Monitoring analysis method for rapidly and accurately diagnosing titanium alloy ion nitriding arc discharge
By constructing a CNN-LSTM-Attention hybrid architecture model and high-speed optical coupling isolation technology, the problem of insufficient early warning in the arc discharge diagnosis and suppression system was solved, realizing early arc discharge prediction and rapid suppression on the titanium alloy surface, ensuring the high quality of titanium alloy workpieces and the stable operation of equipment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- METALS & CHEM RES INST CHINA ACAD OF RAILWAY SCI
- Filing Date
- 2025-12-19
- Publication Date
- 2026-05-12
AI Technical Summary
Existing arc discharge diagnosis and suppression systems are unable to provide early warning after arc discharge occurs, leading to damage to titanium alloy surfaces and equipment components, which affects workpiece quality and equipment stability.
A hybrid architecture model based on convolutional neural networks, long short-term memory networks, and attention mechanisms is adopted. Combined with electrical parameter monitoring, the local spatial features and temporal dependencies of arc discharge are extracted to achieve early prediction of arc discharge. The power supply is quickly cut off before the arc is formed through high-speed optical isolation and semiconductor power switching devices.
It enables accurate prediction and timely suppression of arc discharge, avoids microscopic damage to the titanium alloy surface, ensures the uniformity of the nitriding layer and the stability of the equipment, and improves the workpiece quality and safety in high-end manufacturing.
Smart Images
Figure CN122017477A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of titanium alloy surface strengthening equipment technology, and in particular to a rapid and accurate monitoring and analysis method for diagnosing ion nitriding arc discharge in titanium alloys. Background Technology
[0002] Titanium alloys are widely used in high-end manufacturing fields such as aerospace, medical devices, and new energy, where the surface quality and performance requirements of parts are extremely high. The high-frequency ion nitriding furnace for titanium alloys is the core equipment for surface strengthening of titanium alloys. It uses plasma generated by a high-frequency pulsed power supply to diffuse nitrogen ions into the surface layer of the titanium alloy, forming a high-hardness, high-wear-resistant nitrided layer (such as TiN or Ti2N). This significantly improves the wear resistance, corrosion resistance, and fatigue strength of the titanium alloy, enabling it to meet the requirements of high-end manufacturing.
[0003] Arc discharge (also known as "arc striking") is the most common interference problem in the operation of high-frequency ion nitriding furnaces for titanium alloys. It is extremely harmful to equipment and processes and is a key challenge affecting equipment stability, workpiece quality and production efficiency.
[0004] Arc discharge is a violent discharge phenomenon of plasma under abnormal conditions, generating instantaneous high temperatures of thousands of degrees Celsius and strong electromagnetic pulses. If not suppressed in time, it can ablate workpiece electrodes and furnace cathode electrodes, damage insulating components such as ceramic sleeves and seals, and even cause safety accidents such as furnace short circuits and fires. Oil contamination on the workpiece surface, tip discharge, and sudden changes in gas pressure are all abnormal conditions that can lead to arc discharge. Simultaneously, arc discharge disrupts plasma stability, causing fluctuations in nitrogen ion concentration, resulting in uneven nitrided layer thickness and abnormal hardness gradients. For precision titanium alloy parts, the uniformity of the nitrided layer directly determines its fatigue life and biocompatibility. Therefore, the diagnosis and suppression of arc discharge need to be integrated throughout equipment operation and process optimization.
[0005] Arc flash diagnostic suppression systems are commonly used to reduce the occurrence of arc discharge. These systems effectively protect core components of equipment by detecting arc flash in real time and quickly triggering arc extinguishing mechanisms.
[0006] Arc detection is a prerequisite for arc extinguishing, and its core objective is to rapidly capture arc signals and accurately distinguish between "normal glow discharge" and "abnormal arc discharge." Current arc detection technologies primarily rely on the acoustic, electrical, and electronic differences between glow discharge and arc discharge. The mainstream technologies for arc detection currently include ultraviolet imaging sensors, acoustic sensors, and electrical parameter monitors. Ultraviolet imaging sensors achieve high-sensitivity arc detection (response time <1ms) by capturing the ultraviolet radiation of the arc (wavelength 185-400nm). Acoustic sensors achieve spatial positioning (error <5mm) by detecting the "crackling" sound of the arc. Electrical parameter monitors determine the occurrence of arcs by collecting voltage and current fluctuations in the power supply (such as sudden voltage drops and current surges).
[0007] The core objective of rapid arc extinguishing is to quickly terminate the discharge and repair surface damage to the workpiece after the arc occurs. Current mainstream technologies for rapid arc extinguishing include high-frequency pulse power supply technology. High-frequency pulse power supplies suppress the continued development of the arc by rapidly switching the current on and off. Existing arc extinguishing technologies mainly include series high-resistance resistors, current-cutoff negative feedback arc extinguishing, thyristor bypass arc extinguishing, ion switch arc extinguishing, and pulse power supply arc extinguishing.
[0008] Existing arc light diagnostic suppression systems have a major drawback: within the few to tens of microseconds between the detection and eventual extinguishing of the arc light, a considerable amount of energy is concentrated and bombards a tiny point on the workpiece surface. For reactive metals such as titanium alloys, even a very brief, intense arc light can leave microscopic damage or structural changes on the surface, affecting the quality of the titanium alloy.
[0009] In view of this, based on years of experience in production and design in this and related fields, the inventor has designed a rapid and accurate monitoring and analysis method for diagnosing arc discharge in titanium alloys through repeated experiments, in order to solve the problems existing in the prior art. Summary of the Invention
[0010] The purpose of this invention is to provide a rapid and accurate monitoring and analysis method for diagnosing arc discharge in titanium alloys through ion nitriding, which can predict the possibility of arc discharge and thus suppress its occurrence in advance.
[0011] To achieve the above objectives, this invention proposes a rapid and accurate monitoring and analysis method for diagnosing arc discharge in titanium alloy ion nitriding, wherein the arc discharge pre-diagnosis method includes:
[0012] The electrical parameters of arc discharge during ion nitriding were collected and a database of arc discharge electrical parameters was established.
[0013] A hybrid architecture model for arc discharge prediction is constructed, which can identify abnormal patterns of electrical parameters related to arc discharge. The hybrid architecture model for arc discharge prediction includes at least a convolutional neural network layer, a memory network layer, and an attention mechanism layer.
[0014] The arc light prediction hybrid architecture model is trained using the electrical parameter database;
[0015] During the operation of the high-frequency ion nitriding furnace for titanium alloys, the trained arc discharge prediction hybrid architecture model is used to predict arc discharge.
[0016] Compared with the prior art, the present invention has the following features and advantages:
[0017] This invention proposes a rapid and accurate monitoring and analysis method for diagnosing arc discharge in titanium alloy ion nitriding. It deeply integrates plasma physics mechanisms with artificial intelligence prediction technology. Based on the inherent time difference between cathode foreign matter evaporation and arc discharge, a CNN-LSTM-Attention hybrid architecture model is constructed to accurately extract the local features and long-term temporal dependencies of electrical parameters in the time-frequency domain. The model dynamically focuses on key abnormal patterns through an attention mechanism. Combined with electrical parameters, it can accurately identify precursor features and quantify the probability of arc discharge 50-200 microseconds before the event, achieving a technological leap from post-arc extinguishing to pre-arc prevention. Simultaneously, with a ≤0.3μs ultrafast hard shutdown design based on fiber optic transmission and high-speed optical coupling isolation, it rapidly cuts off the ion power supply and suppresses shutdown overvoltage when high risk is predicted. This fundamentally avoids microscopic damage to the titanium alloy surface caused by concentrated energy bombardment, effectively maintaining plasma stability and uniform nitrogen ion concentration distribution. It significantly improves the consistency of nitrided layer thickness and the rationality of hardness gradient, providing a high-quality titanium alloy surface strengthening solution that meets stringent fatigue life and biocompatibility requirements for high-end manufacturing fields such as aerospace and medical devices. Attached Figure Description
[0018] The accompanying drawings described herein are for illustrative purposes only and are not intended to limit the scope of the invention in any way. Furthermore, the shapes and proportions of the components in the drawings are merely illustrative to aid in understanding the invention and do not specifically limit the shapes and proportions of the components. Those skilled in the art, guided by the teachings of this invention, can select various possible shapes and proportions to implement the invention according to specific circumstances.
[0019] Figure 1 This is a flowchart illustrating the rapid and accurate monitoring and analysis method for diagnosing ion nitriding arc discharge in titanium alloys according to the present invention. Detailed Implementation
[0020] The details of the present invention can be more clearly understood by referring to the accompanying drawings and the description of specific embodiments. However, the specific embodiments of the present invention described herein are for illustrative purposes only and should not be construed as limiting the invention in any way. Under the teachings of this invention, those skilled in the art can conceive of any possible modifications based on the invention, and these should all be considered to fall within the scope of the invention.
[0021] This invention proposes a rapid and accurate monitoring and analysis method for diagnosing ion nitriding arc discharge in titanium alloys, such as... Figure 1 As shown, the arc flash pre-diagnosis method includes:
[0022] The electrical parameters of arc discharge during ion nitriding were collected and a database of arc discharge electrical parameters was established.
[0023] A hybrid architecture model for arc discharge prediction is constructed. This model can identify abnormal patterns of electrical parameters related to arc discharge. The hybrid architecture model for arc discharge prediction includes at least a convolutional neural network layer, a memory network layer, and an attention mechanism layer.
[0024] The hybrid architecture model for arc prediction was trained using an electrical parameter database;
[0025] During the operation of a high-frequency ion nitriding furnace for titanium alloys, a trained arc discharge prediction hybrid architecture model is used to predict arc discharge.
[0026] The present invention proposes a rapid and accurate monitoring and analysis method for diagnosing arc discharge in titanium alloy ion nitriding. Based on the physical mechanism of arc discharge, it collects electrical parameters such as voltage and current during arc discharge and establishes a database. It then constructs a hybrid prediction model for arc discharge, which includes a convolutional neural network layer, a memory network layer, and an attention mechanism layer. This hybrid prediction model is used to train and identify the aforementioned electrical parameters. It can extract local spatial features and temporal dependence features related to the precursors of arc discharge from the fluctuations of electrical parameters. This enables early prediction of arc discharge during furnace operation, providing early warning for the suppression of subsequent arc discharge, thereby ensuring the operational stability of the high-frequency ion nitriding furnace for titanium alloys and the surface quality of titanium alloy parts.
[0027] In an optional embodiment of the present invention, the electrical parameters include at least voltage, current density, ion temperature distribution and / or arc morphology.
[0028] Specifically, in the implementation of this arc flash prediagnosis method, the electrical parameters involved are a set of key physical quantities used to characterize and identify the plasma discharge state. Voltage and current signals can be directly acquired by voltage and current sensors connected to the ion power supply output circuit or the furnace main circuit; current density can be calculated from the measured current value and the known discharge area; ion temperature distribution can be inverted by analyzing the intensity ratio of specific spectral lines using a spectral diagnostic system arranged inside the furnace; and arc morphology can be captured using a high-speed camera or photoelectric sensor facing the discharge area. During the execution of the arc flash prediagnosis method, these electrical parameters are acquired synchronously or asynchronously, forming the raw input data set for subsequent analysis and assessment of arc flash discharge risk using a hybrid arc flash prediction architecture model. By using voltage, current density, ion temperature distribution, and arc morphology as monitoring electrical parameters, the entire process characteristics of arc discharge can be captured, making full use of the physical mechanism characteristics between cathode foreign matter evaporation and arc discharge: "evaporation is slow while arc discharge is fast, and evaporation precedes arc discharge." Due to the same physical mechanism, the voltage and current characteristics from evaporation to arc discharge are repeatable, and these electrical parameters exhibit identifiable abnormal patterns before arc discharge. Through the synergistic analysis of multiple electrical parameters, the defects of single-parameter detection being susceptible to interference are effectively overcome, significantly improving the accuracy and reliability of pre-diagnosis, and enabling early warning before arc discharge causes workpiece damage.
[0029] In an optional embodiment of the present invention, a convolutional neural network layer (CNN layer) performs convolution operations on electrical parameters and extracts local spatial features of electrical parameters, outputting a sequence of feature maps.
[0030] Specifically, the convolutional neural network (CNN) layer undertakes the core task of automatically extracting key local spatial features from the input electrical parameter data. This CNN layer receives electrical parameter data such as voltage and current collected and preprocessed by sensors, and performs sliding convolution operations on the input data using its internally configured multi-layer convolutional kernels. This process automatically identifies and extracts discriminative local spatiotemporal patterns in the electrical signal, such as high-frequency spikes or abnormal fluctuation segments of specific shapes and durations in voltage waveforms. After feature extraction, the CNN layer outputs a deeper sequence of feature maps representing these identified patterns. This sequence of feature maps serves as input to subsequent network layers for further temporal correlation analysis. By employing a CNN layer to perform convolution operations on electrical parameters and extract local spatial features, this implementation achieves effective dimensionality reduction and feature enhancement of the original high-dimensional, high-noise electrical signal. This automated local feature extraction capability transforms the raw signal into a more representative and discriminative sequence of feature maps, laying a clear and efficient data foundation for subsequent processing layers to focus on key temporal evolution patterns. This, in turn, improves the overall accuracy and robustness of the hybrid architecture model in identifying early arc flash signs from complex data.
[0031] In one alternative embodiment of this implementation, the electrical parameter is voltage, and the shape and / or duration of the high-frequency spikes of the voltage are used as local spatial features.
[0032] Specifically, when the acquired electrical parameter is voltage, the local spatial features extracted by the convolutional neural network layer include the shape and / or duration of high-frequency spikes in the voltage signal. The voltage signal is acquired in real-time by a voltage sensor deployed in the ion power supply output circuit or the furnace main circuit. The shape characteristics of the high-frequency spikes refer to the specific geometric contours of abnormal protrusions or depressions in the voltage waveform, such as steepness, top flatness, or oscillation patterns; the duration characteristics refer to the width of the abnormal spike signal maintained on the time axis. These features directly reflect the instantaneous and local abnormal voltage fluctuation patterns caused by physical processes such as the evaporation of foreign matter on the cathode surface and sudden changes in local gas pressure before or during arc discharge. By explicitly using the shape and / or duration of the high-frequency voltage spikes as local spatial features extracted by the convolutional neural network layer, the model can focus on the microscopic transient information in the voltage signal that best characterizes the early precursors of arc discharge. Differences in the shape of the high-frequency spikes can correspond to discharge anomalies caused by different physical inducements, while their duration is directly related to the brief process of abnormal energy accumulation. By extracting and learning these key local features, the prediction model can more sensitively and specifically identify dangerous signs mixed in with normal voltage fluctuations, significantly improving the sensitivity and accuracy of arc discharge prediagnosis. This enables the system to issue early warnings in the early stages of arc discharge formation or even in the incubation stage, buying valuable time for subsequent suppression measures.
[0033] In an optional example, the preprocessed voltage is converted into a two-dimensional time-frequency graph, and the two-dimensional time-frequency graph is input into a convolutional neural network layer, which extracts the local spatial features of the voltage through the ReLU activation function.
[0034] Specifically, the process of converting the preprocessed voltage into a two-dimensional time-frequency graph involves: first, filtering, denoising, and normalizing the original voltage signal to eliminate measurement noise and dimensional differences; then, using the Short Time Fourier Transform (STFT) method to convert the one-dimensional time-series voltage signal into a two-dimensional time-frequency representation, where the horizontal axis represents time, the vertical axis represents frequency, and the pixel grayscale value represents the energy intensity at the corresponding time-frequency point; during the conversion, the window function length is set to 128 sampling points, and the overlap rate is 75% to balance time resolution and frequency resolution; the generated two-dimensional time-frequency graph is fed into a convolutional neural network layer containing three consecutive convolutional layers, each using a 3×3 kernel, performing sliding convolution operations on the time-frequency graph with a stride of 1; after each convolution operation, the ReLU activation function is applied. The convolution results are nonlinearly transformed, expressed as f(x) = max(0,x), which highlights the positive peak features in the voltage signal while suppressing negative noise. Through this processing, the convolutional neural network layer effectively extracts the local spatial features of the voltage signal in the time-frequency domain, including the shape, duration, and energy distribution patterns of high-frequency peaks. These local spatial features are closely related to the local pressure changes caused by the evaporation of cathode foreign matter during ion nitriding, providing key criteria for early prediction of arc discharge. This significantly improves the targeting and effectiveness of feature extraction, providing high-quality feature data support for the subsequent training and real-time prediction of the arc discharge prediction hybrid architecture model. This helps the model learn the precursor laws of arc discharge more accurately, thereby improving the accuracy of arc discharge prediagnosis.
[0035] In one alternative embodiment of this implementation, local spatial features are input into a memory network layer (LSTM layer), which receives the temporal dependencies of the local spatial features and outputs the hidden sequence state.
[0036] Specifically, the memory network layer adopts a Long Short-Term Memory (LSTM) structure, containing 128 hidden units. The memory network layer processes the input feature sequence through its internal gating mechanism, where the forget gate determines which historical information needs to be retained, the input gate controls the inflow of new information, and the output gate adjusts the amount of information in the final output. During processing, the memory network layer can identify the temporal correlations between local spatial features, such as the evolution of voltage fluctuations over time, including the decay trend after a voltage surge and the long-term dependence of pressure changes on voltage. Through a recurrent connection structure, the memory network layer combines the hidden state from the previous time step with the current input local spatial features to generate an updated... The hidden states; after processing the entire feature sequence, the memory network layer outputs a set of hidden sequence states, with each time step corresponding to a 128-dimensional hidden state vector. These hidden state vectors encode the evolution of local spatial features in the time dimension, providing temporal context information for subsequent arc discharge prediction; during ion nitriding, the local pressure change caused by the evaporation of cathode foreign matter will generate specific temporal patterns in the voltage signal. The memory network layer can capture these patterns and establish a correlation with subsequent arc discharge, significantly improving the prediction lead and accuracy, allowing the suppression system sufficient time to take preventive measures, enabling proactive detection, and preventing microscopic damage to the surface of titanium alloy workpieces.
[0037] In an optional example, the hidden sequence state is fed into the attention mechanism layer to increase the weights of key features related to arc discharge.
[0038] Specifically, after receiving the hidden state sequence, the attention mechanism layer uses its internal computation module to evaluate the importance or relevance of the hidden states at different time steps in the sequence to the final determination of the probability of arc discharge. This attention mechanism calculates the attention weight for each time step; a larger weight indicates a more critical feature at that moment. Subsequently, these attention weights are used to perform a weighted summation or reconstruction of the original hidden state sequence. This significantly increases the contribution of key historical moment features highly correlated with arc discharge precursors in the final decision-making process of the arc discharge prediction hybrid architecture model, while suppressing the influence of secondary or irrelevant moment features.
[0039] In an optional embodiment of the present invention, the electrical parameters of the high-frequency ion nitriding furnace for titanium alloy are monitored in real time during operation.
[0040] The electrical parameters obtained from monitoring are input into the trained arc light prediction hybrid architecture model;
[0041] The trained arc discharge prediction hybrid architecture model identifies anomalous patterns associated with arc discharge and calculates the probability of arc discharge occurring.
[0042] First, by connecting sensors such as voltage and current sensors to the ion power supply output circuit and the main circuit of the furnace, key electrical parameters during the nitriding furnace operation are continuously monitored and collected in real time. Then, the data stream of the monitored real-time electrical parameters, after preprocessing consistent with the model training phase, is input into the trained and deployed arc discharge prediction hybrid architecture model. The trained arc discharge prediction hybrid architecture model utilizes its learned parameters and pattern recognition capabilities to sequentially perform convolutional feature extraction, temporal dependency modeling, and attention-focusing analysis on the input real-time data, ultimately identifying abnormal patterns related to arc discharge in the data and outputting a quantified probability value representing the likelihood of arc discharge occurring at the current moment. The arc discharge pre-diagnosis method proposed in this invention forms a closed loop from offline training to online application, realizing dynamic and intelligent monitoring of the nitriding process. By embedding the trained arc discharge prediction hybrid architecture model into the real-time system, it is possible to perform millisecond-level continuous analysis of electrical parameters generated during operation, proactively and in advance detecting potential arc discharge risks. The probability values output by the arc light prediction hybrid architecture model provide clear and quantifiable early warning information for the suppression system. This allows the subsequent suppression system to obtain a valuable intervention window before the arc light energy accumulates violently and causes actual damage. This transforms the traditional "post-event response" mode into a "pre-event prevention" mode, fundamentally improving process safety and workpiece quality assurance capabilities.
[0043] In an optional embodiment of the present invention, discharge voltage, current density, ion temperature distribution and arc morphology are monitored simultaneously, with multiple detections in parallel, so as to detect fluctuations and changes in electrical parameters in a timely manner.
[0044] The present invention also proposes an electronic device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the above-mentioned arc light pre-diagnosis method.
[0045] The present invention also proposes a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the above-described arc light pre-diagnosis method.
[0046] This invention also proposes a method for suppressing arc light in a high-frequency ion nitriding furnace for titanium alloys, wherein the arc light suppression method includes:
[0047] The arc discharge pre-diagnosis method is used to determine the probability of arc discharge in a high-frequency ion nitriding furnace for titanium alloys in operation.
[0048] Based on the probability of arc discharge, predict the risk level of arc discharge and output the corresponding alarm signal;
[0049] Based on the alarm signal, the power supply of the ion power source to the high-frequency ion nitriding furnace for titanium alloy is cut off.
[0050] Specifically, firstly, an arc discharge pre-diagnosis method is used to perform real-time analysis of the operating titanium alloy high-frequency ion nitriding furnace and calculate the probability value of arc discharge occurring at the current moment. Next, based on this probability value and according to preset mapping rules, such as setting multiple probability threshold ranges, the risk level of arc discharge is predicted, and alarm signals of different levels corresponding to this risk level are output. Finally, based on the output alarm signals, control commands are generated and sent to the control unit of the ion power supply or the rapid shutdown circuit to execute the operation of cutting off the power supply to the titanium alloy high-frequency ion nitriding furnace.
[0051] The arc suppression method proposed in this invention for high-frequency ion nitriding furnaces for titanium alloys obtains the arc discharge probability by combining the aforementioned arc discharge pre-diagnosis method, predicts the risk level based on the probability and outputs a corresponding alarm signal, and then cuts off the power supply of the ion power source to the high-frequency ion nitriding furnace for titanium alloys in a timely manner based on the alarm signal. It makes full use of the time difference characteristics between cathode foreign matter evaporation and arc discharge, and actively intervenes in the early stage of arc discharge formation, especially before the current rises sharply and causes damage. This fundamentally prevents microscopic damage and structural changes on the surface of titanium alloy workpieces caused by arc discharge, ensures the uniformity of the nitrided layer and the quality of the workpiece, and avoids the ablation damage of arc discharge to insulating components such as furnace cathode electrodes, ceramic sleeves, and seals, significantly improving the stability and safety of equipment operation.
[0052] In an optional embodiment of the present invention, the ion power supply is provided with a semiconductor power switch device, which is capable of turning off the ion power supply within ≤0.3μs.
[0053] Specifically, the ion power supply incorporates a semiconductor power switch as its core power control unit. When a high-risk alarm signal for arc discharge is triggered, the control unit sends a shutdown command to the semiconductor power switch. Upon receiving the command, the semiconductor power switch, utilizing its inherent high-speed switching characteristics, completes the state switch from on to off within a time interval of less than or equal to 0.3 microseconds, thereby physically cutting off the main power circuit output from the ion power supply to the titanium alloy high-frequency ion nitriding furnace. By achieving a rapid hard shutdown of ≤0.3 microseconds through the semiconductor power switch within the ion power supply, the response speed for arc suppression is increased to the microsecond level. This extremely short shutdown time ensures that after the pre-diagnostic system identifies the precursor to arc discharge and issues an alarm, the power supply can be rapidly cut off before the arc current rises sharply and the energy concentrates to bombard the workpiece surface, causing microscopic damage. This prevents the formation and continuation of destructive arc discharge at its source, greatly protecting the integrity of the workpiece surface and ensuring equipment safety.
[0054] In an optional embodiment of the present invention, when the ion power supply is cut off, the current is provided by the independent freewheeling circuit of the current-limiting inductor at the output terminal of the ion power supply to suppress the turn-off voltage spike.
[0055] Specifically, when the ion power supply is switched off, the energy stored in the current-limiting inductor connected in series at the output of the ion power supply needs to be released. To address this, an independent freewheeling circuit consisting of a fast recovery diode is connected in parallel across the current-limiting inductor. At the instant the semiconductor power switching device is turned off and the main power circuit is disconnected, the induced electromotive force generated in the current-limiting inductor due to the sudden current change drives the stored current through the parallel fast recovery diode to form a low-impedance closed discharge path, thereby achieving freewheeling in the inductor. This allows the magnetic energy in the inductor to be gradually consumed or circulated within this circuit, rather than being released through the load circuit. By setting up an independent freewheeling circuit for the current-limiting inductor, this implementation provides a controllable and safe discharge channel for the inductive energy stored in the line while rapidly switching off the main power supply.
[0056] In an optional embodiment of the present invention, when the ion power supply is cut off, the main circuit current is diverted by a bypass arc-breaking device disposed between the cathode and anode of the titanium alloy high-frequency ion nitriding furnace.
[0057] Specifically, when the ion power supply is cut off, a bypass arc-breaking device located between the cathode and anode of the titanium alloy high-frequency ion nitriding furnace is simultaneously triggered. This bypass arc-breaking device consists of a high-speed semiconductor switch (such as a SiCMOSFET) and its drive control circuit, and is connected in parallel with the furnace's anode and cathode main circuits. Upon receiving an alarm signal from the arc diagnostic system, the drive circuit of the bypass arc-breaking device forces its internal high-speed semiconductor switch to fully conduct within a very short time, thereby establishing a transient, low-impedance bypass conductive channel between the cathode and anode of the furnace while the ion power supply is cut off. This bypass conductive channel provides a shunt path for the discharge current that is still attempting to be maintained or the residual energy in the circuit, causing the current flowing through the plasma region inside the furnace to be rapidly diverted and drastically reduced. By setting and triggering the bypass arc-breaking device between the anode and cathode, a forced current transfer and energy dissipation mechanism is added on top of the main power supply cutoff. When the ion power supply is cut off, the main circuit current is diverted by a bypass arc-breaking device set between the cathode and anode of the high-frequency ion nitriding furnace for titanium alloys. This can rapidly reduce the current density on the workpiece surface before the ion power supply is completely turned off, effectively preventing micro-melting and structural damage to the titanium alloy surface caused by arc discharge. This significantly improves the arc extinguishing success rate and process stability of the system, ensures the uniformity and surface quality of the nitrided layer on the titanium alloy workpiece, extends the service life of the equipment, and reduces the rework and scrap rate of workpieces caused by arc damage.
[0058] In an optional embodiment of the present invention, the suppression of arc energy is achieved through a three-level collaborative mechanism. Specifically, an independent freewheeling circuit with a current-limiting inductor provides a safe energy discharge path, a SiC MOSFET bypass arc-breaking device actively shunts the arc current, and distributed parameter optimization reduces the impact of stray parameters in the system.
[0059] The detailed explanations of the above embodiments are intended only to explain the present invention so as to facilitate a better understanding of the present invention. However, these descriptions should not be construed as limiting the present invention for any reason. In particular, the various features described in different embodiments can be arbitrarily combined with each other to form other embodiments. Unless there is an explicit description to the contrary, these features should be understood to be applicable to any embodiment, and not limited to the described embodiments.
Claims
1. A rapid and accurate monitoring and analysis method for diagnosing ion nitriding arc discharge in titanium alloys, characterized in that, The arc light pre-diagnosis method includes: The electrical parameters of arc discharge during ion nitriding were collected and a database of arc discharge electrical parameters was established. A hybrid architecture model for arc discharge prediction is constructed, which can identify abnormal patterns of electrical parameters related to arc discharge. The hybrid architecture model for arc discharge prediction includes at least a convolutional neural network layer, a memory network layer, and an attention mechanism layer. The arc light prediction hybrid architecture model is trained using the electrical parameter database; During the operation of the high-frequency ion nitriding furnace for titanium alloys, the trained arc discharge prediction hybrid architecture model is used to predict arc discharge.
2. The rapid and accurate monitoring and analysis method for diagnosing ion nitriding arc discharge in titanium alloys as described in claim 1, characterized in that, The electrical parameters include at least voltage, current density, ion temperature distribution, and / or arc morphology.
3. The rapid and accurate monitoring and analysis method for diagnosing ion nitriding arc discharge in titanium alloys as described in claim 1, characterized in that, The convolutional neural network layer performs convolution operations on the electrical parameters and extracts the local spatial features of the electrical parameters, outputting a sequence of feature maps.
4. The rapid and accurate monitoring and analysis method for diagnosing ion nitriding arc discharge in titanium alloys as described in claim 3, characterized in that, The electrical parameter is voltage, and the shape and / or duration of the high-frequency spikes of the voltage are used as the local spatial features.
5. The rapid and accurate monitoring and analysis method for diagnosing ion nitriding arc discharge in titanium alloys as described in claim 4, characterized in that, The preprocessed voltage is converted into a two-dimensional time-frequency graph, and the two-dimensional video graph is input into the convolutional neural network layer. The convolutional neural network layer extracts the local spatial features of the voltage through the ReLU activation function.
6. The rapid and accurate monitoring and analysis method for diagnosing ion nitriding arc discharge in titanium alloys as described in claim 3, characterized in that, The local spatial features are input into the memory network layer, and the memory network receives the temporal dependencies of the local spatial features and outputs the hidden sequence state.
7. The rapid and accurate monitoring and analysis method for diagnosing ion nitriding arc discharge in titanium alloys as described in claim 6, characterized in that, The hidden sequence state is input into the attention mechanism layer to increase the weights of key features related to arc discharge.
8. The rapid and accurate monitoring and analysis method for diagnosing ion nitriding arc discharge in titanium alloys as described in claim 1, characterized in that, During the operation of the high-frequency ion nitriding furnace for titanium alloys, the electrical parameters of the furnace are monitored in real time. The electrical parameters obtained from monitoring are input into the trained arc light prediction hybrid architecture model; The trained arc discharge prediction hybrid architecture model identifies anomalous patterns associated with arc discharge and calculates the probability of arc discharge occurring.