Diamond electro-discharge composite cutting system for precise hard and brittle substrate

Abstract

The present application relates to the technical field of numerical control, and discloses a diamond electro-discharge combined cutting regulation system for precise hard and brittle substrates, which comprises a control module, a wire saw cutting module, a pulse power supply and a thickness adjusting mechanism. The control module is electrically connected with the rest modules to generate process parameters and issue control instructions. The thickness adjusting mechanism adjusts the spacing of the diamond wire saw according to the instructions. The pulse power supply generates high-frequency pulse discharge between the wire saw and the hard and brittle substrate to form a discharge channel. The control module drives the wire saw to mechanically grind the substrate, and controls the pulse power supply to melt and vaporize the surface material of the substrate, and the mechanical grinding is used to remove the discharge metamorphic layer synchronously. The present application realizes the compatibility of pulse discharge and mechanical grinding in the physical space by real-time monitoring and multivariable closed-loop regulation of the processing gap state by the control module, can independently cope with abnormal working conditions, and effectively reduces the cutting resistance and the risk of wire breakage.

Description

Technical Field

[0001] This invention relates to the field of numerical control technology, specifically to a diamond electrical discharge composite cutting control system for precision hard and brittle substrates. Background Technology

[0002] Silicon carbide, sapphire, and other hard and brittle substrates are widely used in the semiconductor and optoelectronic fields. Currently, their slicing process mainly relies on pure mechanical grinding using diamond wire saws. Due to the high hardness of these substrates, the cutting resistance between the wire saw and the material is large and fluctuates wildly during mechanical cutting. This not only leads to increased diamond particle shedding and a higher risk of high-frequency wire breakage, but also easily causes severe mechanical damage such as microcracks on the surface and subsurface of the substrate, increasing the cost of subsequent grinding and polishing.

[0003] To overcome the high resistance of purely mechanical cutting, existing technologies have attempted to incorporate wire electrical discharge machining (EDM). However, relying solely on high-temperature melting to remove material, EDM faces challenges when dealing with deep, narrow kerfs in large, hard, and brittle ingots. The working fluid struggles to effectively penetrate the bottom of the machining area, and the resulting electrolytic corrosion products cannot be promptly removed. This deterioration in chip removal leads to frequent secondary discharges, short circuits, and abnormal arcing within the machining gap, severely limiting cutting efficiency. Furthermore, pure EDM leaves a thick layer of discharge-induced degradation on the substrate surface, making it difficult to meet the surface roughness requirements of precision manufacturing.

[0004] The industry has proposed a composite machining approach that combines electrical discharge machining (EDM) with mechanical grinding. However, in actual operation, EDM requires maintaining a physical discharge gap at the micrometer level, while mechanical grinding with diamond wire saws requires the wire saw to maintain close physical contact with the substrate surface. These two processes inherently interfere with each other on a physical scale. Existing cutting equipment and control systems simply combine the power supply unit and machine tool feed functions, lacking real-time feedback and multi-variable collaborative control mechanisms for abnormal processing gap conditions. If localized material unevenness or chip removal obstruction occurs during cutting, the system often relies on a single stop or wire retraction action for coarse adjustments. It cannot dynamically restore and maintain the physical balance between discharge and mechanical contact through the underlying linkage of electrical and mechanical parameters, easily leading to servo lock-up or physical wire breakage, making it difficult to meet the demands for continuous, efficient, and high-yield mass production of hard and brittle substrates. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a diamond electrical discharge composite cutting control system for precision hard and brittle substrates. This system solves the problems of high cutting resistance, severe diamond wear, and easy generation of microcrack damage layers caused by single mechanical wire saws in existing precision hard and brittle substrate cutting technologies.

[0006] To achieve the above objectives, the present invention provides the following technical solution: This invention provides a diamond electrical discharge machining (EDM) composite cutting control system for precision hard and brittle substrates, including a control module, a wire saw cutting module, a pulse power supply, and a thickness adjustment mechanism. The wire saw cutting module includes a diamond wire saw. The thickness adjustment mechanism adjusts the spacing of the diamond wire saw according to control commands issued by the control module. The output terminal of the pulse power supply is connected between the diamond wire saw and the hard and brittle substrate to generate high-frequency pulse discharge to form a discharge channel. The control module controls the wire saw cutting module to drive the diamond wire saw to mechanically grind the hard and brittle substrate, while simultaneously controlling the pulse power supply to melt and vaporize the surface material of the hard and brittle substrate through the discharge channel. The mechanical grinding simultaneously removes the discharge-modified layer.

[0007] Furthermore, the control module is electrically connected to the wire saw cutting module, the pulse power supply, and the thickness adjustment mechanism, respectively, and is used to receive processing task parameters, generate composite cutting process parameters, and issue control commands.

[0008] Furthermore, the wire saw cutting module is internally configured with a guide wheel structure, which includes a multi-wire guide wheel and a wire guide slider that supports the multi-wire guide wheel. The thickness adjustment mechanism is mechanically connected to the wire guide slider in the guide wheel structure. By controlling the linear displacement of the wire guide slider, the physical span between the multi-wire guide wheels is changed, thereby realizing the adjustment of the working line spacing of the diamond wire saw forming the cutting mesh.

[0009] Furthermore, the control module invokes a big data prediction model and a multi-process parameter optimization algorithm. The multi-process parameter optimization algorithm generates candidate solution vectors in the parameter solution space, and the candidate solution vectors are input into the big data prediction model to obtain predicted values ​​for surface roughness and warpage. A fitness function is then constructed to iteratively optimize the candidate solution vectors. After a set maximum number of iterations or when the convergence condition is met, the control module extracts the globally optimal individual with the highest brightness evaluation value from the population. The brightness evaluation value is calculated by the fitness function. The composite cutting parameter combination of the globally optimal individual is converted into low-level control commands and sent to the pulse power supply and the wire saw cutting module.

[0010] Furthermore, the multi-process parameter optimization algorithm is specifically the firefly algorithm. The control module defines the combination of composite cutting process parameters as individuals in the optimization population, converts the predicted weighted index of surface roughness and warpage into the absolute brightness of the individual, and outputs the parameter solution that optimizes the processing quality index in the parameter solution space by simulating the attraction and position update mechanism between populations.

[0011] Furthermore, the thickness adjustment mechanism includes an independent variable spacing structure, which includes a servo motor and a ball screw. The control module identifies the target slice thickness in the processing task parameters, and after compensating for the target slice thickness by combining the physical dimensions of the cutting line and the cutting gap caused by the discharge erosion effect, maps it to a physical target spacing to generate the control command. The servo motor receives the control command and drives the ball screw, which in turn moves the guide wire slider in the independent variable spacing structure. The control module compares the deviation between the target pulse number and the actual feedback position, and dynamically adjusts the frequency and duty cycle of the subsequent output pulses so that the guide wire slider stops at the working coordinate corresponding to the physical target spacing.

[0012] Furthermore, the independent variable spacing structure also includes a guide mechanism, a linear guide rail, a guide rail brake, and a displacement sensor. The guide rail brake includes a friction plate. After the servo motor reaches the working coordinate, the control module controls the guide rail brake to squeeze the friction plate, thereby locking the guide mechanism onto the linear guide rail.

[0013] Furthermore, after entering the locked state, the control module reads the actual physical distance coordinates through the displacement sensor and identifies the zero-position deviation from the physical target. When the zero-position deviation is greater than the tolerance threshold, the control module releases the locked state and sends a pulse to the servo motor to perform feed compensation. This process is repeated until the zero-position deviation converges to the tolerance threshold. If the zero-position deviation still does not converge to the tolerance threshold after reaching the maximum number of compensation iterations, the control module forcibly interrupts the compensation process and sends a mechanical abnormality alarm signal to the host computer.

[0014] Furthermore, the wire saw cutting module includes a main drive motor, a tension sensor, a tension swing arm, and a tension servo motor. The control module identifies the deviation between the actual feedback value collected by the tension sensor and the set value of the wire tension, and outputs a correction torque to the tension servo motor to drive the tension swing arm to deflect, so as to compensate for tension fluctuations and restore a constant tension state. When the actual feedback value is lower than the abnormal wire breakage threshold and the duration exceeds the preset safety time window, the control module executes the wire breakage safety protection logic, cuts off the current output of the tension servo motor, and shuts off the main drive motor.

[0015] Furthermore, the pulse power supply includes multiple inverter sub-modules connected in parallel. The control module is equipped with a programmable delay compensation unit to identify the actual time delay deviation of the gate drive signal sent to each inverter sub-module, and to perform time phase calibration through the programmable delay compensation unit to compensate for the actual time delay deviation to obtain the total transient superimposed current of synchronous superposition. When any actual time delay deviation exceeds the maximum allowable time delay threshold, the control module immediately stops sending the synchronous trigger signal and shuts down all the inverter sub-modules.

[0016] Furthermore, the control module identifies the average discharge voltage of the processing gap. When the average discharge voltage is lower than the short-circuit precursor threshold, it issues a deceleration command to the wire saw cutting module. When the average discharge voltage is higher than the open-circuit precursor threshold, it increases the feed rate of the wire saw cutting module.

[0017] Furthermore, the control module internally sets up a cutting attenuation logic and a dynamic calculation relationship for the feed rate. During the end cutting stage of the diamond wire saw cutting the hard and brittle substrate, the control module dynamically calculates the feed rate attenuation coefficient based on the current remaining cutting depth, and gradually reduces the feed rate of the wire saw cutting module according to the dynamic calculation relationship for the feed rate, thereby unloading the transient mechanical impact force acting on the bottom of the substrate.

[0018] Furthermore, the control system also includes a frequency-to-voltage conversion circuit and a multi-channel synchronous analog-to-digital converter. The wire saw cutting module includes an encoder installed at the tail end of the multi-wire guide wheel drive shaft. The frequency-to-voltage conversion circuit converts the high-frequency pulse sequence output by the encoder into an analog voltage signal, which is then converted into a digital input to the control module by the multi-channel synchronous analog-to-digital converter. If the analog voltage signal remains below the wire breakage reference threshold for a period of time when the main drive motor receives the running command and the wire feeding speed set by the process exceeds the preset activation boundary, and the duration exceeds the set safety judgment window, the control module determines that the encoder or the frequency-to-voltage conversion circuit has failed, cuts off the closed-loop adjustment branch, forcibly locks the control parameters to a safe preset constant template, and outputs a channel abnormality code to the external terminal.

[0019] Furthermore, the control module includes a trigger threshold and a recovery threshold. When the average gap voltage is detected to have fallen below the trigger threshold and the proportion of short-circuit pulses exceeds the abnormal proportion boundary, the control module reduces the discharge pulse width command value and increases the wire feed speed command value. When the average gap voltage rises back above the recovery threshold after adjustment, the control module determines that the gap has recovered, slightly increases the discharge pulse width command value and slightly decreases the wire feed speed command value.

[0020] Furthermore, when the discharge pulse width command value reaches the minimum sustaining pulse width lower limit, or the wire feed speed command value reaches the maximum safe wire feed speed, and the abnormal state duration exceeds the set limit tolerance window, the control module executes the limit safety back-off logic, forces the pulse power supply to shut down, and sends a micro-reverse push command to the feed axis of the wire saw cutting module, so that the diamond wire saw is removed from the interference region.

[0021] Furthermore, the wire saw cutting module includes a feed axis; The control module continuously identifies the current coordinates of the feed axis and the average discharge voltage of the machining gap; When the current coordinates exceed the pre-cut thickness threshold and the average discharge voltage shows an upward trend, the control module sends a deceleration command to the wire saw cutting module and a pulse width reduction command to the pulse power supply to restore cutting stability. When the current coordinate reaches the total cutting thickness and the average discharge voltage approaches the rated open-circuit voltage of the pulse power supply, or when the current coordinate exceeds the set overtravel limit position and remains at that position for more than the set safe waiting time, the control module determines that the cutting task is completed and cuts off each subsystem in a fixed hard sequence.

[0022] The diamond electrical discharge machining (EDM) composite cutting control system for precision hard and brittle substrates provided by this invention has the following beneficial effects: This invention achieves simultaneous composite cutting of high-frequency pulse discharge and mechanical grinding by coordinating a control module with a pulse power supply and a wire saw cutting module. The system utilizes high-frequency pulse discharge to locally thermally erode and soften the surface of hard and brittle substrates, reducing cutting resistance during mechanical feeding. Simultaneously, the wire saw's feed motion synchronously scrapes away the altered layer and slag generated by the discharge. This method physically accommodates the discharge gap and mechanical contact interface, effectively reducing the probability of wire breakage during processing and improving the smoothness of the sliced ​​surface.

[0023] This invention establishes a state monitoring mechanism based on discharge gap voltage and short-circuit pulse ratio, improving machining continuity under complex conditions. When abnormal conditions such as deteriorated chip removal in deep gaps are detected, the control module simultaneously reduces the discharge pulse width and increases the wire feed speed, suppressing short circuits by reducing heat input and enhancing fluid chip removal. Combined with a pre-defined limit safety retreat judgment and micro-scale reverse logic, the system can autonomously restore normal machining gaps under severe chip removal conditions, avoiding machining interruptions caused by conventional single-stop operations.

[0024] This invention incorporates a thickness adjustment mechanism with displacement closed-loop and mechanical locking, ensuring the dimensional control accuracy and operational stability of the entire equipment. The control module performs micro-feed cyclic compensation by comparing the physical target distance with the actual coordinates of the displacement sensor, and locks the guide mechanism using a guide rail brake after reaching the target coordinates. Combined with tension closed-loop compensation logic and hardware failure prevention judgment of the underlying frequency-voltage conversion circuit, malfunctions caused by false alarms from a single sensor are eliminated, improving the mechanical accuracy of the system under long-term continuous cutting conditions. Attached Figure Description

[0025] Figure 1 This is a perspective view of the present invention; Figure 2 This is a schematic diagram of the system architecture of the present invention; Figure 3 This is a schematic diagram of the method flow of the present invention; Figure 4 This is a schematic diagram of the parallel power supply topology and spatiotemporal coupling composite cutting principle of the present invention; Figure 5 This is a closed-loop adaptive control logic diagram of pulse width and wire feed speed based on state data anomaly determination of the present invention. Figure 6 This is a dynamic decay curve of the feed rate during the cutting stage of the present invention. Figure 7 This is an iterative response curve of the control parameters under abnormal chip removal conditions according to the present invention; Figure 8 This is a comparison chart of the core quality indicators of slice processing according to the present invention; Figure 9 This is a comparison chart of the processing efficiency and end chipping rate of the present invention.

[0026] The components include: 1. Control module; 2. Wire saw cutting module; 3. Pulse power supply; 4. Thickness adjustment mechanism. Detailed Implementation

[0027] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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.

[0028] See Figure 1 and Figure 2 The present invention provides a diamond electrical discharge composite cutting control system for precision hard and brittle substrates, comprising: a control module 1, a wire saw cutting module 2, a pulse power supply 3, and a thickness adjustment mechanism 4.

[0029] Control module 1 is electrically connected to wire saw cutting module 2, pulse power supply 3, and thickness adjustment mechanism 4. Control module 1 receives processing task parameters, runs a process optimization algorithm to generate composite cutting process parameters, and sends control commands to each module in the system.

[0030] The wire saw cutting module 2 drives the diamond wire saw to feed wire, providing mechanical grinding force for the processing. Specifically, the wire saw cutting module 2 has a guide wheel structure inside, which includes multi-wire guide wheels for guiding the diamond wire saw and wire guide sliders that support the multi-wire guide wheels. The physical span between the multi-wire guide wheels determines the working wire spacing of the diamond wire saw to form the cutting mesh.

[0031] The output of pulse power supply 3 is connected between the diamond wire saw and the hard, brittle substrate. Pulse power supply 3 provides electrical discharge energy to achieve thermoelectric ablation processing.

[0032] The thickness adjustment mechanism 4 is connected to the guide wheel structure of the wire saw cutting module 2. Specifically, the thickness adjustment mechanism 4 is connected to the wire guide slider in the guide wheel structure. The thickness adjustment mechanism 4 drives the wire guide slider to produce linear displacement according to the control command, thereby adjusting the spacing of the diamond wire saw in real time.

[0033] When the diamond EDM composite cutting control system for precision hard and brittle substrates is in operation, the wire saw cutting module 2 and the pulse power supply 3 work together on the hard and brittle substrate. The diamond particles on the diamond wire saw mechanically grind the hard and brittle substrate, while the pulse power supply 3 generates high-frequency pulse discharges in the gap between the diamond wire saw and the hard and brittle substrate. The high temperature formed by the discharge channel melts and vaporizes the surface material of the hard and brittle substrate and throws it out. The mechanical grinding action of the diamond wire saw removes the discharge-modified layer in real time.

[0034] like Figure 3 As shown, this embodiment of the invention provides a method for controlling diamond electrical discharge machining (EDM) composite cutting of precision hard and brittle substrates, comprising the following steps: S100, the control module 1 acquires the input initial material characteristic parameters and target processing indicators. The initial material characteristic parameters include the material type, diameter and resistivity of the hard and brittle substrate. The target processing indicators include the target slice thickness, the upper limit of surface roughness, the allowable value of warpage and the upper limit of microcrack depth. S200, the control module 1 calls the built-in big data prediction model and multi-process parameter optimization algorithm, performs matrix operation based on the parameters input in step S100, and calculates the composite cutting parameter combination, which includes pulse power supply voltage, pulse width, wire feed speed and wire tension setting value. S300, the control module 1 converts the target slice thickness into an electrical signal command and transmits it to the thickness adjustment mechanism 4. The thickness adjustment mechanism 4 drives the internal independent variable spacing structure to move, so that the spacing of the multi-wire guide wheels in the wire saw cutting module 2 is adjusted to the target position corresponding to the target slice thickness. S400, control module 1 starts wire saw cutting module 2 and pulse power supply 3. Wire saw cutting module 2 controls diamond wire saw to start high-speed wire feeding. Pulse power supply 3 outputs high-frequency pulse current according to the pulse power supply voltage and pulse width determined in step S200. Diamond wire saw performs electrical discharge grinding composite cutting on hard and brittle substrate. In the S500 composite cutting process, the sensor component collects processing status data in real time and feeds it back to the control module 1. The processing status data includes real-time discharge frequency, average processing current and actual tension of diamond wire saw. S600, the control module 1 performs anomaly detection on the received processing status data. If the processing status data deviates from the set threshold range, the control module 1 adjusts the pulse width of the pulse power supply 3 or the wire feed speed of the wire saw cutting module 2 in real time to restore the processing status data to the normal working range until the overall cutting task is completed. To ensure that subsequent optimization of processing parameters has an accurate data foundation and boundary range, in this embodiment, control module 1 performs the acquisition of pre-processing parameters and transforms the actual process requirements into mathematical boundary conditions for the optimization algorithm. The specific implementation process includes the following steps: S101, the control module 1 receives and records the initial material characteristic parameters of the hard and brittle substrate through a human-machine interface or communication interface. In one specific implementation, the initial material characteristic parameters include the material type, substrate diameter, and resistivity of the hard and brittle substrate. Hard and brittle substrates include silicon carbide, single-crystal silicon, and sapphire semiconductor materials. The material type reflects the material's hardness and fracture toughness. In this embodiment, the control module 1 has a pre-set material property database. By automatically searching this database, the theoretical hardness value of the corresponding material can be retrieved. This value serves as an important reference for the control module 1 in setting the mechanical grinding cutting force threshold. The resistivity parameter is used to characterize the material's conductivity. In the electrical discharge machining (EDM) mechanism, the formation of the discharge channel depends on the combined impedance of the inter-electrode medium and the substrate. Therefore, the control module 1 calculates the equivalent impedance of the initial discharge gap in the processing area based on the resistivity, and then presets the initial no-load breakdown voltage lower limit of the pulse power supply 3. The data parsing process of the communication interface and the specific communication protocol settings of the human-machine interface can be conventionally configured by those skilled in the art based on the actual machine tool control hardware platform. This is well-known technology in the field and will not be elaborated upon here.

[0035] S102, based on the acquired material characteristics, the system needs to further establish the target boundaries for processing. Control module 1 synchronously acquires the target processing indicators. These indicators include the target slice thickness, upper limit of surface roughness, allowable warpage, and upper limit of microcrack depth. The target slice thickness serves as the basic position reference for the closed-loop control of the thickness adjustment mechanism 4, affecting the geometric dimensions of the slice output. For precision processing scenarios such as semiconductor silicon wafers, this target slice thickness is set between 0.2mm and 1mm. The upper limit of surface roughness and the allowable warpage serve as boundary constraints on processing quality, limiting the parameter search space of the subsequent process parameter optimization algorithm. Control module 1 converts the acquired upper limit of surface roughness, allowable warpage, and upper limit of microcrack depth into floating-point numerical variables readable by the internal control program, used for result convergence determination during algorithm iteration.

[0036] S103, as a preferred method, control module 1 aggregates the initial material characteristic parameters and the target processing indicators to construct a multi-dimensional constraint set. Control module 1 establishes a set of processing state constraint space vectors in memory. The mathematical expression of the multi-dimensional constraint set is as follows: ; In the formula, Represents a column vector of multidimensional constraint sets; Indicates the diameter of the substrate; Indicates the resistivity of the substrate; Indicates the thickness of the target slice; This indicates the upper limit of surface roughness. Indicates the allowable value for warpage; superscript This represents the transpose operation of a vector.

[0037] Since the parameters in the aforementioned vectors have different dimensions and orders of magnitude, to avoid algorithmic dead zones in matrix operations, control module 1 normalizes the column vectors using the range transformation method after vector assembly, mapping their internal values ​​to the [0,1] interval. The processed multidimensional constraint set column vectors are then passed to the subsequent process optimization algorithm module as fixed preconditions for algorithm optimization. This constraint construction process helps ensure that the combination of composite cutting process parameters calculated and output by control module 1 will not cause overload of pulse power supply 3 at the actual physical execution level, and also helps guide the final cut product data to fall within the quality range defined by the target processing index.

[0038] In this embodiment, control module 1 invokes a built-in big data prediction model, specifically by constructing an RBF neural network to establish a nonlinear mapping relationship between composite cutting process parameters and processing quality indicators. The specific construction and operation of this big data prediction model includes the following steps: S201, Control Module 1 constructs the historical sample dataset for the prediction model and sets the network input and output nodes. Based on the actual cutting scenario, the key control parameters issued by Control Module 1 are used as the model's input features. Specifically, the input data dimension is set to 4, covering the pulse power supply voltage and pulse width of the pulse power supply 3, and the wire feed speed and wire tension settings of the wire saw cutting module 2. Correspondingly, the model output data dimension is set to 3, representing the physical state after slicing, including surface roughness, warpage, and subsurface microcrack depth. The sample data and corresponding labels used for model training originate from the historical database accumulated from previous electrical discharge composite cutting experiments on semiconductor hard and brittle substrates. Due to the differences in the numerical span of parameters of different dimensions (e.g., pulse voltage of tens of volts and surface roughness at the micrometer level), to eliminate the numerical calculation bias caused by data of different dimensions and orders of magnitude on network weight updates, Control Module 1 preprocesses the historical sample dataset using a max-min normalization algorithm before feeding the data into the model, uniformly mapping all input parameters and output label values ​​to the [0,1] interval.

[0039] S202, Control Module 1 establishes the internal hierarchical structure and forward data flow of the RBF neural network. Since the EDM (Electrical Discharge Machining) process is a highly nonlinear multi-physics coupled system, conventional linear fitting is insufficient to accurately express its inherent laws. As a preferred approach, this model adopts a three-layer feedforward topology: an input layer, a hidden layer, and an output layer. The input layer directly receives the normalized four-dimensional process parameter feature vector and passes it to the hidden layer. The hidden layer uses a Gaussian radial basis function as a nonlinear activation function to perform spatial transformation on the input signal. The physical meaning of this operation is to quantify the spatial similarity between the current processing condition and historical typical conditions (i.e., cluster centers). The hidden layer... The formula for calculating the output response of each node is as follows: ; In the formula, Indicates the hidden layer number 1 The output response value of each node; This represents an input feature vector containing four process parameters; Indicates the first The center vector of each hidden layer node; Indicates the first The basis width parameter of the Gaussian function of each hidden layer node determines the radial range of the basis function around the center point; The Euclidean distance between the input vector and its corresponding center vector is represented. The data flow proceeds to the output layer after nonlinear mapping in the hidden layer. The output layer uses a linear activation function, and the final quality index prediction result is calculated by weighted summation of the output response values ​​from the hidden layer. The output layer's... The formula for calculating the predicted value of each node is as follows: ; In the formula, This represents the predicted values ​​for the surface roughness, warpage, or microcrack depth of the output layer. This represents the total number of hidden layer nodes. This total number of nodes can be initially set during the model initialization phase based on the empirical formula of the square root of the total number of samples, or dynamically generated through subsequent sample clustering algorithms. Indicates the hidden layer number 1 The node to the output layer The connection weight of each node.

[0040] S203, after clarifying the network topology and forward propagation mechanism, control module 1 performs iterative training of model parameters to establish a regression mapping model for the composite cutting process. The training objective of the prediction model is to minimize the error between the network's output prediction index and the actual historical label values. Control module 1 selects mean squared error as the key loss function in the model training process, and its calculation formula is as follows: ; in, This represents the overall mean square error loss value of the system. This represents the total number of samples participating in the training. Indicates the first The sample at the th Network prediction values ​​at each output node Indicates the first The sample at the th The true historical label value corresponding to each output node. In the specific training steps, the algorithm inside control module 1 uses the K-means clustering algorithm to perform cluster analysis on the sample input features to determine the center vector of each hidden layer node. With base width parameter After establishing the center point, the connection weight matrix between the hidden layer and the output layer is solved using the least squares method based on the principle of minimizing the mean square error. The network parameters are continuously updated by inputting historical samples until the mean squared error loss value is reached. The model training is completed when the threshold is lowered to below the preset convergence threshold.

[0041] In the scenario of precision machining parameter prediction, the convergence threshold is set to 10. -4 Up to 10 -5Order of magnitude. To ensure the prediction results have direct process evaluation value, after obtaining the predicted values ​​in the [0,1] interval calculated by the output layer, control module 1 performs an inverse normalization operation corresponding to the preprocessing step, restoring the values ​​to physical state values ​​with actual physical dimensions. For the specific mathematical derivation of the K-means clustering algorithm and the least squares method, as well as the matrix iteration operation process at the code level, those skilled in the art can refer to the standard machine learning algorithm framework for conventional programming deployment. The specific operation rules are well-known in the field and will not be elaborated here. The trained RBF neural network model data structure is stored in the storage unit of control module 1, used to receive the set process parameter combinations in real time before actual processing and quickly output the corresponding processing quality prediction and evaluation results. In this embodiment, because the solution space of the composite cutting process parameters exhibits high nonlinearity and contains multiple local optima, the conventional gradient descent method is prone to stagnation. Therefore, control module 1, based on the constructed RBF prediction model, uses the firefly algorithm to solve for the optimal composite cutting parameters in the multidimensional constraint space and converts them into actual low-level control commands. The specific implementation process includes the following steps: S204, based on the previously established multi-dimensional constraint set, control module 1 calls the multi-process parameter optimization algorithm to initialize the optimization population in the parameter solution space to generate candidate solution vectors for the composite cutting process parameters. As a preferred method, the multi-process parameter optimization algorithm specifically adopts the firefly algorithm, defining each individual in the firefly population as a candidate solution vector for the composite cutting process parameters, denoted as... In the formula, Indicates the first A candidate solution vector; Indicates the pulse power supply voltage; Indicates the pulse width; Indicates the wire feed speed; This represents the setpoint for line tension. To evaluate the merits of each candidate solution, control module 1 performs the same max-min normalization preprocessing on the candidate solution vector as during model training. The processed candidate solution vector is then input into the trained RBF neural network to obtain the corresponding predicted values ​​for surface roughness and warpage. Based on this, control module 1 constructs a fitness function aimed at optimizing processing quality, transforming the network's predicted metrics into brightness evaluation values ​​(i.e., absolute brightness values) that characterize the individual's merits. The specific fitness function expression can be configured as follows: ); In the formula, For individuals The absolute brightness value, and These are the predicted roughness and warpage values ​​after inverse normalization; and These are the weighting coefficients set. The purpose of constructing this fitness function is to ensure that the smaller the predicted roughness and warpage, the higher the corresponding absolute brightness value. Simultaneously, if the model's predicted value exceeds the upper limit of surface roughness or the allowable value of warpage set in the multidimensional constraint set, the internal program of control module 1 will apply a penalty coefficient to the fitness of that candidate solution, for example, by directly forcing its absolute brightness value to a very small constant (such as 10). -6 This eliminates invalid solution intervals that do not meet the actual engineering requirements.

[0042] S205, after clarifying the evaluation criteria for candidate solutions, control module 1 performs iterative optimization in the parameter space. The core optimization mechanism of the firefly algorithm is based on position updates according to the relative brightness and spatial distance between individuals. Control module 1 calculates the position of any two individuals... and Euclidean distance between Because light intensity signals attenuate with increasing spatial distance, individual... For individuals The formula for calculating attractiveness is as follows: ; In the formula, Indicates dynamic attraction; This represents the maximum attraction when the distance is zero, and is set to 1. The light absorption coefficient, used to control the rate of decrease in attractiveness with distance, typically takes values ​​between 0.1 and 10 in the normalized parameter space. When an individual... The absolute brightness is greater than that of the individual At that time, individual To individuals The position is moved in the direction of , and the position update formula is as follows: ; In the formula, and Each represents an individual In the Next and first The position vector in the next iteration; The step size factor controls the intensity of random perturbations, aiming to balance the algorithm's early global exploration with later local convergence. Set as the dynamic decay value within the range of [0.01, 0.2]; This represents a random number that follows a uniform distribution in the [0,1] region. This update formula not only utilizes the direction guidance of high-quality solutions in the current population but also introduces a random perturbation term to avoid the algorithm getting trapped in a local optimum dead zone. Regarding the initial hyperparameter calibration process for the maximum attraction, light absorption coefficient, and step size factor in the algorithm, those skilled in the art can perform conventional trial and error and empirical value selection based on the dimensional characteristics of the specific parameter space. Hyperparameter tuning is a well-known technique in this field and will not be elaborated upon here.

[0043] S206, after the set maximum number of iterations or after meeting the convergence condition, control module 1 extracts the globally optimal individual in the population with the highest fitness value (i.e., the highest brightness evaluation value). The coordinate position of this individual is the theoretically optimal combination of composite cutting parameters. In order to match the continuous theoretical calculation results with the underlying hardware execution mechanism of the machine tool, control module 1 discretizes and calibrates the obtained optimal parameter combination. Based on the hardware analog-to-digital conversion resolution of the servo driver and power module, it rounds the continuous variables into digital instructions that meet the accuracy of the hardware interface, thereby converting them into a digital control matrix that can be directly read by the servo system and the power system. Control module 1 sends the pulse power supply voltage and pulse width command electrical signals in the control matrix to the pulse power supply 3 according to the preset communication refresh cycle, and simultaneously sends the wire feed speed and wire tension setpoints to the wire saw cutting module 2.

[0044] In this embodiment, the control module 1, based on the pre-determined target processing indicators, converts them into position commands executable by the underlying hardware, and achieves precise adjustment of the cutting line spacing through a closed-loop servo system. The specific implementation process includes the following steps: S301, Control Module 1 acquires the set target slice thickness and maps it to the physical target spacing of the wire saw guide mechanism. In actual EDM (Electrical Discharge Machining) operations, the physical spacing between the cutting wires is not equivalent to the final slice thickness. Due to the material removal mechanism, the actual set spacing must compensate for the physical dimensions of the cutting wires and the cutting kerf caused by the discharge erosion effect. As a preferred method, Control Module 1 internally calls a spacing compensation algorithm to convert the thickness requirement into a spatial coordinate constraint, the mathematical expression of which is as follows: ; In the formula, Indicates the physical target spacing of the guidewire mechanism; Indicates the thickness of the target slice; Indicates the standard wire diameter of the cutting line; This refers to the single-sided equivalent discharge gap during the electrical discharge machining process. In the precision hard and brittle substrate slicing process, the standard wire diameter is determined based on the specifications of the selected abrasive-coated metal wire, and the value range is mostly between 0.1 mm and 0.25 mm. The single-sided equivalent discharge gap is positively correlated with the output energy of the pulse power supply. In this embodiment, the control module 1 can preset it to a constant between 10 micrometers and 50 micrometers based on the currently set empirical value of the pulse voltage, or obtain the gap value that dynamically matches the current process parameters by looking up a table.

[0045] S302, after establishing the physical target spacing, in order to drive the machine tool actuator, control module 1 needs to convert this continuous spatial geometry into discrete electrical control signals. Specifically, control module 1 converts the spacing parameter into the underlying drive pulse sequence of the thickness adjustment mechanism 4. In this embodiment, the thickness adjustment mechanism 4 specifically adopts an independent variable spacing structure, which includes a servo motor and a ball screw coaxially connected to the servo motor rotor. The nut seat on the screw is fixedly connected to the guide wire slider that carries the multi-wire guide wheel. Control module 1 calculates the target number of pulses that the servo motor needs to execute by analyzing the reduction ratio and lead parameters of the mechanical transmission mechanism. The conversion formula for the position command is as follows: ; In the formula, This indicates the target number of pulses issued by control module 1; This indicates the current initial spacing position of the independent variable spacing structure. Before each cutting operation, the reference coordinates are established by either homing the system or reading data from the absolute encoder. Indicates the physical target spacing of the guidewire mechanism; This indicates the rated pulse resolution per revolution of the servo motor encoder; This represents the fixed lead of the ball screw. Since the pitch adjustment process involves increasing or decreasing the pitch, the algebraic calculation yields... There is a distinction between positive and negative values. To avoid a logic dead zone in the underlying pulse generator that prevents it from recognizing negative pulses, control module 1 performs a check before issuing commands. The process involves sign extraction and absolute value taking. Specifically, the sign bit is extracted and converted into a direction control signal for the motor's forward and reverse rotation, while the absolute value is... The number of driving pulses as the actual output.

[0046] S303, control module 1 initiates closed-loop adaptive drive control based on the generated control commands. During operation, control module 1 outputs a target pulse number to the servo driver of the thickness adjustment mechanism 4. The servo motor receives a high-frequency pulse sequence and a corresponding directional level signal. After receiving the drive current, it drives the ball screw to rotate, converting the rotational motion into linear displacement of the guide wire slider in the independent variable-pitch structure, thereby changing the span between adjacent cutting lines. To suppress backlash errors in the mechanical transmission chain and thermal drift caused by long-term operation, an absolute encoder at the tail end of the servo motor collects the actual displacement coordinates of the guide wire slider in real time and sends them back to control module 1 as feedback signals. Control module 1 compares the deviation between the target pulse count and the actual feedback position, dynamically adjusting the frequency and duty cycle of subsequent output pulses to ensure the guide wire slider smoothly and accurately stops at the working coordinates corresponding to the physical target spacing.

[0047] This closed-loop drive mechanism helps to resist the interference of external mechanical vibration on the thickness adjustment accuracy. For the specific tuning steps of the servo motor PID closed-loop control parameters and the digital filtering processing of the position feedback signal, those skilled in the art can perform conventional programming and debugging according to the hardware control manual of the selected driver. The specific implementation methods are well-known in the field and will not be elaborated here.

[0048] In this embodiment, after the spacing of the multi-line guide wheels is adjusted to the correct position, to address the position drift caused by high-frequency discharge vibration and line tension fluctuations during the composite cutting process, control module 1 executes a closed-loop compensation procedure for mechanical rigid locking and zero-position deviation to ensure the physical stability of the slice thickness. The specific implementation process of this stage includes the following steps: S304, after detecting that the servo motor of the thickness adjustment mechanism 4 has reached the target pulse position and remained stable, the control module 1 sends a locking trigger command to the clamping execution unit in the independent variable spacing structure. As a preferred method, the clamping execution unit uses a pneumatic-hydraulic pressurized guide rail brake. The control module 1 outputs a high-level signal to open the corresponding electromagnetic reversing valve, using high-pressure fluid to drive the piston inside the brake, causing the friction plate to rigidly lock the guide mechanism carrying the multi-wire guide wheel onto the linear guide rail. This mechanical locking operation replaces the position-holding method that relies solely on the electromagnetic torque of the servo motor. The system's mechanical self-locking not only reduces drive energy consumption during long-term operation but also helps to counteract axial movement generated during subsequent high-speed wire cutting and cuts off the path of high-frequency mechanical resonance transmitted backward along the transmission screw.

[0049] S305, during the application of mechanical clamping force, the nonlinear extrusion deformation of the friction plates and the release of contact stress between mechanical components will cause a slight axial displacement of the multi-line guide wheel, resulting in its actual physical coordinates deviating from the set value. Simultaneously, to reduce the cumulative error caused by the manufacturing and thermal deformation of the long-stroke transmission screw, after the reversing valve has built up pressure, control module 1 directly reads the current actual physical distance coordinates of the multi-line guide wheel using a high-precision displacement sensor (e.g., a linear grating ruler) installed on the side of the linear guide rail. This fully closed-loop measurement method skips the transmission chain, avoiding the blind spot where the motor encoder cannot detect the deformation of the mechanical end. Control module 1 calculates the zero-position deviation with directional polarity by performing a difference calculation between the actual physical distance coordinates fed back by the displacement sensor and the set physical target distance, denoted as... .

[0050] S306, after extracting the zero-position deviation, control module 1 obtains the absolute value of the deviation and compares it with a preset system tolerance threshold to determine whether to trigger closed-loop compensation logic. In this case, the tolerance threshold is configured based on 5% to 10% of the target slice thickness tolerance band. When the current locking status is less than or equal to the tolerance threshold, the control module 1 determines that the current locking status is qualified, confirms that the spacing of the multi-line guide wheels has been established, and enters the standby cutting process.

[0051] when When the value exceeds the tolerance threshold, it indicates that the locking displacement is out of tolerance or the cumulative error is too large, and control module 1 activates the adaptive compensation mechanism. Control module 1 sends an unloading command to the electromagnetic reversing valve to release the pneumatic-hydraulic pressurized guide rail brake and release the friction lock state. Since the underlying driver can only receive positive pulse signals, control module 1... The positive and negative polarities generate corresponding motor forward and reverse direction level signals, and the absolute number of pulses to be compensated is calculated independently. To overcome the static friction of the guide rail and the elastic hysteresis of the mechanical transmission chain during system restart, the formula for calculating the number of compensation pulses is as follows: ; In the formula, This indicates the absolute number of compensation pulses issued by control module 1; This indicates the measured zero-point deviation. This indicates the rated pulse resolution per revolution of the servo motor encoder; Indicates the fixed lead of the ball screw; This represents the nonlinear compensation coefficient introduced to overcome mechanical static friction and hysteresis. This compensation coefficient is preset in the storage unit of control module 1 based on historical displacement compensation calibration data, and its value is between 1.05 and 1.2 to prevent undercompensation. Control module 1 sends a position correction command containing the number of compensation pulses and the corresponding direction level to the servo driver, driving the servo motor to drive the multi-line guide wheel to perform feed compensation, and then triggers the rigid locking program in S304 again. This position reading, deviation comparison, and fine-tuning compensation process will be executed iteratively. To prevent mechanical jamming from causing the program to enter an infinite loop, control module 1 has a built-in maximum number of compensation iterations (e.g., set to 3 to 5 times). If this maximum number is reached... If the signal still fails to converge to the tolerance threshold, control module 1 will forcibly interrupt the compensation process and send a mechanical abnormality alarm signal to the host computer. For the pressure regulation logic of the gas-liquid booster fluid circuit and the high-speed reading protocol of the displacement sensor data bus, those skilled in the art can refer to standard pneumatic control manuals and industrial fieldbus specifications for routine development and configuration. The specific technical implementations are well-known in the field and will not be elaborated upon here.

[0052] In this embodiment, the control module 1 receives and parses the composite cutting process parameters obtained from the aforementioned optimization steps, and sends motion and tension control commands to the wire saw cutting module 2. By establishing a dual closed-loop circuit of speed and tension, the mechanical motion stability of the diamond wire saw during processing is maintained. The specific implementation process includes the following steps: S401, control module 1 sends a speed control command to the main driver of wire saw cutting module 2. Wire saw cutting module 2 responds to the command and activates the main drive motor, which drives the take-up and untake-up rollers to rotate alternately, thereby driving the diamond wire saw to perform high-speed reciprocating wire feeding motion between the multi-wire guide rollers. During the reciprocating wire feeding reversal phase, due to mechanical inertia, directly switching the rotation direction can cause a drastic jump in wire tension, increasing the risk of wire breakage. To avoid this potential hardware damage, control module 1 internally decomposes the speed slope during the reversal phase, specifically using an S-shaped acceleration / deceleration curve control algorithm for a smooth transition. That is, at the reversal moment, the main drive motor controls its operating speed to decrease to zero along an S-shaped curve according to a preset time constant, and during reverse startup, it follows the same S-shaped acceleration curve to increase to the target wire feeding speed. This target wire feeding speed matches the wire feeding speed setpoint optimized and calibrated in the aforementioned steps. As a preferred approach, the reciprocating wire feeding speed is controlled between 5 m / s and 15 m / s to balance material removal efficiency and wire fatigue life.

[0053] In S402, while the diamond wire saw is reciprocating at high speed, a tension sensor installed on the wire saw's transmission path captures the axial dynamic tension of the wire in real time. Due to the high-frequency discharge impact and liquid bursting associated with electrical discharge erosion, the diamond wire saw generates complex micro-vibrations, causing high-frequency pulse interference to interfere with the raw electrical signal acquired by the tension sensor. To obtain physical indicators that accurately reflect the macroscopic tension state of the wire saw, the signal processing unit of control module 1 inputs the analog signal fed back from the tension sensor into a pre-filter for anti-aliasing and peak reduction processing, and then introduces a sliding window averaging filter algorithm for digital smoothing. The sample window size of the sliding window is dynamically configured based on the spindle speed and discharge frequency, set to 10 milliseconds to 50 milliseconds, thereby filtering out transient pseudo-random fluctuations and extracting the true feedback value.

[0054] S403, Control Module 1 compares the smoothed actual feedback value with the previously optimized set tension value in real time, and implements adaptive fine-tuning of the tension by constructing a constant tension maintenance closed-loop system. Control Module 1 internally contains a tension servo adjustment mechanism, which includes a tension lever and a tension servo motor coaxially connected to the tension lever. Based on the deviation between the target tension and the actual tension, Control Module 1 dynamically calculates and outputs the correction torque of the tension servo motor. The formula for calculating the control torque is as follows: ; In the formula, This indicates the corrected torque output by control module 1 to the tension servo motor; This represents the tension feedback value after filtering and smoothing. This represents the line tension setpoint determined through process parameter optimization; This represents the torque proportional gain coefficient; This represents the torque differential gain coefficient; This indicates the processing run time. The formula determines the compensation direction of the servo motor by extracting the positive and negative deviations between the setpoint and the feedback value.

[0055] Torque proportional gain coefficient The amplitude used to respond to tension deviation, under steady-state conditions, has an empirical value ranging from 0.5 to 2.0; torque differential gain coefficient. The damping is used to predict the trend of tension changes and provides damping, with a value ranging from 0.01 to 0.1. To prevent the amplifier from overloading and burning out due to the corrective torque exceeding the physical limit of the motor in the event of sudden severe jamming, a hardware protection limiting mechanism is set in control module 1. The absolute value is limited to within 120% of the rated torque. However, simple torque limiting cannot cope with the risk of system collapse caused by extreme operating conditions such as wire breakage. In this embodiment, if the diamond wire saw suddenly breaks, the tension feedback value... The tension will momentarily approach zero. Driven by the huge algebraic deviation, the constant tension closed-loop circuit will force the tension servo motor to operate continuously at the maximum torque limit, which can easily cause the tension swing arm to violently impact the hard limit and deform. To address this, a wire breakage safety protection logic has been added to the control module 1, which monitors the tension feedback value in real time. Is the wire tension below the set abnormal wire breakage threshold, which is usually configured as the wire tension setting? 20%; when detected When the wire breaks below the threshold and the duration exceeds the preset safety time window (set to 30 to 50 milliseconds), control module 1 determines that the wire has broken, immediately cuts off the current output of the tension servo motor and shuts down the main drive motor, thereby eliminating the dead zone of closed-loop control under abnormal operating conditions. Under normal processing conditions, the tension servo motor responds with corrective torque and drives the tension swing arm to make a small-angle adaptive deflection, compensating for tension fluctuations caused by thermal elongation or EDM wear of the wire saw, which helps to suppress axial movement during the slicing of hard and brittle substrates. For the speed synchronization decoupling control logic between the main drive motor and the tension servo motor, as well as the specific self-tuning process of the closed-loop parameters, those skilled in the art can configure the conventional parameters according to the official application manual of the selected motion control card. The specific debugging methods are well-known in the field and will not be elaborated here.

[0056] See Figure 4 In this embodiment, the control module 1 coordinates the power output and mechanical motion, and synchronously triggers the operation through the parallel topology within the pulse power supply 3. This establishes a coupled removal mechanism in the spatiotemporal dimension, combining high-frequency electrical spark discharge erosion with diamond particle mechanical grinding, to suppress subsurface microcracks and warpage in the hard and brittle substrate. The specific implementation process of this stage includes the following steps: S404, Control Module 1 sends a synchronous trigger signal to Pulse Power Supply 3, driving its internal multi-parallel topology to collaboratively output high-frequency pulse current. To obtain the required peak current within a set pulse width, Pulse Power Supply 3 consists of multiple identical inverter sub-modules connected in parallel. Each inverter sub-module uses silicon carbide field-effect transistors (MOSFETs) as high-frequency switching elements to achieve rapid turn-on and turn-off responses. In actual operation, if the triggering time difference between the drive signals is caused by differences in hardware circuit trace lengths or component parameter drift, it can easily lead to current concentration effects, causing the first-conducting switching element to be damaged due to overcurrent. Therefore, Control Module 1 is internally configured with a programmable delay compensation unit to perform nanosecond-level time phase calibration on the gate drive signals sent to each inverter sub-module. Due to differences in hardware traces and device response, the output of each sub-module has an inherent delay, and the total transient superposition current after merging satisfies the standard time-domain superposition mathematical relationship: ; In the formula, Indicates that pulse power supply 3 is in The total transient superposition current output at any given moment; Indicates the physical number of parallel submodules; This represents the transient output current function of a single submodule under ideal, delay-free conditions. Indicates the first The time delay deviation when the inverter submodule is actually turned on.

[0057] To ensure the safe operation of the parallel system and obtain a steep pulse rising edge, control module 1 adjusts the logic wiring and internal registers to control the pulse rising edge of each branch. Dynamic compensation is performed to control the maximum time delay deviation of each submodule within 5 nanoseconds, thereby ensuring that the current waveforms of each branch are highly coincident on the time axis, achieving the designed peak current output. This multi-channel synchronous superposition method enables the pulse power supply 3 to stably output pulse current with a pulse width between 100ns and 500ns, providing a high-energy-density transient heat source for subsequent material removal. To prevent the time delay deviation from exceeding the adaptive adjustment range of the programmable delay compensation unit due to severe hardware aging or occasional failures, the control module 1 is equipped with a safety current interruption protection. When the actual time delay deviation of a certain drive signal is detected to exceed the preset maximum allowable time delay threshold (e.g., set to 10 to 20 nanoseconds), the control module 1 immediately stops sending the synchronization trigger signal and shuts down all inverter submodules, thereby avoiding the risk of local overcurrent burnout and eliminating the control dead zone caused by synchronization abnormalities. For the heat dissipation structure design of the pulse power supply 3 and the electromagnetic compatibility layout of the high-frequency drive chip, those skilled in the art can refer to the switching power supply design manual and high-frequency circuit board wiring specifications for conventional hardware development. The specific implementation methods are well-known technologies in this field and will not be elaborated here.

[0058] S405, while maintaining a stable output of high-frequency pulsed current, control module 1 triggers the synergistic effect of discharge thermoelectric erosion and mechanical grinding in the time dimension through timing regulation. Within a single pulse cycle, the synchronously triggered high-frequency pulsed current induces dielectric breakdown in the processing gap between the diamond wire saw and the hard and brittle substrate, forming a discharge plasma channel. The high temperature generated by the plasma channel causes the surface material of the hard and brittle substrate to melt and vaporize, forming micro-discharge pits, and leaving a heat-affected layer around the pits that has softened due to the high temperature. Immediately following this discharge period, the wire saw cutting module 2 drives the diamond wire saw to perform high-speed axial movement, so that the diamond particles embedded on the selected wire surface immediately perform mechanical scraping and grinding to remove the softened heat-affected layer. Since the thermoelectric erosion effect reduces the surface hardness and yield strength of the material in advance, the mechanical cutting resistance experienced by the diamond particles in the subsequent scraping is reduced. This helps to suppress the propagation of brittle fracture cracks in the material to the subsurface depth, thereby delaying the stress accumulation trend of microcracks and warpage in the micro-time sequence.

[0059] S406, based on achieving time-axis coordination, control module 1 further maintains an adaptive balance between the discharge gap and the mechanical contact interface in the spatial dimension. In the geometric distribution of the processing space, the actual discharge gap and the exposed height of the diamond particles are mutually constrained. If the feed speed applied by the wire saw cutting module 2 is too fast, the wire on the multi-wire guide wheel will excessively compress the processing area, causing direct contact between the diamond particle matrix and the hard, brittle substrate, thus triggering a continuous short circuit and extinguishing the discharge plasma, leading to the failure of the composite cutting mechanism. To avoid processing failure caused by spatial overlap, control module 1 monitors the average discharge voltage of the processing gap in real time. Excessive feed reduction can lead to an excessively large machining gap, causing an open circuit and reducing the efficiency of electro-erosion and grinding. During this stage, control module 1 performs bidirectional closed-loop adjustment of the machining state in the spatial dimension: when the average discharge voltage is lower than the preset short-circuit precursor threshold (set to 5V to 15V), a spatial contact overload is determined. Control module 1 then sends a deceleration command to the wire saw cutting module 2 to reduce the feed rate, or increases the pulse stop time to improve the chip removal and washing effect of the working fluid, thereby dynamically restoring a reasonable spark discharge gap. Conversely, when the average discharge voltage is higher than the preset open-circuit precursor threshold (set to 80% to 90% of the rated open-circuit voltage of the pulse power supply), the gap is determined to be too large. Control module 1 then increases the feed rate of the wire saw cutting module 2 by a preset step size, causing the machining gap to move closer to the optimal spark discharge range. This bidirectional closed-loop dynamic adjustment eliminates the machining dead zone caused by the blind spot of unidirectional adjustment. By relying on coordinated control in both spatiotemporal dimensions, the system combines continuous mechanical grinding with discrete electro-erosion, achieving a relative balance between material removal rate and surface finish. In this embodiment, the control module 1 combines hardware-level signal conditioning circuitry with underlying synchronous sampling algorithms to capture the electrical and mechanical motion states in parallel during the composite cutting process. This helps alleviate feedback distortion caused by time-scale mismatch of multi-source physical field signals and high-frequency electromagnetic noise interference. The specific implementation process of this stage includes the following steps: S501, during the parallel execution of composite cutting, multiple sensors acquire heterogeneous physical state signals of the processing area online. Specifically, a sampling sensor installed at the output of the pulse power supply 3 captures the discharge voltage waveform and average processing current between the diamond wire saw and the hard, brittle substrate in real time. This data is then converted into an average gap voltage and short-circuit pulse ratio that can quantify the processing gap state through a subsequent high-frequency feature extraction algorithm. As a preferred embodiment, this sampling sensor includes a high-frequency voltage divider resistor network and a Hall current transformer. Simultaneously, a tension sensor positioned on the wire saw transmission path continuously detects the axial dynamic tension of the diamond wire saw, while an encoder installed at the tail end of the multi-wire guide wheel drive shaft records the current pulse signal in real time, reflecting the actual wire feed speed of the wire saw cutting module 2. Because the processing gap is accompanied by discrete electrical spark discharges and continuous mechanical grinding, the amplitude range, impedance characteristics, and noise floor of the aforementioned original signals differ, making direct input to the microprocessor of the control module 1 difficult.

[0060] In S502, to eliminate high-frequency spike interference and unify signal dimensions, the raw analog signal and pulse sequence acquired by the sensor are shunted and input into a hardware filter circuit and a frequency-to-voltage conversion circuit for parallel conditioning. For discharge voltage waveforms and current signals containing dense spark discharge noise, the signal processing unit inputs them into a pre-amplified passive differential low-pass filter network for anti-aliasing and peak clipping, thereby filtering out megahertz-level radio frequency interference. For the high-frequency pulse sequence output by the encoder, to avoid the microprocessor consuming core computing resources due to frequent responses to external counting interrupts, this pulse signal is directly guided to the hardware frequency-to-voltage conversion circuit. This circuit uses the charge pump charging and discharging principle to linearly convert discrete pulse frequencies into continuous analog voltage signals. Its hardware conversion mechanism satisfies the following mathematical relationship: ; In the formula, This represents the analog voltage signal output by the frequency-to-voltage conversion circuit. This indicates the real-time pulse frequency output by the encoder; This represents the transfer gain coefficient of the frequency-to-voltage conversion circuit.

[0061] In practical engineering applications, the transfer gain coefficient The circuit is calibrated based on a preset maximum wire feed speed and the full-scale voltage of the subsequent analog-to-digital converter (ADC), with values ​​ranging from 0.001V / Hz to 0.005V / Hz. This ensures that the frequency signal can be effectively mapped within the optimal linear range of the hardware sampling. The conditioned multi-channel analog signals are then converted into digital quantities by a multi-channel synchronous analog-to-digital converter. For the register clock configuration of the ADC and the specific RC parameters of the differential amplifier circuit, those skilled in the art can refer to standard analog electronics manuals for conventional hardware development. The specific technical implementation is well-known in the field and will not be elaborated upon here.

[0062] S503, control module 1 receives the converted digital quantity and calls its internal software clock alignment algorithm to perform timing synchronization of multi-dimensional state parameters. Simultaneously, it performs signal integrity checks to avoid control dead zones. Since the response period of electrical signals is in the microsecond range, while the response period of mechanical signals such as tension and speed is in the millisecond range, control module 1 constructs a ring-shaped FIFO buffer, integrates and averages the high-frequency electrical characteristics based on a microsecond-level sampling rate, and performs step interpolation of the mechanical signals on the time axis, thereby assembling a multi-dimensional processing state feature vector aligned at the same timestamp.

[0063] Considering the differences in electrical performance among different types of sensors, and to avoid false alarms and control dead zones caused by the speed signal synchronously approaching zero volts when the equipment stops normally or the wire saw decelerates to zero speed during reciprocating cycles, control module 1 employs differentiated validity verification logic for heterogeneous signals. Specifically, for tension sensors that maintain an initial preload throughout the processing, their output voltage remains within a set positive bias voltage range (e.g., 1V to 5V) under normal operating conditions. If control module 1 detects that the voltage of this channel is consistently lower than a preset wire breakage threshold (preferably set to 0.2V), it determines that the tension sensor has experienced a hardware wire breakage fault. For analog voltage signals reflecting wire feed speed... The control module 1 establishes a command linkage mechanism, which means that the analog voltage signal is activated only when the main drive motor receives the running command and the wire feeding speed set by the process exceeds the preset activation boundary. The lower limit value was compared; if the measured value was within the above-mentioned operational guidance period, If the voltage remains below the disconnection threshold (e.g., 0.2V) for an extended period exceeding the set safety threshold (set to 50 to 100 milliseconds), the encoder or frequency-to-voltage conversion circuit is considered to have failed.

[0064] When any sensor link is determined to meet its corresponding failure condition, control module 1 determines that the current sensor feedback link is abnormal, immediately cuts off the closed-loop adjustment branch, forcibly locks the control parameters to a safe preset constant template, and outputs a channel abnormality code to the external terminal, awaiting manual intervention to troubleshoot the hardware fault. Through the above-mentioned collaborative processing from hardware conditioning to software differential verification, the system provides reliable input features for subsequent process status determination.

[0065] Reference Figure 5 In this embodiment, the control module 1, based on the acquired time-aligned multidimensional feature vector, evaluates the physical stability of the processing gap in real time, and suppresses discharge concentration and chip deterioration by dynamically adjusting the energy input of the pulse power supply 3 and the mechanical scouring capability of the wire saw cutting module 2. The specific implementation process of this stage includes the following steps: S601, the control module 1 periodically extracts the average gap voltage and short-circuit pulse ratio within the current time window from the data buffer, serving as the core input for determining abnormal processing conditions. Under normal composite cutting steady-state conditions, the working fluid can promptly remove electro-erosion products and mechanical debris, resulting in a stable high-frequency spark discharge characteristic in the gap. When debris accumulates or micro-laps occur in localized areas, the average gap voltage drops, while the short-circuit pulse ratio rises sharply. To avoid single-point threshold determination failure due to measurement noise or transient disturbances, which could lead to high-frequency oscillations in control commands, the control module 1 internally constructs a state monitoring model with hysteresis characteristics. Specifically, the control module 1 sets trigger thresholds and recovery thresholds; when the average gap voltage drops below the trigger threshold (e.g., set to 8V) and the short-circuit pulse ratio exceeds the abnormal ratio boundary (e.g., 30%), the system confirms it as a precursor to chip removal deterioration and enters an abnormal control state; during the control process, only when the average gap voltage recovers and crosses the recovery threshold (e.g., set to 12V) does the system determine that the processing state has escaped the deterioration range and prepares to enter the steady-state recovery process. This hysteresis interval setting effectively isolates the state jitter dead zone near the boundary.

[0066] S602, after confirming the entry into an abnormal control state, control module 1 activates a dual-variable coordinated control mechanism for pulse width and wire feed speed. Deterioration in chip removal is usually accompanied by localized heat accumulation, and simply reducing the mechanical feed often results in a delayed response. Therefore, control module 1 sends a pulse width reduction command to pulse power supply 3 to reduce the heat input of a single discharge; simultaneously, it sends an acceleration command to wire saw cutting module 2, utilizing the higher axial speed of the diamond wire saw to drive more fresh cutting fluid into the gap, enhancing the hydrodynamic flushing effect. Control module 1 uses a discrete iterative step-size algorithm to calculate the correction parameters for the current control cycle, as shown in the following formula: ; ; In the formula, Indicates the first The discharge pulse width command value output in each control cycle; This indicates the discharge pulse width command value of the previous control cycle; This indicates a preset fixed step size for pulse width reduction; Indicates the first The wire feed speed command value output in each control cycle; This indicates the wire feed speed command value from the previous control cycle; This represents the proportional gain constant used to adjust the wire feed speed. This indicates the percentage of short-circuit pulses measured during the current control cycle.

[0067] As a preferred method, pulse width reduction uses a fixed step size. The value ranges from 10ns to 20ns, and the proportional gain constant is... The value of needs to be matched and calibrated in conjunction with the torque characteristics of the linear drive motor. Its empirical range is between 0.1 and 1.0 to ensure smooth speed increment transition. Control module 1 will use the calculated value... and The signal is converted into a low-level drive message and synchronously sent to the corresponding actuator via the fieldbus. When the logic judgment system in S601 crosses the recovery threshold and enters the steady-state recovery process, if the control module 1 fails to reset the previously compressed pulse width and increased wire feed speed, the system will continue to operate under a low-efficiency parameter combination, forming an irreversible performance loss dead zone. To address this, the control module 1 stops the abnormal iteration of the above formula during the steady-state recovery phase and instead adopts smooth backtracking logic. It incrementally increases the discharge pulse width command value and incrementally decreases the wire feed speed command value cycle by cycle with a fixed recovery step size until the two parameters are re-aligned with the set values ​​established in the previous process optimization phase, thus forming a complete control closed loop with a beginning and an end.

[0068] In the S603, during closed-loop adaptive control, control module 1 monitors in parallel whether the command value approaches the hardware's safe physical boundary. If the chip removal deterioration is caused by dense impurities within the hard and brittle substrate, conventional pulse width and wire feed speed adjustments cannot reverse the gap deterioration trend. This will cause the algorithm to continuously iterate until the pulse width is compressed to zero or the wire feed speed exceeds the system's maximum allowable mechanical limit, ultimately leading to a control deadlock state where parameters reach their limit and causing a wire break. To cut off this logical dead zone, control module 1 embeds limit safety backoff logic. Control module 1 compares in real time. With the set minimum sustain pulse width lower limit (e.g., 50 ns), and The maximum safe wire feed speed is set (e.g., 20 m / s). When any control parameter is detected to have reached its corresponding boundary and the abnormal state is maintained for a period of time exceeding the set limit tolerance window (1 to 2 seconds), the control module 1 determines that the adaptive control has failed.

[0069] At this point, control module 1 immediately stops the step-size iteration, forces pulse power supply 3 to shut off its output, and sends a small reverse push command (e.g., lift upwards by 0.5mm to 1.0mm in the opposite direction of feed) to the feed axis of wire saw cutting module 2, forcing the diamond wire saw to leave the interference area along its original kerf trajectory. After the physical gap is re-established and the working fluid is fully flushed, the system resumes discharge and feed, thus maintaining the continuity of the overall processing flow while ensuring the physical safety of the equipment. For the packet protocol of the fieldbus message and the overload protection configuration of the servo motor, those skilled in the art can refer to the industrial Ethernet communication specifications for routine adjustments; the specific technical implementation is well-known in the field and will not be elaborated here.

[0070] In this embodiment, the control module 1 tracks the processing progress in real time and performs specific timing control during the cutting stage, thereby reducing the risk of edge chipping at the end of the hard and brittle substrate due to cantilever gravity at the moment of cutting through, and preventing equipment hardware damage caused by disordered shutdowns. The specific implementation process of this stage includes the following steps: S604, the control module 1 continuously reads the current position coordinates of the feed axis of the wire saw cutting module 2 and determines online whether the cutting task has entered the pre-cutting stage based on electrical characteristics. In the end region where the hard and brittle substrate is about to be completely cut through, the processing contact area decreases, and the structural rigidity of the material itself decreases accordingly. If the normal constant feed speed and discharge energy are maintained in this stage, it is easy to cause brittle tearing at the end of the material, which may lead to workpiece chipping or breakage of the diamond wire saw. To avoid this risk of physical damage, the control module 1 adopts a dual verification mechanism of position and voltage to trigger cut-out protection. In specific implementation, when the current coordinate of the feed axis exceeds the set pre-cutting thickness threshold (set to 95% to 98% of the total cutting thickness), the control module 1 simultaneously monitors the average discharge voltage of the processing gap. If the voltage value shows a continuous upward trend with the increase of the feed depth, the system is determined to have officially entered the cut-out stage. At this time, control module 1 synchronously sends a deceleration command to wire saw cutting module 2 and a pulse width reduction command to pulse power supply 3, so that the thermal input energy of a single pulse (such as peak current or discharge pulse width) decreases proportionally to the mechanical feed speed. As a preferred method, the feed speed in the cutting stage is dynamically calculated according to the following mathematical relationship: ; In the formula, This indicates the real-time feed rate command during the cut-out phase. This indicates the feed rate setting value during the normal steady-state cutting phase; Indicates the feed attenuation coefficient; Indicates the current coordinate of the feed axis; This indicates the set threshold thickness before the cutout; This indicates the total cutting thickness of a hard and brittle substrate.

[0071] Feed attenuation coefficient The value is configured between 0.5 and 0.8, which allows the feed rate to smoothly transition to a lower safe cutting level just before cutting through.

[0072] S605, during the execution of the cut-out attenuation logic, control module 1 continuously performs a final state verification upon task completion and executes a strict energy cut-off sequence after confirming complete cut-through. In engineering practice, due to the slight flatness error of the clamping surface of the hard and brittle substrate, relying solely on position coordinates... To determine whether all wires have been cut through, there is a risk of premature shutdown leaving uncut strands. Therefore, control module 1 incorporates an electrical open-circuit characteristic into the task completion criteria. When... achieve When the control module 1 continuously monitors the average discharge voltage for multiple sampling cycles and it approaches the rated open-circuit voltage of the pulse power supply 3 (e.g., reaching more than 98% of the rated open-circuit voltage), the system confirms that the hard and brittle substrate has been completely cut off. However, if there is a false short circuit caused by conductive abrasive debris overlapping inside the processing gap, the average discharge voltage will never reach the open-circuit threshold. If the system waits indefinitely, the wire saw will dry grind in place, increasing the risk of wire breakage. To avoid this shutdown decision dead zone, the control module 1 introduces overtravel redundancy forced completion logic. When the current coordinate of the feed axis... Exceeding the set overtravel limit position (e.g.) With an added redundancy of 1mm to 2mm, and after lingering at this position for more than the set safe waiting time (e.g., 2 to 3 seconds), regardless of whether the average discharge voltage meets the standard, the control module 1 will forcibly determine that the cutting task is completed.

[0073] Upon confirmation of task completion, control module 1 immediately intervenes in the equipment shutdown sequence. To prevent hardware damage caused by timing errors, control module 1 shuts down each subsystem in a fixed, hard-sequence manner. Control module 1 sends a nanosecond-level shutdown signal to pulse power supply 3, forcibly blocking the trigger pulses of all inverter submodules to cut off the heat source input during the gap. Subsequently, the system maintains the working fluid pump and the main drive motor of the wire saw cutting module 2 to continue running for a preset washing delay time (set to 3 to 5 seconds), using the flowing cutting fluid and wire feed motion to remove residual heat and debris from the processing area. Only after the washing delay time ends does control module 1 issue a command to shut down the working fluid pump and control the main drive motor to smoothly decelerate to zero speed according to an S-shaped deceleration curve.

[0074] S606, after all power sources have safely come to a standstill, control module 1 instructs the wire saw cutting module 2 to perform mechanical zeroing and tension unloading actions. To facilitate the unloading and transfer of the sliced ​​material, control module 1 instructs the feed axis to reverse and lift at a set rapid traverse rate until the mechanical origin limit sensor is triggered. During the feed axis retraction, control module 1 instructs the tension closed-loop mechanism to switch to low-pressure holding mode, causing the tension servo motor to output a weak safety torque (10% of the rated torque) that is only enough to prevent the diamond wire saw from degrooving, in order to release the axial elastic stress accumulated in the wire during long-term processing. For the encoder phase-finding algorithm for servo motor origin return and the hardware interlocking circuit design of the emergency stop relay, those skilled in the art can refer to the standard CNC system programming manual and electrical safety specifications for conventional configuration. The specific technical implementation is well-known in the field and will not be elaborated here. At this point, the entire closed-loop control process from multi-dimensional status monitoring to safe completion of the task is completed, and the system returns to the initial standby state and waits for the instruction of the next composite cutting task.

[0075] Specific application examples are as follows: Taking the processing of a 6-inch (150mm diameter) 4H-SiC (silicon carbide) single crystal ingot as an example, the specific formula call and numerical derivation process of the control system of this invention are as follows: 1. Establishing Objective Constraints and Evaluating Predictive Models Control module 1 acquires initial parameters, including substrate diameter. mm, resistivity cm; Set target slice thickness m, upper limit of surface roughness m, upper limit of warpage m. Assemble the multidimensional constraint set vector. .

[0076] In the firefly algorithm In the optimization process, the candidate solution for the process parameters represented by a certain individual is: 65V ns, m / s [N]. Control module 1 inputs the solution into a pre-trained RBF neural network to predict the roughness. m, warpage m. The fitness function is called to calculate its absolute brightness (with weights set). , ): ; After multiple iterations and updates, the system finally locked in the globally optimal solution with the highest brightness and issued it as the underlying control command.

[0077] 2. Target Spacing Mapping and Servo Drive Calculation To ensure that the final slice thickness accurately falls within 350μm, control module 1 needs to calculate the execution pulses of the servo motor.

[0078] Given cutting wire diameter m, the system determines the single-sided equivalent discharge gap under the current voltage (65V) based on experience or by referring to tables. m. Substitute into the spacing compensation formula: ; Assuming the independent variable spacing structure is currently in its initial position m, servo motor resolution 10000P / rev, lead screw m. Substitute into the position instruction conversion formula: ; Control module 1 outputs 80 positive high-frequency pulses to drive the multi-line guide wheel to move precisely by 40μm, completing the physical closed-loop adjustment of the gap.

[0079] 3. Closed-loop adaptive iteration for abnormal states When the cutting process reaches the middle stage, control module 1 monitors the current cycle. The average gap voltage dropped to 7V (below the trigger threshold of 8V), and the proportion of short-circuit pulses increased. (Exceeding the abnormal boundary by 30%), the chip removal is judged to be worse.

[0080] Reference Figure 7 The system triggers an abnormal iteration in the 2nd second, and the system immediately calls the collaborative control formula (with a step size). ns, proportional gain ): Pulse width reduced to: ns-10ns=290ns.

[0081] The wire feed rate is: m / s 12.17 m / s; The system cools down the wire by compressing the pulse width and increases the wire feed speed to enhance chip removal and flushing, so that the gap state can be restored to the normal range within 2 seconds.

[0082] 4. Adaptive feed decay during the cut-out phase Reference Figure 6 When the current coordinate of the feed axis When it reaches 144mm (exceeding the precursor threshold) (95% of the total thickness), entering the cutting stage where the edges are extremely prone to chipping.

[0083] At this time, the normal steady-state feed rate mm / h, attenuation coefficient set to Substitute into the feed dynamic calculation formula: 0.43; As can be seen, the system smoothly reduces the feed rate from 0.5 mm / h to 0.43 mm / h through this mathematical relationship, and further reduces it as it approaches the 150 mm endpoint, thus unloading the mechanical impact force at the moment of cutting through and achieving zero edge chipping.

[0084] Reference Figure 8 and Figure 9 The system of this invention was used to process 6-inch 4H-SiC ingots on a self-developed platform, and a comparative experiment of 100 wafers was conducted with traditional processing methods. The results are shown in the table below: Compared to traditional purely mechanical wire saws, this invention increases cutting efficiency by approximately 3.37 times (up to 1.52 mm / h). At the same time, through predictive models and spacing compensation, the slice thickness variation is strictly controlled within ±3.2 μm, and the average warpage is reduced to 8.6 μm, breaking the traditional trade-off between processing efficiency and surface precision.

[0085] Based on the pioneering cutting attenuation logic and dynamic calculation relationship of feed rate, the system unloads the end mechanical impact force during the cutting stage. In a continuous 100-piece cutting experiment, a 0% bottom edge chipping rate was achieved, improving the yield and material utilization of expensive crystal rods.

[0086] Under harsh conditions of deep trench chip removal, the system's closed-loop coordinated control can instantly mitigate the risk of wire breakage, ensuring stable operation without downtime for extended periods. The system is suitable for the mass production needs of hard and brittle substrates such as silicon carbide and gallium nitride, and has extremely high industrialization economic benefits.

[0087] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A diamond electrical discharge machining (EDM) composite cutting control system for precision hard and brittle substrates, comprising a control module (1), a wire saw cutting module (2), a pulse power supply (3), and a thickness adjustment mechanism (4), characterized in that, The wire saw cutting module (2) includes a diamond wire saw; The thickness adjustment mechanism (4) adjusts the spacing of the diamond wire saw according to the control command issued by the control module (1); The output end of the pulse power supply (3) is connected between the diamond wire saw and the hard and brittle substrate to generate high-frequency pulse discharge to form a discharge channel; The control module (1) controls the wire saw cutting module (2) to drive the diamond wire saw to mechanically grind the hard and brittle substrate, and at the same time controls the pulse power supply (3) to melt and vaporize the surface material of the hard and brittle substrate through the discharge channel and throw it out. The mechanical grinding simultaneously removes the discharge-modified layer.

2. The diamond electrical discharge machining (EDM) composite cutting control system for precision hard and brittle substrates according to claim 1, characterized in that, The control module (1) calls the big data prediction model and the multi-process parameter optimization algorithm, uses the multi-process parameter optimization algorithm to generate candidate solution vectors in the parameter solution space, inputs the candidate solution vectors into the big data prediction model to obtain the predicted values ​​of surface roughness and warpage, and constructs a fitness function to iteratively optimize the candidate solution vectors; After the maximum number of iterations is set or the convergence condition is met, the control module (1) extracts the global optimal individual with the highest brightness evaluation value in the population. The brightness evaluation value is calculated by the fitness function. The composite cutting parameter combination of the global optimal individual is converted into a low-level control command and sent to the pulse power supply (3) and the wire saw cutting module (2).

3. The diamond electrical discharge machining (EDM) composite cutting control system for precision hard and brittle substrates according to claim 1, characterized in that, The thickness adjustment mechanism (4) includes an independent variable pitch structure, which includes a servo motor and a ball screw. The control module (1) identifies the target slice thickness in the processing task parameters, and after compensating the target slice thickness by combining the physical size of the cutting line and the cutting gap generated by the discharge erosion effect, it maps it to the physical target spacing to convert the control command. The servo motor receives the control command to drive the ball screw, which in turn moves the guide wire slider in the independent variable pitch structure. The control module (1) compares the deviation between the target pulse number and the actual feedback position and dynamically adjusts the frequency and duty cycle of the subsequent output pulses so that the guide wire slider stops at the working coordinate corresponding to the physical target pitch.

4. The diamond electrical discharge machining (EDM) composite cutting control system for precision hard and brittle substrates according to claim 3, characterized in that, The independent variable spacing structure also includes a guide mechanism, a linear guide rail, a guide rail brake, and a displacement sensor, wherein the guide rail brake includes a friction plate; After the servo motor reaches the working coordinate, the control module (1) controls the guide rail brake to squeeze the friction plate, thereby locking the guide mechanism onto the linear guide rail.

5. The diamond electrical discharge machining (EDM) composite cutting control system for precision hard and brittle substrates according to claim 4, characterized in that, After entering the locked state, the control module (1) reads the actual physical distance coordinates through the displacement sensor and identifies the zero-position deviation of the distance from the physical target; When the zero-position deviation is greater than the tolerance threshold, the control module (1) releases the locking state and sends a pulse to the servo motor to perform feed compensation, and iterates in a loop until the zero-position deviation converges to the tolerance threshold. If the zero-position deviation still fails to converge to the tolerance threshold after the maximum number of compensation iterations is reached, the control module (1) forcibly interrupts the compensation process and sends a mechanical abnormality alarm signal to the host computer.

6. The diamond electrical discharge machining (EDM) composite cutting control system for precision hard and brittle substrates according to claim 1, characterized in that, The wire saw cutting module (2) includes a main drive motor, a tension sensor, a tension swing arm, and a tension servo motor; The control module (1) identifies the deviation between the actual feedback value collected by the tension sensor and the set value of the line tension, and outputs a correction torque to the tension servo motor to drive the tension swing arm to deflect, so as to compensate for the tension variation and restore the constant tension state. When the actual feedback value is lower than the abnormal disconnection threshold and the duration exceeds the preset safe time window, the control module (1) executes the disconnection safety protection logic, cuts off the current output of the tension servo motor and shuts off the main drive motor.

7. The diamond electrical discharge machining (EDM) composite cutting control system for precision hard and brittle substrates according to claim 1, characterized in that, The pulse power supply (3) includes multiple inverter sub-modules connected in parallel; The control module (1) is equipped with a programmable delay compensation unit, which identifies the actual time delay deviation of the gate drive signal sent to each inverter submodule, and performs time phase calibration through the programmable delay compensation unit to compensate for the actual time delay deviation to obtain the total transient superimposed current of synchronous superposition. When any of the actual time delay deviations exceeds the maximum allowable time delay threshold, the control module (1) immediately stops sending the synchronization trigger signal and shuts down all the inverter submodules.

8. The diamond electrical discharge machining (EDM) composite cutting control system for precision hard and brittle substrates according to claim 1, characterized in that, The control module (1) identifies the average discharge voltage of the processing gap; When the average discharge voltage is lower than the short circuit precursor threshold, a deceleration command is sent to the wire saw cutting module (2); When the average discharge voltage is higher than the open circuit precursor threshold, the feed rate of the wire saw cutting module (2) is increased.

9. The diamond electrical discharge machining (EDM) composite cutting control system for precision hard and brittle substrates according to claim 6, characterized in that, The control system also includes a frequency-to-voltage conversion circuit and a multi-channel synchronous analog-to-digital converter; the frequency-to-voltage conversion circuit converts the high-frequency pulse sequence output by the encoder into an analog voltage signal, which is then converted into a digital input to the control module (1) by the multi-channel synchronous analog-to-digital converter.

10. The diamond electrical discharge machining (EDM) composite cutting control system for precision hard and brittle substrates according to claim 9, characterized in that, The wire saw cutting module (2) includes an encoder installed at the tail end of the multi-wire guide wheel drive shaft; If the analog voltage signal remains below the wire breakage reference threshold and the duration exceeds the preset activation boundary during the period when the main drive motor receives the running command and the wire feeding speed set by the process exceeds the preset activation boundary, the control module (1) determines that the encoder or frequency-voltage conversion circuit has failed, cuts off the closed-loop adjustment branch, forcibly locks the control parameters on the safe preset constant template, and outputs the channel abnormality code to the external terminal.

Images