodal signals from the visual sensor, force and torque sensor, and vibration sensor. The sampling frequency is set to 1kHz, and the timestamp alignment error is controlled within 1 millisecond. The cutting contact heat energy is calculated based on the contact area between the tool and the workpiece, the cutting speed, the cutting force, and the heat transfer coefficient. The displacement potential is calculated in combination with the structural stiffness. A short-time Fourier transform (STFT) is performed on the force, torque, and vibration signals with a window width of 0.1 seconds and a 50% overlap. The STFT entropy is calculated to measure signal purity. The displacement potential and STFT are input into a fuzzy decision tree model to generate a thermal-noise coupling coefficient, which ranges from 0 to 1. Data is filtered based on the STFT. Data with a STFT greater than or equal to 0.8 is considered usable and passed to subsequent steps. The step three includes the following: Step 301: Synchronous acquisition of multimodal signals within the noise suppression window Within the noise suppression window determined in step 2, synchronously collect signals from the visual sensor, force and torque sensor, and vibration sensor. The visual sensor is responsible for collecting image data, the force and torque sensor is responsible for collecting force and torque signals, and the vibration sensor is responsible for collecting acceleration signals. All signals use time as the independent variable, with a uniform sampling frequency of 1000 times per second. The timestamps of each sensor are aligned, with an error of less than 1 millisecond, to ensure data synchronization. The rationale for synchronous multimodal signal acquisition within the noise suppression window is that electromagnetic interference is suppressed, resulting in high-quality signals suitable for subsequent data analysis. This provides high-quality raw data for thermal-signal quantification calculations, thereby improving the reliability of the analysis results.

[0041] The cutting contact heat energy is calculated based on the contact area between the tool and the workpiece, the cutting speed and the cutting force. The contact area is measured in square meters, the cutting speed is measured in meters per second, and the cutting force is measured in Newtons. The cutting contact heat energy is obtained by multiplying the contact area, cutting speed, cutting force and the heat conversion coefficient, where the heat conversion coefficient is a dimensionless parameter that represents the ratio of cutting energy to heat energy. Combined with the structural stiffness (in Newtons per meter), the displacement potential is calculated. The displacement potential represents the structural displacement trend caused by thermal energy and is measured in meters. The specific calculation method is to divide the cutting contact heat energy by the structural stiffness. The displacement caused by thermal energy will affect the processing accuracy. By quantifying the displacement potential, the degree of influence of thermal deformation can be evaluated. This provides a key indicator reflecting the thermal effect for the generation of the thermal noise coupling coefficient.

[0042] Step 302: Heat-Signal Quantification: Signal Purity Measurement The force, torque, and vibration signals were subjected to a short-time Fourier transform (SFT) with a transform window width of 0.1 seconds and a window overlap ratio of 50% to obtain the time-frequency spectrum of each signal. The time-frequency spectrum reflects the energy distribution of the signal at different times and frequencies. The short-time Fourier entropy of each signal was calculated. The entropy value was obtained by taking the logarithmically weighted sum of the normalized power spectral density, where the normalized power spectral density is the square modulus of the time-frequency spectrum divided by the total energy. The smaller the entropy value, the more concentrated the signal is at a specific frequency, and the higher the signal purity; the larger the entropy value, the more dispersed the signal frequency distribution, and the more severely interfered with the signal. Electromagnetic interference can cause signal spectrum distortion, and the entropy value can effectively quantify the impact of interference on signal quality. This provides a quantitative basis for the generation of the thermal noise coupling coefficient.

[0043] The displacement potential and signal entropy are input into the fuzzy decision tree model. The fuzzy decision tree generates a thermal noise coupling coefficient based on predefined fuzzy rules and membership functions. The coefficient value range is set to 0 to 1, where 1 indicates that the data is completely available and 0 indicates that the data is completely unavailable.

[0044] Examples of fuzzy rules include: when the displacement potential is small and the signal entropy is low, the thermal-noise coupling coefficient is high; when the displacement potential is large or the signal entropy is high, the thermal-noise coupling coefficient is low. Through multi-layer fuzzy judgment, the thermal-noise coupling coefficient is output for each time point. The thermal-noise coupling coefficient can be used to comprehensively consider the combined effects of thermal deformation and electromagnetic interference on data quality. Fuzzy decision trees can effectively handle uncertainty and nonlinear relationships, providing an accurate comprehensive standard for assessing data availability.

[0045] Multimodal signals are screened based on the thermal-noise coupling coefficient: if the thermal-noise coupling coefficient is greater than or equal to 0.8, the data is determined to be usable data; if the thermal-noise coupling coefficient is less than 0.8, the data is determined to be unusable data and needs to be eliminated or corrected in subsequent steps.

[0046] The timestamps and corresponding signal values ​​of the available data are passed to step four as the input of the digital twin compensation model. Only by ensuring the high quality of the data input to the digital twin compensation model can the decline in compensation effect due to data errors be avoided; the accuracy and reliability of the digital twin compensation model can be improved.

[0047] During use, the system simultaneously acquires multimodal signals, calculates displacement potential and signal entropy, and generates a thermal-noise coupling coefficient, enabling accurate assessment of data availability. Displacement potential reflects the impact of thermal deformation on machining accuracy, while signal entropy quantifies the damage to signal quality caused by electromagnetic interference. The fuzzy decision tree integrates the coupling effects of the two, providing high-quality input data for the digital twin compensation model. Thermal deformation and electromagnetic interference are the main factors affecting machining accuracy. Quantifying their coupling effects can effectively improve data reliability and machining accuracy, enhance the stability of the flexible machining unit, improve machining quality, and provide support for high-precision manufacturing.

[0048] Step 4: The digital twin compensation model receives available data confirmed by the thermal noise coupling coefficient and extracts real-time status information during the machining process, including tool position and workpiece temperature. It uses multi-physics field coupling simulation to predict machining errors, calculates micro-corrections for tool position to offset the predicted errors, and adjusts the tool position in real time through the industrial robot control system. Cooling requirements are calculated based on workpiece temperature and cutting contact heat energy, generating pulse cooling instructions to control coolant flow. Tool position micro-corrections and pulse cooling instructions are injected into the next toolpath planning segment at a unified timestamp, enabling coordinated regulation of heat source, power, and cooling. Real-time adjustment data is recorded and passed on to subsequent steps to support process optimization. The step 4 includes the following contents: Step 401: Receiving and parsing available data The available data from step three, confirmed by the thermal-noise coupling coefficient, is received. This data includes timestamps and corresponding multimodal signal values, such as visual sensor images, force signals, torque signals, and vibration signals. This data is parsed to extract real-time status information during the machining process, including tool position and workpiece temperature. Tool position is represented as a three-dimensional coordinate vector in millimeters; workpiece temperature is expressed in degrees Celsius. This data serves as input parameters for the digital twin compensation model. Receiving and parsing available data ensures high-quality data input to the digital twin compensation model, preventing degradation of compensation effectiveness due to data errors. It also provides accurate real-time status information for subsequent compensation calculations, improving the reliability of machining accuracy.

[0049] The digital twin compensation model uses multi-physics field coupling simulation to integrate thermodynamics, structural mechanics, and electromagnetics to simulate the impact of thermal deformation and electromagnetic interference on machining accuracy during the machining process.

[0050] Model input parameters include real-time tool position, workpiece temperature, cutting force (in Newtons), vibration acceleration (in meters per second squared), as well as sensor layout and workpiece material properties (such as thermal conductivity and specific heat capacity) in the virtual environment. The model output is the predicted machining error, expressed as a three-dimensional error vector in microns, reflecting dimensional deviations caused by thermal deformation and electromagnetic interference. The rationale for constructing a digital twin compensation model is that the impact of multiple physical fields on machining accuracy must be comprehensively considered. Traditional single-field models cannot accurately predict errors; through multi-physical field coupling simulation, machining errors can be more accurately predicted, providing a reliable basis for subsequent compensation.

[0051] Step 402: Calculate tool position micro-correction amount Based on the machining error output by the digital twin compensation model, a micro-correction for the tool position is calculated to offset the predicted error. The micro-correction for the tool position is a displacement vector in three-dimensional space, measured in micrometers, and is calculated by inverting the predicted machining error.

[0052] The industrial robot control system uses micro-correction to adjust the tool position in real time to ensure machining accuracy. The rationale behind calculating micro-correction is that adjusting the tool position directly compensates for machining errors and improves machining accuracy. This allows for real-time dynamic adjustment of the tool path, minimizing machining errors and ultimately improving the dimensional consistency of the workpiece.

[0053] The cooling demand is calculated based on the workpiece temperature and the cutting contact heat energy (measured in joules, derived from cutting force and cutting speed). The cooling demand is the non-negative portion of the difference between the cutting contact heat energy and the workpiece temperature control threshold (measured in joules, determined by the workpiece material's heat capacity and allowable temperature rise). Pulsed cooling instructions are generated to control the coolant flow rate (in liters per second): when the cooling demand is greater than 0, the coolant flow rate is at its maximum; otherwise, it is 0. Dynamically controlling the coolant flow rate effectively controls the workpiece temperature and reduces the impact of thermal deformation on machining accuracy. This enables precise temperature control, reduces errors caused by thermal deformation, and improves machining quality.

[0054] Micro-corrections for tool position and pulse cooling instructions are injected into the next toolpath planning segment at a unified timestamp, enabling coordinated adjustment of heat source, power, and cooling. Adjusting the toolpath reduces cutting forces and contact heat energy, while pulse cooling controls workpiece temperature to maintain thermal equilibrium. The rationale for unified timestamp injection and coordinated adjustment is to ensure the coordination of tool position adjustment and cooling measures, avoiding compensation failures caused by time asynchrony. This enables real-time coordinated optimization of process parameters, improving machining accuracy and stability.

[0055] The real-time adjustment records during the machining process, including tool position micro-corrections, pulse cooling instructions, and predicted machining errors, are passed to step five as the basis for calculating residual dimensional errors and energy consumption indicators. These data support the adaptive update of the energy consumption-electromagnetic predictor in step five, improving the machining accuracy of subsequent workpieces. The real-time adjustment records provide feedback information for subsequent process optimization, achieving continuous process improvement. They can enhance the overall coordination of the process flow and improve machining quality and efficiency.

[0056] During use, the digital twin compensation model receives available data and outputs micro-corrections for tool position and pulse cooling instructions, enabling coordinated adjustment of heat source, power, and cooling. Based on multi-physics coupled simulation, the digital twin compensation model predicts machining errors and compensates them through fine-tuning of tool position. Pulse cooling instructions control workpiece temperature based on real-time cooling requirements, minimizing the effects of thermal deformation. The injection of unified timestamps ensures the coordination and real-time nature of adjustment measures, providing reliable data support for step five.

[0057] By comprehensively considering the impact of thermal deformation and electromagnetic interference on machining accuracy, and through real-time compensation and control, machining accuracy can be improved. This can significantly improve the machining quality and stability of the flexible machining unit, providing technical support for high-precision manufacturing.

[0058] Step 5. After the machining is completed, use high-precision measuring equipment to measure the actual size of the workpiece and compare it with the design size to calculate the residual dimensional error. At the same time, collect the energy consumption of the electric spindle, servo drive and cooling system during the machining process to calculate the total energy consumption; compare the power peak time domain sequence output by the energy consumption-electromagnetic predictor in step 1 with the actual power peak time domain sequence collected to calculate the power prediction error, and compare the machining error predicted by the digital twin compensation model in step 4 with the actual measured residual dimensional error to calculate the machining error prediction deviation; A neural network-based machine learning model is used to adaptively adjust weight parameters by optimizing an objective function consisting of a weighted combination of the squared integral of the power prediction error over the machining time period, the squared modulus of the machining error prediction deviation, and the total energy consumption. The updated weight parameters are applied to an energy consumption-electromagnetic predictor to generate a time-domain sequence of power spikes for the next workpiece. This sequence is then used to optimize the servo switching strategy and control parameters of the distributed noise suppression unit in step two, as well as the tool position fine-tuning and pulse cooling instructions of the digital twin compensation model in step four. The machining results from step four provide feedback data, including residual dimensional error and energy consumption data, forming a closed-loop optimization mechanism. The step five includes the following: Step 501: Summary of residual size error and energy consumption index After machining is completed, the actual dimensions of the workpiece are measured using high-precision measuring equipment and compared with the designed dimensions to calculate the residual dimensional error. The residual dimensional error refers to the deviation between the actual dimensions of the workpiece in the X, Y, and Z directions and the designed dimensions, and is expressed in microns. Energy consumption metrics are collected during the machining process, including those for the electric spindle, servo drive, and cooling system, expressed in kilowatt-hours. Total energy consumption is calculated by adding the energy consumption of the electric spindle, servo drive, and cooling system. Residual dimensional error and total energy consumption are used as feedback data for subsequent analysis and optimization. Residual dimensional error reflects the actual results of machining accuracy, while energy consumption metrics reflect the energy efficiency of the machining process. Aggregating this data provides realistic feedback for the optimization predictor. This provides accurate, real-world data for subsequent analysis, ensuring that optimization measures are targeted and reliable.

[0059] Step 502: Comparative analysis of feedback data and prediction results The power spike time-domain sequence output by the energy consumption-electromagnetic predictor in step 1 is compared with the power spike time-domain sequence actually collected during the machining process to calculate the power prediction error. The power prediction error is calculated by subtracting the predicted power value from the actual power value, in kilowatts. In addition, the machining error predicted by the digital twin compensation model in step 4 is compared with the actual measured residual dimensional error to calculate the machining error prediction deviation. The machining error prediction deviation is calculated by subtracting the predicted machining error from the residual dimensional error, in microns. The reason for conducting a comparative analysis of the feedback data and the predicted results is that by comparing the actual data with the predicted data, the accuracy of the energy consumption-electromagnetic predictor and the digital twin compensation model can be evaluated, errors in the prediction process can be identified, and specific improvement directions for optimizing the predictor can be provided.

[0060] The energy consumption-electromagnetic predictor uses a machine learning model based on a neural network, and its weight parameters are adaptively adjusted by optimizing the objective function. The objective function consists of the following parts: A weighted combination of the squared integral of the power prediction error during the processing time period, the squared modulus of the processing error prediction deviation, and the total energy consumption. The squared integral represents the cumulative effect of the power prediction error during the entire processing time period, and the squared modulus represents the magnitude of the processing error prediction deviation. The adjustment coefficients in the weighted combination are used to control the relative importance of power prediction accuracy, processing accuracy, and energy consumption optimization. These adjustment coefficients need to be pre-calibrated based on specific process requirements. The weight update method uses the gradient descent method, with the learning rate set between 0.001 and 0.01, and the gradient is calculated using numerical or analytical methods. By optimizing the objective function and adjusting the weight parameters, the prediction accuracy of the predictor and the energy efficiency of the processing process can be improved. Continuous optimization of the predictor is achieved, and the accuracy and energy efficiency of the processing process are gradually improved.

[0061] The updated weight parameters are applied to the energy consumption-electromagnetic predictor to generate a time-domain sequence of power spikes for the next workpiece. Based on this more accurate power prediction data, the servo switching strategy and control parameters of the distributed noise suppression unit in step two, as well as the tool position fine-tuning and pulse cooling instructions of the digital twin compensation model in step four, are further optimized. Improving prediction accuracy through the updated weight parameters provides more reliable process parameters for subsequent machining processes, enabling continuous improvement of the machining process, thereby enhancing machining accuracy and energy efficiency.

[0062] The feedback data for step five is derived from the machining results of step four, including the actual measured residual dimensional error and the energy consumption data recorded in real time during machining. The energy consumption-electromagnetic predictor constructed in step one provides the initial prediction value. In step five, a closed-loop optimization mechanism is formed through error comparison and weight update. By comparing and optimizing actual machining data with predicted data, a closed-loop feedback mechanism for the machining process is established to ensure continuous process improvement. This enhances the coordination of the entire process, thereby improving machining quality and efficiency.

[0063] First, residual dimensional errors and energy consumption indicators are summarized and compared with the predicted results to calculate the power prediction error and machining error prediction deviation. Based on this error data, the weight parameters of the energy consumption-electromagnetic predictor are adaptively updated by optimizing the objective function, thereby improving the prediction accuracy of the next workpiece. This process forms a closed-loop feedback mechanism, ensuring that the flexible machining cell for new energy vehicle parts gradually optimizes machining accuracy and energy efficiency while suppressing thermal deformation and electromagnetic interference. Through feedback from actual data and optimization of the predictor, continuous improvement of the machining process can be achieved, improving machining accuracy and energy efficiency while enhancing process stability and reliability.

[0064] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0065] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0066] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is only for some logical functions. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0067] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0068] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A flexible processing technology for industrial robots of new energy vehicle parts, characterized by: include, An energy consumption-electromagnetic predictor is constructed using historical load curves and virtual simulation results to output the time domain sequence of power peaks in the next tool path section. Instantly switch the servo switch strategy and activate the distributed noise suppression unit according to the power peak time sequence to perform shielding and harmonic suppression; Multimodal signals are collected synchronously within the noise suppression window, and the thermal-noise coupling coefficient is generated by calculating the displacement potential of the coupling between cutting contact heat energy and structural stiffness and the short-time Fourier entropy of the signal to determine the data availability. The digital twin compensation model receives data confirmed by the thermal noise coupling coefficient, outputs a slight correction value for the tool position and a pulse cooling instruction, and injects it into the next tool path according to a unified timestamp, realizing the coordinated adjustment of heat source, power and cooling; After machining is completed, the residual dimensional error and energy consumption index are summarized and fed back to the energy consumption-electromagnetic predictor to adaptively update the weights to improve the prediction accuracy of the next workpiece.

2. The flexible processing technology for industrial parts processing by robots according to claim 1 is characterized in that: Collect and preprocess the power consumption data of the electric spindle, servo drive, and virtual simulation. The preprocessing steps include removing outliers, filling missing values ​​through linear interpolation, and normalizing. The power peak, peak duration and peak interval features are extracted from the historical load curve data, and the simulated power peak, simulated peak duration and simulated peak interval features are extracted from the virtual simulation data.

3. The flexible processing technology for industrial parts processing by robots according to claim 2, characterized in that: A long short-term memory network is used to construct a prediction model. The input features include the power peak, peak duration, and peak interval of historical load curve data, the simulated power peak, simulated peak duration, and simulated peak interval of virtual simulation data, and the current tool path parameters. The output is the time domain sequence of power spikes in the next tool path segment.

4. The flexible processing technology for industrial parts processing by robots according to claim 3 is characterized in that: The model is trained using the training set, the mean square error between the predicted power sequence and the actual sequence is used as the loss function, the gradient descent method is used to optimize the parameters, and the model performance is evaluated through five-fold cross validation; The current machining conditions and preprocessed features are input into the trained model to generate a time domain sequence of power spikes in the next tool path section, and this sequence is used as the time reference.

5. The flexible processing technology for industrial parts processing by robots according to claim 4 is characterized in that: The controller receives and parses the power spike time domain sequence, extracts the start time, peak amplitude and duration of each power spike, and decides whether to switch the servo drive to low power mode based on whether the peak amplitude exceeds a preset threshold within a preset time window before the power spike arrives.

6. The flexible processing technology for industrial parts processing by robots according to claim 5, characterized in that: Low power mode is achieved by reducing the feed speed and adjusting the electric spindle speed, and normal mode is restored after the power peak ends. Meanwhile, the distributed noise suppression unit is activated within the power peak time window. The electromagnetic shielding device is activated and the active filter is enabled to attenuate electromagnetic noise and offset harmonic currents. The signal-to-noise ratio of the sensor signal is monitored within the noise suppression window. When the signal-to-noise ratio is lower than a preset threshold, the active filter gain is adjusted until the signal-to-noise ratio reaches the preset threshold. Finally, the power peak time window and the corresponding noise suppression status are passed to the subsequent steps.

7. The flexible processing technology for industrial parts processing by robots according to claim 6, characterized in that: Synchronously collect multimodal signals from vision sensors, force and torque sensors, and vibration sensors; calculate cutting contact heat energy based on the contact area between the tool and the workpiece, cutting speed, cutting force, and heat transfer coefficient, and calculate displacement potential in combination with structural stiffness; Short-time Fourier transform is performed on the force signal, torque signal and vibration signal, and short-time Fourier entropy is calculated to measure the signal purity; the displacement potential and short-time Fourier entropy are input into the fuzzy decision tree model to generate the thermal-noise coupling coefficient and the data is filtered according to the thermal-noise coupling coefficient.

8. The flexible processing technology for industrial parts processing by robots according to claim 7, characterized in that: The digital twin compensation model receives available data confirmed by the thermal noise coupling coefficient and extracts real-time status information during the machining process, including tool position and workpiece temperature; Use multi-physics field coupling simulation to predict machining errors, calculate tool position corrections to offset the predicted errors, and adjust tool positions in real time through the industrial robot control system; The cooling requirements are calculated based on the workpiece temperature and cutting contact heat energy, and pulse cooling instructions are generated to control the coolant flow rate. The tool position micro-correction amount and pulse cooling instructions are injected into the next tool path planning according to a unified timestamp to achieve coordinated adjustment of heat source, power and cooling.

9. The flexible processing technology for industrial parts processing by robots according to claim 8, characterized in that: Use high-precision measuring equipment to measure the actual size of the workpiece and compare it with the designed size to calculate the residual dimensional error. At the same time, collect the energy consumption of the electric spindle, servo drive and cooling system during the processing to calculate the total energy consumption; The power peak time domain sequence output by the energy consumption-electromagnetic predictor is compared with the actual collected power peak time domain sequence to calculate the power prediction error. The processing error predicted by the digital twin compensation model is compared with the actual measured residual size error to calculate the processing error prediction deviation.

10. The flexible processing technology for industrial parts processing by robots according to claim 9, characterized in that: A neural network-based machine learning model is used to adaptively adjust weight parameters by optimizing an objective function, where the objective function is a weighted combination of the square of the integral of the power prediction error during the processing time period, the square of the modulus of the processing error prediction deviation, and the total energy consumption. The updated weight parameters are applied to the energy consumption-electromagnetic predictor to generate the time domain sequence of power spikes for the next workpiece. Based on this, the servo switching strategy and the control parameters of the distributed noise suppression unit, as well as the tool position fine-tuning and pulse cooling instructions of the digital twin compensation model are optimized.

Images