Intelligent control method and system for nickel plating layer thickness of diamond fretsaw

Abstract

The invention discloses an intelligent control method and system for the thickness of a nickel plating layer of a diamond fretsaw, and relates to the technical field of diamond fretsaw electroplating control, and the method comprises the steps: collecting the current pulse, line displacement and bath solution thermal imaging multi-source sensing data flow of the operation of the diamond fretsaw, eliminating the transmission delay difference through time-space alignment, and generating a synchronous sampling data set; and identifying axial position and circumferential angle labels, and constructing a process parameter matrix with a spatial index. And inputting the matrix into a pre-training thickness prediction network to output an instantaneous deposition rate cloud picture, calculating an accumulated thickness deviation field along the length direction, generating a partition voltage modulation sequence in combination with wire walking path planning, and issuing the partition voltage modulation sequence to a rectification control unit to complete coating thickness closed-loop adjustment. According to the method, multi-source data synchronization and spatialization representation are achieved, precise closed-loop regulation and control are completed through partition voltage modulation, and the plating thickness control effect is improved.

Description

Technical Field

[0001] This invention belongs to the field of diamond wire saw electroplating control technology, specifically a method and system for intelligent control of the thickness of nickel plating on diamond wire saws. Background Technology

[0002] In the conventional process of controlling the nickel plating thickness of diamond wire saws, only current pulse signals are collected as the core monitoring parameters, and simple auxiliary feedback is provided by the line displacement signal. The bath temperature is detected by single-point contact only, without the introduction of thermal imaging signals to achieve full-area monitoring. Multiple types of sensor data are collected and transmitted independently, and fixed voltage output is achieved by conventional rectifier equipment. The plating thickness detection is mostly completed by offline sampling, which cannot meet the real-time control requirements of continuous electroplating production.

[0003] In existing control schemes, the transmission paths of current pulse, wire displacement, and bath temperature signals differ, resulting in inherent transmission delays. Various sensor data cannot be synchronized in time, and sampled data cannot accurately correspond to the actual operating position of the diamond wire saw. The system cannot identify the process parameter states corresponding to the axial position and circumferential angle of the wire saw; process parameters lack spatial indexing; the deposition state cannot be finely characterized; the instantaneous deposition rate on the wire saw surface cannot be obtained in real time; and coating thickness deviations accumulate continuously along the length of the wire saw.

[0004] The deposition rate cloud map cannot be generated through the prediction model to quantify the thickness deviation. The voltage regulation cannot be combined with the wire guide path planning to perform zoned adaptive control. The rectifier control unit can only perform overall voltage regulation and cannot perform differentiated closed-loop adjustment for thickness deviation in different areas. Therefore, it is difficult to guarantee the uniformity and control accuracy of the coating thickness. Summary of the Invention

[0005] This invention aims to solve at least one of the technical problems existing in the prior art; Therefore, this invention proposes an intelligent control method for the thickness of the nickel plating layer on a diamond wire saw, comprising: Collect multi-source sensor data streams during the operation of the diamond wire saw to be controlled. The multi-source sensor data streams include current pulse signals, wire displacement signals, and bath liquid thermal imaging signals. Perform a spatiotemporal alignment operation on the multi-source sensor data stream to eliminate transmission delay differences and generate a synchronous sampling dataset; The synchronous sampling dataset is analyzed to identify the axial position label and circumferential angle label of the diamond wire saw to be controlled, and a process parameter matrix with spatial index is constructed. The process parameter matrix is ​​input into a pre-trained thickness prediction network, which outputs an instantaneous deposition rate cloud map of the diamond wire saw to be controlled. Based on the instantaneous deposition rate cloud map, the cumulative thickness deviation field along the length direction of the diamond wire saw to be controlled is calculated; The accumulated thickness deviation field is invoked, and combined with the wire path planning information of the diamond wire saw to be controlled, to generate a partitioned voltage modulation sequence; The partition voltage modulation sequence is sent to the rectifier control unit to complete the closed-loop adjustment of the coating thickness of the diamond wire saw to be controlled.

[0006] Furthermore, the acquisition of multi-source sensor data streams during the operation of the diamond wire saw to be controlled includes: The Hall current sensor array is activated to capture the current pulse signal as the diamond wire saw to be controlled passes the anode plate; Activate the laser vibration meter to track the surface micro-vibration of the diamond wire saw to be controlled and generate a wire displacement signal; Turn on the infrared thermal imager to scan the surface temperature distribution of the plating solution in the electroplating tank and output the thermal imaging signal of the solution. The current pulse signal, line displacement signal, and tank liquid thermal imaging signal are encapsulated into a raw data packet and marked with a timestamp of the acquisition. The original data packets are aggregated by an industrial Ethernet switch to form the multi-source sensor data stream.

[0007] Furthermore, a spatiotemporal alignment operation is performed on the multi-source sensor data stream to eliminate transmission delay differences and generate a synchronous sampling dataset, including: Extract the hardware trigger flag bits of each signal channel in the multi-source sensing data stream; Calculate the transmission link delay difference between the current pulse signal, the line displacement signal, and the tank liquid thermal imaging signal; Based on the transmission link delay difference, the line displacement signal and the tank liquid thermal imaging signal are digitally interpolated and resampled. Using the sampling time of the current pulse signal as a reference, all signals are corrected by time axis shift. The corrected signal data is packaged and reassembled according to a unified timestamp to obtain the synchronous sampling dataset.

[0008] Furthermore, the synchronous sampling dataset is analyzed to identify the axial position labels and circumferential angle labels of the diamond wire saw to be controlled, and a process parameter matrix with spatial index is constructed, including: The encoder feedback value in the synchronous sampling dataset is read and converted to obtain the axial travel distance of the diamond wire saw to be controlled; Analyze the high-frequency jitter components in the line displacement signal and demodulate the rotation phase of the diamond wire saw to be controlled; The axial travel distance is mapped to an axial position label, and the rotation phase is mapped to a circumferential angle label; The amplitude of the current pulse signal, the frequency of the line displacement signal, and the temperature value of the tank liquid thermal imaging signal are associated with the corresponding axial position label and circumferential angle label. Fill the empty nodes of the process parameter matrix to form a complete multidimensional array structure.

[0009] Further, the process parameter matrix is ​​input into a pre-trained thickness prediction network to output an instantaneous deposition rate contour map of the diamond wire saw to be controlled, including: Load the neural network weight file stored in the edge computing node and initialize the thickness prediction network; The process parameter matrix is ​​expanded into sector-shaped data slices according to the circumferential angle labels; The sector-shaped data slices are sequentially fed into the thickness prediction network for forward inference calculation; The feature map of the intermediate layer of the network is extracted and restored to a two-dimensional mesh corresponding to the geometry of the surface of the diamond wire saw to be controlled. Each node in the two-dimensional grid is assigned a deposition rate value, and the instantaneous deposition rate cloud map is plotted.

[0010] Further, based on the instantaneous deposition rate cloud map, the cumulative thickness deviation field along the length direction of the diamond wire saw to be controlled is calculated, including: Set the target nominal thickness value of the diamond wire saw to be controlled as the thickness reference plane; Traverse each grid cell of the instantaneous deposition rate cloud map and calculate the difference between the actual deposition rate and the target deposition rate; By combining the dwell time parameter of the diamond wire saw to be controlled, the difference is integrated to obtain the local overthickness or underthickness. The integral results of all circumferential meshes are summarized along the axial direction of the diamond wire saw to be controlled to generate the cross-sectional average deviation curve. The average deviation curve of the cross section is extended in the spatial domain along the length direction to construct the three-dimensional cumulative thickness deviation field; The integral results of all circumferential meshes along the axial direction of the diamond wire saw to be controlled are summarized to generate the cross-sectional average deviation curve, including: For any cross-section of the diamond wire saw to be controlled, extract the local overthickness or underthickness of all circumferential meshes on the arbitrary cross-section; Calculate the arithmetic mean of all the local overthickness or underthickness as the average thickness deviation of the arbitrary cross-section; Record the axial coordinate position corresponding to the average thickness deviation; Repeat the extraction, calculation, and recording steps until the entire length of the diamond wire saw to be controlled is traversed; Connect all axial coordinate positions and their corresponding average thickness deviation points to draw a continuous average deviation curve for the cross-section.

[0011] Furthermore, the accumulated thickness deviation field is invoked, and combined with the wire path planning information of the diamond wire saw to be controlled, to generate a partitioned voltage modulation sequence, including: The wire cutting path planning information is analyzed to obtain the linear velocity trajectory of the diamond wire saw to be controlled in each electroplating zone; Retrieve the coordinates of abnormal regions in the cumulative thickness deviation field whose deviation values ​​exceed a preset threshold; Calculate the required current density compensation coefficient for the coordinates of the abnormal region; The application duration of the current density compensation coefficient is dynamically adjusted based on the rate of change of the linear velocity trajectory. The current density compensation coefficient and the application duration are encoded into a time sequence instruction to form the partitioned voltage modulation sequence.

[0012] Further, the partitioned voltage modulation sequence is sent to the rectifier control unit to complete the closed-loop adjustment of the coating thickness of the diamond wire saw to be controlled, including: Establish a communication handshake protocol with the rectifier control unit to verify the online status of the device; The partitioned voltage modulation sequence is converted into an analog control signal or a digital pulse width modulation signal; At the moment when the diamond wire saw to be controlled enters the corresponding electroplating zone, the rectifier control unit is triggered to output the adjusted voltage level; The output current feedback value of the rectifier control unit is read back in real time and compared with the expected current value; If the comparison result exceeds the tolerance range, the bias parameters of the thickness prediction network are updated, and the adjustment process for the next cycle is restarted.

[0013] Further, the high-frequency jitter components in the wire displacement signal are analyzed, and the rotation phase of the diamond wire saw to be controlled is demodulated, including: Bandpass filtering is performed on the line displacement signal to separate the vibration waveform of a specific frequency band; The zero-crossing position of the vibration waveform is detected, and the number of oscillations per unit time is counted. The surface linear velocity fluctuation period of the diamond wire saw to be controlled is calculated based on the number of oscillations. The surface linear velocity fluctuation period is phase-matched with the preset wire saw circumference; Output the absolute rotation angle of the diamond wire saw to be controlled relative to the initial reference point, as the circumferential angle label.

[0014] Furthermore, the present invention also includes an intelligent control system for the thickness of nickel plating on a diamond wire saw, the system including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, it implements the steps of the intelligent control method for the thickness of nickel plating on a diamond wire saw as described above.

[0015] Compared with the prior art, the beneficial effects of the present invention are: Spatiotemporal alignment is performed on current pulse signals, wire displacement signals, and bath thermal imaging signals. Synchronous sampling datasets eliminate transmission delay differences between different sensor signals. The synchronous sampling datasets are analyzed to identify the axial position and circumferential angle labels of the diamond wire saw to be controlled, constructing a process parameter matrix with spatial indexing. Multi-source sensor data achieves precise matching in both temporal and spatial dimensions. Process parameters can directly correspond to the electroplating state of the diamond wire saw at different axial and circumferential positions. Parameter distortion caused by sensor signal transmission deviations is avoided, and the process parameter matrix can completely map the spatialized operating state of the entire wire saw electroplating process.

[0016] A process parameter matrix with spatial index is input into a pre-trained thickness prediction network, outputting an instantaneous deposition rate cloud map of the diamond wire saw to be controlled. Based on the instantaneous deposition rate cloud map, the cumulative thickness deviation field along the length direction of the wire saw is calculated. The cumulative thickness deviation field is combined with wire feed path planning information to generate a zoned voltage modulation sequence, which is then sent to the rectifier control unit. The instantaneous deposition rate is visualized globally in the form of a cloud map, the thickness deviation along the length direction of the wire saw can be accurately quantified, the voltage modulation command can be adapted to the thickness deviation state according to different zones of the wire saw, the control command of the rectifier control unit is matched with the wire feed path in real time, and the coating thickness adjustment forms a complete real-time closed loop. The thickness distribution deviation can be specifically corrected through zoned control. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the steps of an intelligent control method for the thickness of a nickel plating layer on a diamond wire saw according to the present invention. Figure 2 A flowchart for acquiring multi-source sensor data streams; Figure 3 A flowchart for constructing a process parameter matrix with spatial indexes to parse a synchronously sampled dataset; Figure 4 Box plot of axial distribution of nickel plating thickness deviation for diamond wire saws; Figure 5 Accumulated thickness deviation field and closed-loop control decision cloud map for nickel plating of diamond wire saw. Detailed Implementation

[0018] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] See Figure 1 This invention provides an intelligent control method for the nickel plating thickness of a diamond wire saw. The method includes: acquiring multi-source sensor data streams during the operation of the diamond wire saw to be controlled, the multi-source sensor data streams encompassing current pulse signals, wire displacement signals, and bath thermal imaging signals. A spatiotemporal alignment operation is performed on the multi-source sensor data streams to eliminate transmission delay differences and generate a synchronous sampling dataset. The synchronous sampling dataset is parsed to identify the axial position label and circumferential angle label of the diamond wire saw to be controlled, constructing a process parameter matrix with spatial index. The process parameter matrix is ​​input into a pre-trained thickness prediction network, outputting an instantaneous deposition rate cloud map of the diamond wire saw to be controlled. Based on the instantaneous deposition rate cloud map, the cumulative thickness deviation field of the diamond wire saw to be controlled along its length direction is calculated. The cumulative thickness deviation field is called, combined with the wire path planning information of the diamond wire saw to be controlled, to generate a partitioned voltage modulation sequence. The partitioned voltage modulation sequence is sent to the rectifier control unit to complete the closed-loop adjustment of the plating thickness of the diamond wire saw to be controlled.

[0020] In one embodiment of the present invention, the acquisition of multi-source sensor data streams during the operation of the diamond wire saw to be controlled includes: (See reference) Figure 2 The Hall current sensor array is activated to capture the current pulse signal as the diamond wire saw to be controlled passes the anode plate. A laser vibrometer is activated to track the surface micro-vibrations of the diamond wire saw to be controlled, generating a wire displacement signal. An infrared thermal imager is turned on to scan the surface temperature distribution of the plating solution in the electroplating tank, outputting a thermal imaging signal of the plating solution. The current pulse signal, wire displacement signal, and plating solution thermal imaging signal are encapsulated into a raw data packet and timestamped. The raw data packet is aggregated through an industrial Ethernet switch to form the multi-source sensor data stream.

[0021] In practice, multi-source sensor data streams are collected during the operation of the diamond wire saw to be controlled, including current pulse signals, wire displacement signals, and bath thermal imaging signals. After the electroplating production line starts, a Hall current sensor array placed near multiple anode plates is activated. The sampling frequency of the Hall current sensor array is set to 10kHz to capture the current pulse signals caused by electrochemical reactions as the diamond wire saw passes through each anode plate area. The amplitude of the current pulse signals fluctuates between 0.1A and 5A. In practice, a laser vibrometer installed near the wire saw guide wheel is activated. The laser beam emitted by the laser vibrometer illuminates the moving wire saw surface at a 60-degree angle. By analyzing the Doppler frequency shift of the reflected laser, the micro-vibrations on the surface of the diamond wire saw to be controlled are tracked at a sampling rate of 1MHz, thereby generating a wire displacement signal containing axial and radial displacement components. In practical implementation, an infrared thermal imager mounted above the electroplating tank is simultaneously activated. The imager scans the surface of the plating solution within the tank at a rate of 30 frames per second, with a thermal sensitivity better than 0.05℃. It outputs a thermal imaging signal reflecting the temperature field distribution within the tank, with each pixel corresponding to a temperature measurement value. It can be understood that at the data generation source, the current pulse signal, the line displacement signal, and the thermal imaging signal are synchronously encapsulated by their respective acquisition cards into raw data packets containing raw binary data and acquisition metadata. Each raw data packet is marked with an acquisition timestamp accurate to the microsecond level. In some embodiments, raw data packets generated by multiple acquisition nodes are aggregated in real time through a gigabit industrial Ethernet switch. The industrial Ethernet switch employs a time synchronization mechanism based on a precise time protocol to aggregate and initially sort raw data packets from different physical links at the data link layer, forming a multi-source sensor data stream that is sequential in time but maintains physically independent channels. Optionally, each probe of the Hall current sensor array is arranged in a one-to-one mapping with the anode plate, and the current pulse signal directly reflects the instantaneous current density of the corresponding anode area. It is understandable that the displacement signal of the wire tracked by the laser vibrometer not only includes mechanical vibration but also implicitly contains dynamic imbalance information caused by the rotation of the wire saw and uneven coating. In practical implementation, the spatial resolution of the output thermal imaging signal of the plating bath from the infrared thermal imager is matched with the size of the plating bath, enabling a correspondence between the temperature distribution and the geometric positions of the anode and cathode. The temperature measurement value of the infrared thermal imager... With pixel coordinates and time Related, among which Represents the infrared thermal image located in The pixel at the coordinates The measured surface temperature of the plating solution at that moment. In some embodiments, the encapsulated raw data packet has a unified data structure, which includes a frame header, sensor identifier, acquisition timestamp, data payload length, and data payload body. When aggregating multi-source sensor data streams, the industrial Ethernet switch adds a switch port reception timestamp to each passing raw data packet for subsequent transmission delay analysis.

[0022] In one embodiment of the present invention, a spatiotemporal alignment operation is performed on the multi-source sensor data stream to eliminate transmission delay differences and generate a synchronous sampling dataset. This includes: extracting the hardware trigger flags of each signal channel in the multi-source sensor data stream; calculating the transmission link delay difference between the current pulse signal, the line displacement signal, and the tank liquid thermal imaging signal; performing digital interpolation resampling on the line displacement signal and the tank liquid thermal imaging signal based on the transmission link delay difference; performing time-axis shift correction on all signals using the sampling time of the current pulse signal as a reference; and packaging and reassembling the corrected signal data according to a unified timestamp to obtain the synchronous sampling dataset.

[0023] In practice, spatiotemporal alignment is performed on the multi-source sensor data stream to eliminate transmission delay differences and generate a synchronous sampling dataset. Hardware trigger flags for each signal channel in the multi-source sensor data stream are extracted. Specifically, when each sensor acquisition card captures the rising edge of an external synchronization trigger pulse, it sets a dedicated hardware trigger flag in the header of the generated raw data packet. For example, when all sensors receive the same 1Hz square wave synchronization trigger pulse from the central controller, the acquisition cards for the Hall current sensor array, laser vibrometer, and infrared thermal imager will each set their hardware trigger flags in the data packets of their respective next sampling period. The transmission link delay difference between the current pulse signal, the line displacement signal, and the tank liquid thermal imaging signal is calculated. In some embodiments, this is done by comparing the timestamps of different raw data packets carrying the same hardware trigger flag arriving at the network interface card of the data processing server, and comparing them with the acquisition timestamps inherent in the raw data packets. This allows for the calculation of the end-to-end delay of each signal from the acquisition end to the processing end. For example, in a single measurement, the end-to-end delay of the current pulse signal might be 2.1 milliseconds, the end-to-end delay of the line displacement signal might be 3.8 milliseconds, and the end-to-end delay of the tank liquid thermal imaging signal might be 15.4 milliseconds. The transmission link delay difference is the difference between these delay times. It is understood that the transmission link delay difference is a dynamically changing value, affected by factors such as network load and switch queue length. In practical implementation, based on the calculated transmission link delay difference, digital interpolation resampling is performed on the line displacement signal and the tank liquid thermal imaging signal. Optionally, a cubic spline interpolation algorithm is used to upsample the original discrete data sequences of the line displacement signal and the tank liquid thermal imaging signal, which have a long transmission delay, to generate new data points that perfectly match the sampling time of the current pulse signal. For example, if the original sampling rate of the line displacement signal is 1MHz, it contains 2000 original data points within a 2ms time window. After interpolation resampling, 20 new data points aligned with the 10kHz sampling rate of the current pulse signal are generated within the same time window. The timing of the new data points... Sampling time with reference signal (current pulse signal) Alignment, satisfying the following relationship: in: Indicates the first The timestamps of the line displacement signal or tank liquid thermal imaging signal data points generated after interpolation and resampling. Indicates the first The timestamps of the raw data points of each current pulse signal and This is the index number. In some embodiments, the sampling time of the current pulse signal is used as a reference to perform time axis translation correction on all signals. This means that the data sequences obtained after interpolation and resampling of the line displacement signal and the tank liquid thermal imaging signal are shifted along the time axis in the past direction according to their respective transmission link delay differences, so that the time zero point of all signal data after correction corresponds to the actual time of the same physical event. It can be understood that time axis translation correction is essentially to unify the timing of all signals to the same absolute time reference frame. In specific implementation, the corrected signal data are packaged and recombined according to a unified timestamp to obtain a synchronous sampling dataset. Optionally, the synchronous sampling dataset is stored in matrix form, where each row of the matrix corresponds to a unified, high-precision timestamp, and the columns of the matrix store the amplitude of the current pulse signal, the axial and radial component values ​​of the line displacement signal, and the temperature statistics of the region of interest in the tank liquid thermal imaging signal at that time, respectively, thereby ensuring that any row of data extracted at any time strictly corresponds to the multi-dimensional observation state of the same physical instant.

[0024] In one embodiment of the present invention, the synchronous sampling dataset is parsed to identify the axial position label and circumferential angle label of the diamond wire saw to be controlled, and a process parameter matrix with spatial index is constructed, including: (see reference) Figure 3 The encoder feedback value in the synchronous sampling dataset is read and converted to obtain the axial travel distance of the diamond wire saw to be controlled. The high-frequency jitter component in the wire displacement signal is analyzed, and the rotation phase of the diamond wire saw to be controlled is demodulated. This process specifically involves performing bandpass filtering on the wire displacement signal to separate the vibration waveform of a specific frequency band, detecting the zero-crossing position of the vibration waveform, counting the number of oscillations per unit time, calculating the surface linear velocity fluctuation period of the diamond wire saw to be controlled based on the number of oscillations, matching the surface linear velocity fluctuation period with the preset wire saw circumference, and outputting the absolute rotation angle of the diamond wire saw to be controlled relative to the initial reference point as the circumferential angle label. The axial travel distance is mapped to an axial position label, and the rotation phase is mapped to a circumferential angle label. The amplitude of the current pulse signal, the frequency of the wire displacement signal, and the temperature value of the bath thermal imaging signal are associated with the corresponding axial position label and circumferential angle label. The empty nodes of the process parameter matrix are filled to form a complete multidimensional array structure.

[0025] In practical implementation, the synchronous sampling dataset is analyzed to identify the axial position and circumferential angle labels of the diamond wire saw to be controlled. A process parameter matrix with spatial index is constructed, and the encoder feedback value in the synchronous sampling dataset is read. The encoder is coaxially connected to the main drive wheel and generates a fixed number of pulses per revolution. In practical implementation, the encoder generates 10,000 pulses per revolution. By accumulating the number of pulses and multiplying by a fixed coefficient, the axial travel distance of the diamond wire saw to be controlled from the initial zero point is calculated. For example, if 250,000 pulses are accumulated, the corresponding axial travel distance is 25.0 meters. The high-frequency jitter components in the wire displacement signal are analyzed, and the rotation phase of the diamond wire saw to be controlled is demodulated. In practical implementation, bandpass filtering is performed on the wire displacement signal to separate the vibration waveform of a specific frequency band. It can be understood that the passband frequency range of the bandpass filter is set to 500Hz to 1500Hz to retain the high-frequency characteristic vibration caused by the periodic passage of diamond particles on the wire saw surface through the measurement point, while suppressing low-frequency mechanical vibration and electrical noise. In some embodiments, the zero-crossing position of the vibration waveform after bandpass filtering is detected, i.e., the time point when the signal voltage crosses zero level from positive to negative or from negative to positive. The number of oscillations per unit time is counted; for example, 12 complete oscillation cycles are detected within a 1-millisecond time window. The surface linear velocity fluctuation period of the diamond wire saw to be controlled is calculated based on the statistically obtained number of oscillations. It can be understood that the surface linear velocity fluctuation period is inversely proportional to the oscillation period. In a specific implementation, the calculated surface linear velocity fluctuation period is phase-matched with a preset wire saw circumference. The preset wire saw circumference is a known fixed parameter, for example, a wire saw circumference of 0.5 meters. By mapping the fluctuation period to a length range of 0 to 0.5 meters, the relative position on the circumference of the wire saw corresponding to the current vibration waveform sample point can be calculated. The absolute rotation angle of the diamond wire saw to be controlled relative to the initial reference point is output as a circumferential angle label. The calculation formula is: in: This represents the cumulative number of sampling points from the initial time to the current time. It is the preset wire saw circumference. It is the first The instantaneous surface linear velocity at each sampling point is calculated based on the number of oscillations. This refers to the sampling time interval. `mod360` indicates a modulo operation on 360, with the result in degrees. In specific implementations, the axial travel distance is mapped to an axial position label, and the rotation phase is mapped to a circumferential angle label. For example, a point with an axial travel distance of 25.0 meters and an absolute rotation angle of 123.5 degrees has an axial position label of "25000mm" and a circumferential angle label of "123.5°". The amplitude of the current pulse signal, the frequency of the line displacement signal, and the temperature value of the bath liquid thermal imaging signal are associated with the corresponding axial position and circumferential angle labels. Optionally, the frequency of the line displacement signal refers to the dominant frequency of the high-frequency vibration waveform within a local time window after bandpass filtering. Null nodes in the process parameter matrix are filled to form a complete multidimensional array structure. In some embodiments, due to non-uniform sampling or data loss, there may be missing positions in the process parameter matrix. Nearest neighbor interpolation is used, employing the average value of the values ​​of the four adjacent grid points (top, bottom, left, and right) to fill the missing nodes. For an example of constructing the process parameter matrix, refer to Table 1, which shows process parameter segments at different circumferential angles around an axial position of 25000mm.

[0026] Table 1: Data Fragment Table of Process Parameter Matrix Optionally, the process parameter matrix is ​​organized in memory as a three-dimensional tensor, with the first dimension indexing the axial position, the second dimension indexing the circumferential angle, and the third dimension storing the values ​​of different sensing signals.

[0027] In one embodiment of the present invention, the process parameter matrix is ​​input into a pre-trained thickness prediction network to output an instantaneous deposition rate cloud map of the diamond wire saw to be controlled. This includes: loading a neural network weight file stored in the edge computing nodes and initializing the thickness prediction network; expanding the process parameter matrix into fan-shaped data slices according to the circumferential angle labels; sequentially feeding the fan-shaped data slices into the thickness prediction network for forward inference calculation; extracting the feature maps of the intermediate layers of the network and restoring them into a two-dimensional grid corresponding to the geometry of the surface of the diamond wire saw to be controlled; assigning deposition rate values ​​to each node in the two-dimensional grid to obtain the instantaneous deposition rate cloud map; and calculating the cumulative thickness deviation field of the diamond wire saw to be controlled along its length direction based on the instantaneous deposition rate cloud map. This includes: setting the target nominal thickness value of the diamond wire saw to be controlled as a thickness reference plane; and traversing each grid cell of the instantaneous deposition rate cloud map to calculate the difference between the actual deposition rate and the target deposition rate. By integrating the dwell time parameter of the diamond wire saw to be controlled, the difference is calculated to obtain the local overthickness or underthickness. The integration results of all circumferential meshes are summarized along the axial direction of the diamond wire saw to generate a cross-sectional average deviation curve. Specifically, for any cross-section of the diamond wire saw, the local overthickness or underthickness of all circumferential meshes on that cross-section is extracted, and the arithmetic mean of all such local overthickness or underthickness is calculated as the average thickness deviation of that cross-section. The axial coordinate position corresponding to the average thickness deviation is recorded. This extraction, calculation, and recording process is repeated until the entire length of the diamond wire saw is traversed. All axial coordinate positions and their corresponding average thickness deviation points are connected to form a continuous cross-sectional average deviation curve. The cross-sectional average deviation curve is then spatially expanded along its length to construct a three-dimensional cumulative thickness deviation field.

[0028] In practice, the process parameter matrix is ​​input into a pre-trained thickness prediction network, which outputs an instantaneous deposition rate cloud map of the diamond wire saw to be controlled. The neural network weight file stored in the edge computing node is loaded, and the thickness prediction network is initialized. In this implementation, the thickness prediction network is a convolutional neural network with an encoder-decoder structure. Its weight file contains the connection weights and bias parameters of each layer of the network. The weight file is trained using historical production data and loaded into memory through the network port. The process parameter matrix is ​​expanded into fan-shaped data slices according to circumferential angle labels. In some embodiments, the dimensions of the process parameter matrix are (number of axial positions, number of circumferential angles, number of sensor features). It is divided into multiple two-dimensional slices according to the circumferential angle dimension, each slice corresponding to a fixed circumferential angle, with dimensions of (number of axial positions, number of sensor features). The fan-shaped data slices are sequentially fed into the thickness prediction network for forward inference calculation. Each fan-shaped data slice independently undergoes feature extraction through the encoder part of the network, and then feature mapping and regression are performed through the decoder part. Feature maps from the intermediate layers of the network are extracted and reconstructed into a two-dimensional mesh corresponding to the geometry of the diamond wire saw surface to be controlled. In specific implementations, the decoder of the thickness prediction network outputs a two-dimensional array. The number of rows in this array equals the number of grid points at the axial position, and the number of columns equals the number of circumferential angle divisions. Each element in the two-dimensional array corresponds one-to-one with a region on the wire saw surface at a specific axial and circumferential coordinate. Deposition rate values ​​are assigned to each node in the two-dimensional mesh, and an instantaneous deposition rate contour map is plotted. For example, a grid node located at an axial index of 100 and a circumferential index of 30 is assigned a value of 0.85 micrometers per minute. The instantaneous deposition rate contour map is presented on the display in pseudo-color format, where colors from blue to red represent deposition rates from low to high.

[0029] Based on the instantaneous deposition rate contour map, the cumulative thickness deviation field along the length direction of the diamond wire saw to be controlled is calculated. A target nominal thickness value for the diamond wire saw to be controlled is set as the thickness reference plane. The target nominal thickness value is a constant, for example, set to 20.0 micrometers. Each grid cell of the instantaneous deposition rate contour map is traversed, and the difference between the actual deposition rate and the target deposition rate is calculated. The target deposition rate is a constant rate value derived from the total process time and the target thickness. Combining the dwell time parameter of the diamond wire saw to be controlled, the difference is integrated to obtain the local overthickness or underthickness. The dwell time parameter refers to the time taken for a point on the wire saw surface to pass through the effective electroplating area. The calculation formula is: in: It is an index of a two-dimensional grid cell. and These are the time points when the wire saw surface point corresponding to the grid cell enters and leaves the electroplating area. Is the grid cell at time... The actual deposition rate, This refers to the target deposition rate. The integral results of all circumferential meshes along the axis of the diamond wire saw to be controlled are summarized to generate a cross-sectional average deviation curve. In practice, for any cross-section of the diamond wire saw to be controlled, the local overthickness or underthickness of all circumferential meshes on that cross-section is extracted. The arithmetic mean of all local overthickness or underthickness is calculated as the average thickness deviation of that cross-section. The axial coordinate position corresponding to the average thickness deviation is recorded. This extraction, calculation, and recording process is repeated until the entire length of the diamond wire saw to be controlled is traversed. All axial coordinate positions and their corresponding average thickness deviation points are connected to form a continuous cross-sectional average deviation curve. It can be understood that the cross-sectional average deviation curve reflects the overall trend of coating thickness deviation along the length of the wire saw. The cross-sectional average deviation curve is spatially extended along the length direction to construct a three-dimensional cumulative thickness deviation field. Optionally, the cumulative thickness deviation field is a three-dimensional array, whose three dimensions represent axial position, circumferential angle, and thickness deviation value, respectively. At each axial position, the average thickness deviation value corresponding to that position is repeated along the circumference to form a deviation distribution loop for that axial section. In some embodiments, partial data of the cumulative thickness deviation field, as shown in Table 2, illustrates the distribution of deviation values ​​in the axial position range from 1000 mm to 1002 mm.

[0030] Table 2: Data Segment Table of Cumulative Thickness Deviation Field Optionally, when constructing the cumulative thickness deviation field, linear interpolation can be used to fill the missing grid points in the axial and / or circumferential directions.

[0031] See Figure 4This study presents the distribution characteristics of the average thickness deviation of the coating cross-section in diamond wire saws at different axial positions (0-200mm to 1800-2000mm). Its core purpose is to verify the engineering consistency between the thickness prediction network inference results and the cumulative thickness deviation field calculation. In terms of statistical dimensions, each axial interval box plot includes the median (horizontal line inside the box), the interquartile range (box height, representing the fluctuation range of the middle 50% of the data), and whiskers (covering the normal fluctuation range within 1.5 times the interquartile range). The pink / blue boxes represent the core distribution area of ​​the thickness deviation in each axial interval, the endpoints of the whiskers define the normal deviation range, and discrete points exceeding the whiskers are potential abnormal deviation points. Analysis of the distribution characteristics shows that within the range of 0-200mm to 1200-1400mm, the median of the box fluctuates slightly around the 0μm baseline. The medians in the 200-400mm and 400-600mm ranges are slightly higher than 0μm, reflecting local positive deviation characteristics. The 1200-1400mm and 1800-2000mm ranges present blue boxes with medians slightly lower than 0μm, corresponding to a negative deviation distribution. Furthermore, the interquartile range of the boxes in these two ranges is relatively narrower, reflecting a higher concentration of local deviation. From the perspective of fluctuation range and outliers, the coverage range of the beard lines in the entire axial range is within ±0.2μm, and no extreme outliers exceeding 1.5 times the interquartile range have appeared, indicating that the overall coating thickness deviation is within the process controllable range. Among them, the 800-1000mm range has the longest beard line length, which represents the largest fluctuation range of the thickness deviation in this axial segment and is the high fluctuation area of ​​the deviation distribution. In contrast, the beard line length in the 1000-1200mm range is significantly shortened, and the deviation fluctuation amplitude drops sharply, which corresponds to the stabilization calculation results of the cumulative thickness deviation field in this range. Combined with the process control logic, the distribution characteristics of the box plot can be directly mapped to the generation basis of the partitioned voltage modulation sequence: for the high fluctuation range of 800-1000mm, it is necessary to increase the adjustment range of the current density compensation coefficient and reduce its interquartile range to reduce deviation fluctuation; for the negative deviation ranges of 1200-1400mm and 1800-2000mm, it is necessary to adjust the output voltage of the current unit in a targeted manner to push the median to converge towards the 0μm baseline, and finally achieve full-axial closed-loop uniform control of the coating thickness.

[0032] In one embodiment of the present invention, the cumulative thickness deviation field is invoked, combined with the wire path planning information of the diamond wire saw to be controlled, to generate a partitioned voltage modulation sequence, including: parsing the wire path planning information to obtain the linear velocity trajectory of the diamond wire saw to be controlled in each electroplating zone; retrieving the coordinates of abnormal areas in the cumulative thickness deviation field where the deviation value exceeds a preset threshold; calculating the required current density compensation coefficient for the coordinates of the abnormal areas; dynamically adjusting the application duration of the current density compensation coefficient according to the rate of change of the linear velocity trajectory; encoding the current density compensation coefficient and the application duration into a time sequence instruction to form the partitioned voltage modulation sequence; and sending the partitioned voltage modulation sequence to the rectifier control unit to complete the closed-loop adjustment of the plating thickness of the diamond wire saw to be controlled, including: establishing a communication handshake protocol with the rectifier control unit to verify the online status of the equipment; converting the partitioned voltage modulation sequence into an analog control signal or a digital pulse width modulation signal; and triggering the rectifier control unit to output the adjusted voltage level when the diamond wire saw to be controlled enters the corresponding electroplating zone. The output current feedback value of the rectifier control unit is read back in real time and compared with the expected current value. If the comparison result exceeds the tolerance range, the bias parameters of the thickness prediction network are updated, and the adjustment process for the next cycle is restarted.

[0033] In practical implementation, the cumulative thickness deviation field is invoked, combined with the wire path planning information of the diamond wire saw to be controlled, to generate a zoned voltage modulation sequence. The wire path planning information is parsed to obtain the linear velocity trajectory of the diamond wire saw in each electroplating zone. The wire path planning information is stored in the database of the programmable logic controller, which includes the preset speed curves of the wire saw as it passes through different anode plate areas from the feed wheel to the take-up wheel. For example, in anode area A1, the linear velocity is planned to be a constant 5.0 meters per minute, and in anode area A2, the linear velocity is planned to linearly increase from 5.0 meters per minute to 5.5 meters per minute. The coordinates of abnormal areas in the cumulative thickness deviation field whose deviation values ​​exceed a preset threshold are retrieved. The preset threshold is set according to process requirements. For example, areas with an absolute thickness deviation value exceeding 0.5 micrometers are defined as abnormal areas. The retrieval process traverses the three-dimensional array of the cumulative thickness deviation field and records the axial position coordinates and circumferential angle coordinates of all grid points that meet the condition "(deviation value > +0.5 micrometers) or (deviation value < -0.5 micrometers)". For the coordinates of the retrieved abnormal regions, calculate the required current density compensation coefficient. The calculation formula is: in: This is the cumulative thickness deviation value at the coordinate point of the abnormal area. It is the target nominal thickness value. This is a process correction factor related to the electroplating current efficiency, for example, set to 0.95. The application duration of the current density compensation coefficient is dynamically adjusted according to the rate of change of the linear velocity trajectory. In some embodiments, if the linear velocity is increasing, the application duration of the compensation coefficient is shortened accordingly; if the linear velocity is decreasing, the application duration of the compensation coefficient is extended accordingly. The adjustment can be based on the formula In progress, among which It is the basic compensation time calculated based on the deviation. It is the reference linear velocity. This is the current actual linear velocity. The current density compensation coefficient and the application duration are encoded into a time-series instruction to form a partitioned voltage modulation sequence. Optionally, the partitioned voltage modulation sequence is a list, and each element in the list contains the following fields: start timestamp, target anode zone number, voltage regulation value, and duration. The voltage regulation value is obtained by multiplying the current density compensation coefficient by the reference operating voltage.

[0034] The partitioned voltage modulation sequence is sent to the rectifier control unit to complete the closed-loop adjustment of the coating thickness of the diamond wire saw to be controlled, establish a communication handshake protocol with the rectifier control unit, and verify the online status of the equipment. In specific implementations, a query command is sent to the rectifier control unit via the MODBUSTCP protocol, and the system waits for the rectifier control unit to return a response frame containing its device identifier and "ready" status. The partitioned voltage modulation sequence is converted into an analog control signal or a digital pulse width modulation signal. It can be understood that when the rectifier control unit supports analog input, the voltage regulation value is converted into a 0-10V analog voltage signal; when the rectifier control unit supports digital pulse width modulation signal input, the voltage regulation value and duration are jointly encoded into a pulse width modulation signal with a specific duty cycle and frequency. When the diamond wire saw to be controlled enters the corresponding electroplating zone, the rectifier control unit is triggered to output the adjusted voltage level. In some embodiments, the axial position encoder reading of the wire saw is compared with the starting position of the command in the partitioned voltage modulation sequence in real time. When the two match, an execution command is sent to the rectifier control unit. The system reads back the output current feedback value of the rectifier control unit in real time and compares it with the expected current value, which is calculated based on the adjusted voltage level and a known circuit impedance model. If the comparison result exceeds the tolerance range, the bias parameters of the thickness prediction network are updated, and the adjustment process for the next cycle is restarted. Optionally, the tolerance range is set to ±2% of the expected current value. If the read-back current feedback value exceeds this range for three consecutive sampling cycles, the deviation is recorded. After the current adjustment cycle ends, this deviation data is used as a new training sample for online fine-tuning of the bias parameters of the output layer in the thickness prediction network. Subsequently, the system automatically starts the next complete adjustment process from data acquisition to voltage delivery based on the updated network parameters.

[0035] See Figure 5 The diagram uses an axial position as the length dimension and a circumferential angle as the circumferential dimension to construct a spatial grid. Color gradations represent the deviation between the actual coating thickness and the target nominal thickness of each grid cell. Based on this, the system traverses and retrieves the coordinates of abnormal areas exceeding the ±0.5μm process threshold, and calculates the current density compensation coefficient by combining the linear velocity trajectory within the electroplating zone. For excessively thick positive deviation areas distributed circumferentially, a corresponding anode area voltage reduction modulation command is generated to decrease the deposition rate; for excessively thin negative deviation areas, a voltage boost compensation command is generated. Simultaneously, the compensation application duration is dynamically adjusted based on the linear velocity change rate, and the compensation coefficient and duration are encoded into a time-series command and converted into an analog quantity or PWM control signal. The system establishes a communication handshake with the rectifier control unit via ModbusTCP. When the linear saw reaches the corresponding axial position, voltage regulation is triggered, and the current feedback value is read back in real time. If the deviation exceeds the ±2% tolerance range, the thickness prediction network bias parameters are fine-tuned online using the deviation data, completing the closed-loop regulation of the entire process. This deviation field cloud map fully realizes the quantification of the spatial distribution of coating thickness, the precise location of abnormal areas, and the accurate mapping of control commands, providing key topological basis and decision support for intelligent control of coating uniformity.

[0036] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A method for intelligent control of the nickel plating thickness on a diamond wire saw, characterized in that, include: Collect multi-source sensor data streams during the operation of the diamond wire saw to be controlled. The multi-source sensor data streams include current pulse signals, wire displacement signals, and bath liquid thermal imaging signals. Perform a spatiotemporal alignment operation on the multi-source sensor data stream to eliminate transmission delay differences and generate a synchronous sampling dataset; The synchronous sampling dataset is analyzed to identify the axial position label and circumferential angle label of the diamond wire saw to be controlled, and a process parameter matrix with spatial index is constructed. The process parameter matrix is ​​input into a pre-trained thickness prediction network, which outputs an instantaneous deposition rate cloud map of the diamond wire saw to be controlled. Based on the instantaneous deposition rate cloud map, the cumulative thickness deviation field along the length direction of the diamond wire saw to be controlled is calculated; The accumulated thickness deviation field is invoked, and combined with the wire path planning information of the diamond wire saw to be controlled, to generate a partitioned voltage modulation sequence; The partition voltage modulation sequence is sent to the rectifier control unit to complete the closed-loop adjustment of the coating thickness of the diamond wire saw to be controlled.

2. The intelligent control method for the thickness of the nickel plating layer on a diamond wire saw as described in claim 1, characterized in that, The acquisition of multi-source sensor data streams during the operation of the diamond wire saw to be controlled includes: The Hall current sensor array is activated to capture the current pulse signal as the diamond wire saw to be controlled passes the anode plate; Activate the laser vibration meter to track the surface micro-vibration of the diamond wire saw to be controlled and generate a wire displacement signal; Turn on the infrared thermal imager to scan the surface temperature distribution of the plating solution in the electroplating tank and output the thermal imaging signal of the solution. The current pulse signal, line displacement signal, and tank liquid thermal imaging signal are encapsulated into a raw data packet and marked with a timestamp of the acquisition. The original data packets are aggregated by an industrial Ethernet switch to form the multi-source sensor data stream.

3. The intelligent control method for the thickness of the nickel plating layer on a diamond wire saw as described in claim 2, characterized in that, Perform a spatiotemporal alignment operation on the multi-source sensor data stream to eliminate transmission delay differences and generate a synchronous sampling dataset, including: Extract the hardware trigger flag bits of each signal channel in the multi-source sensing data stream; Calculate the transmission link delay difference between the current pulse signal, the line displacement signal, and the tank liquid thermal imaging signal; Based on the transmission link delay difference, the line displacement signal and the tank liquid thermal imaging signal are digitally interpolated and resampled. Using the sampling time of the current pulse signal as a reference, all signals are corrected by time axis shift. The corrected signal data is packaged and reassembled according to a unified timestamp to obtain the synchronous sampling dataset.

4. The intelligent control method for the thickness of the nickel plating layer on a diamond wire saw as described in claim 3, characterized in that, The synchronous sampling dataset is parsed to identify the axial position labels and circumferential angle labels of the diamond wire saw to be controlled, and a process parameter matrix with spatial index is constructed, including: The encoder feedback value in the synchronous sampling dataset is read and converted to obtain the axial travel distance of the diamond wire saw to be controlled; Analyze the high-frequency jitter components in the line displacement signal and demodulate the rotation phase of the diamond wire saw to be controlled; The axial travel distance is mapped to an axial position label, and the rotation phase is mapped to a circumferential angle label; The amplitude of the current pulse signal, the frequency of the line displacement signal, and the temperature value of the tank liquid thermal imaging signal are associated with the corresponding axial position label and circumferential angle label. Fill the empty nodes of the process parameter matrix to form a complete multidimensional array structure.

5. The intelligent control method for the thickness of the nickel plating layer on a diamond wire saw as described in claim 4, characterized in that, The process parameter matrix is ​​input into a pre-trained thickness prediction network, which outputs an instantaneous deposition rate contour map of the diamond wire saw to be controlled, including: Load the neural network weight file stored in the edge computing node and initialize the thickness prediction network; The process parameter matrix is ​​expanded into sector-shaped data slices according to the circumferential angle labels; The sector-shaped data slices are sequentially fed into the thickness prediction network for forward inference calculation; The feature map of the intermediate layer of the network is extracted and restored to a two-dimensional mesh corresponding to the geometry of the surface of the diamond wire saw to be controlled. Each node in the two-dimensional grid is assigned a deposition rate value, and the instantaneous deposition rate cloud map is plotted.

6. The intelligent control method for the thickness of the nickel plating layer on a diamond wire saw as described in claim 5, characterized in that, Based on the instantaneous deposition rate contour map, the cumulative thickness deviation field along the length direction of the diamond wire saw to be controlled is calculated, including: Set the target nominal thickness value of the diamond wire saw to be controlled as the thickness reference plane; Traverse each grid cell of the instantaneous deposition rate cloud map and calculate the difference between the actual deposition rate and the target deposition rate; By combining the dwell time parameter of the diamond wire saw to be controlled, the difference is integrated to obtain the local overthickness or underthickness. The integral results of all circumferential meshes are summarized along the axial direction of the diamond wire saw to be controlled to generate the cross-sectional average deviation curve. The average deviation curve of the cross section is extended in the spatial domain along the length direction to construct the three-dimensional cumulative thickness deviation field; The integral results of all circumferential meshes along the axial direction of the diamond wire saw to be controlled are summarized to generate the cross-sectional average deviation curve, including: For any cross-section of the diamond wire saw to be controlled, extract the local overthickness or underthickness of all circumferential meshes on the arbitrary cross-section; Calculate the arithmetic mean of all the local overthickness or underthickness as the average thickness deviation of the arbitrary cross-section; Record the axial coordinate position corresponding to the average thickness deviation; Repeat the extraction, calculation, and recording steps until the entire length of the diamond wire saw to be controlled is traversed; Connect all axial coordinate positions and their corresponding average thickness deviation points to draw a continuous average deviation curve for the cross-section.

7. The intelligent control method for the thickness of the nickel plating layer on a diamond wire saw as described in claim 6, characterized in that, By invoking the accumulated thickness deviation field and combining it with the wire path planning information of the diamond wire saw to be controlled, a partitioned voltage modulation sequence is generated, including: The wire cutting path planning information is analyzed to obtain the linear velocity trajectory of the diamond wire saw to be controlled in each electroplating zone; Retrieve the coordinates of abnormal regions in the cumulative thickness deviation field whose deviation values ​​exceed a preset threshold; Calculate the required current density compensation coefficient for the coordinates of the abnormal region; The application duration of the current density compensation coefficient is dynamically adjusted based on the rate of change of the linear velocity trajectory. The current density compensation coefficient and the application duration are encoded into a time sequence instruction to form the partitioned voltage modulation sequence.

8. The intelligent control method for the thickness of the nickel plating layer on a diamond wire saw as described in claim 7, characterized in that, The partition voltage modulation sequence is sent to the rectifier control unit to complete the closed-loop adjustment of the coating thickness of the diamond wire saw to be controlled, including: Establish a communication handshake protocol with the rectifier control unit to verify the online status of the device; The partitioned voltage modulation sequence is converted into an analog control signal or a digital pulse width modulation signal; At the moment when the diamond wire saw to be controlled enters the corresponding electroplating zone, the rectifier control unit is triggered to output the adjusted voltage level; The output current feedback value of the rectifier control unit is read back in real time and compared with the expected current value; If the comparison result exceeds the tolerance range, the bias parameters of the thickness prediction network are updated, and the adjustment process for the next cycle is restarted.

9. The intelligent control method for the thickness of the nickel plating layer on a diamond wire saw as described in claim 8, characterized in that, Analyzing the high-frequency jitter components in the wire displacement signal and demodulating the rotation phase of the diamond wire saw to be controlled includes: Bandpass filtering is performed on the line displacement signal to separate the vibration waveform of a specific frequency band; The zero-crossing position of the vibration waveform is detected, and the number of oscillations per unit time is counted. The surface linear velocity fluctuation period of the diamond wire saw to be controlled is calculated based on the number of oscillations. The surface linear velocity fluctuation period is phase-matched with the preset wire saw circumference; Output the absolute rotation angle of the diamond wire saw to be controlled relative to the initial reference point, as the circumferential angle label.

10. An intelligent control system for the nickel plating thickness of a diamond wire saw, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent control method for the thickness of nickel plating on a diamond wire saw as described in any one of claims 1 to 9.

Images