Intelligent control method and system of laser oscillator

Abstract

The invention relates to the technical field of laser processing process control, and discloses an intelligent control method and system for a laser oscillator, and the method comprises the steps: collecting an optical signal, determining the current spectrum ratio measurement, and adjusting the laser oscillator according to the current spectrum ratio measurement; according to the method and the system, the technical problem that a static control target is easy to lose efficacy when facing working condition drift is solved, the control precision is improved by monitoring the adjusting quantity of the controller, and the control precision is improved. A fast and slow loop self-adaptive control logic is constructed, so that the system obtains the capability of self-calibrating a control target value, thereby being capable of self-adapting to working condition drift caused by material batches or equipment aging, and ensuring long-period process consistency.

Description

Technical Field

[0001] This invention relates to an intelligent control method and system for a laser oscillator, belonging to the field of process control technology for laser processing. Background Technology

[0002] Currently, especially in high-energy beam processing applications involving laser oscillators, a common technical practice is to determine a set of static parameters, including laser power, motion speed, and focal position, through preliminary process calibration. These parameters are then fixed in the process control system, with the expectation of achieving stable reproduction of the calibration state in subsequent mass production. However, this control method, which relies on static parameters, is inherently limited by the assumption that the production conditions are equivalent to the calibration conditions. In real, long-term industrial environments, factors such as minute differences in reflectivity or trace elements between material batches, transient fluctuations in protective airflow, micron-level tolerances in workpiece assembly, and even environmental electromagnetic interference and equipment aging are all unavoidable sources of dynamic disturbance. These disturbances continuously impact the static and fixed process window.

[0003] To address such disturbances, the industry has begun to explore the introduction of online monitoring and feedback. However, existing online monitoring and feedback control systems often focus solely on ensuring the safety of the control link's continuity, failing to address the adaptability issues of the process objectives themselves. For example, Chinese invention patent CN102377102B discloses a laser oscillator control device. This scheme discloses a mechanism that uses a controller to send alternating signals and monitor the laser oscillator's return signals to determine if the communication status is abnormal and to stop laser irradiation when an abnormality occurs. While this approach solves the safety issues caused by communication interruptions, it is essentially a passive, hard-stop logic based on communication continuity. It cannot identify whether the operating conditions have slowly drifted, nor can it adaptively correct the static process target values ​​upon which the control system relies. Once the operating conditions change, such as material batch changes, the system will continue to execute the outdated static parameters, failing to provide adaptive correction. A truly intelligent control mechanism that adapts to operational condition drift is needed, but conventional feedback control methods themselves bring several new technical problems. First, such control systems usually rely on an offline calibrated, fixed control target. When the operational condition drifts slowly and permanently, such as when a new batch of materials is used, the control system lacks a mechanism to identify that its target has failed and continues to adjust based on the failed target, continuously producing defective products. Second, conventional feedback only adjusts based on the deviation that has already occurred. It has control lag for procedural disturbances that are bound to occur by the control program, such as speed changes at movement corners, and cannot compensate for them in advance. Finally, during the acquisition process, the signals of high-sensitivity sensors inevitably become mixed with environmental noise unrelated to the process itself. Existing control systems lack dynamic noise identification and purification mechanisms, which can easily misjudge noise as process instability, leading to incorrect adjustment actions.

[0004] Therefore, the technical problem to be solved by this invention is how to construct a new type of intelligent control method that can not only suppress random disturbances, but also simultaneously achieve adaptive control of operating condition drift, compensation for program disturbances, and dynamic purification of environmental noise. Summary of the Invention

[0005] This invention provides an intelligent control method and system for a laser oscillator, the main purpose of which is to solve the problem that existing control methods generally lack adaptive correction, compensation and dynamic purification capabilities when facing operating condition drift, programmed disturbances and environmental noise.

[0006] To achieve the above objectives, the present invention provides an intelligent control method for a laser oscillator, comprising: Step 101: Real-time acquisition of optical signals generated in the interaction zone between the laser oscillator and the workpiece to obtain timing signals; Step 102: During the first control period, the timing signal is processed to determine the current spectral ratio metric used to characterize the stability of the interaction region process. Step 103: Compare the current spectrum ratio metric with the target spectrum ratio metric, where the target spectrum ratio metric represents the preset steady-state process window; Step 104: Based on the comparison result of step 103, generate an instantaneous adjustment amount according to the first control logic, and adjust one or more operating parameters of the laser oscillator based on the instantaneous adjustment amount; Step 105: Monitor and statistically analyze the instantaneous adjustment amount generated in step 104 within the second control cycle. The second control cycle is longer than the first control cycle in terms of time scale. Step 106: Based on the results of the statistical analysis in step 105, determine whether there is a systematic deviation in the instantaneous adjustment amount from the preset zero-point reference. Step 107: If a systematic deviation is determined, the preset value of the target spectral ratio measurement in step 103 is automatically and progressively corrected. The magnitude and direction of the correction are controlled by the degree to which the result of the statistical analysis in step 105 deviates from the preset zero-point reference, until the result of the statistical analysis in step 105 returns to the preset zero-point reference.

[0007] Preferably, determining the current spectral ratio metric in step 102 includes: performing spectral analysis on the time-series signal to obtain its signal energy in at least the high-frequency band and its signal energy in at least the low-frequency band; ,in, As a measure of the current spectrum ratio, This refers to the signal energy within the high-frequency band. This refers to the signal energy within the low-frequency band.

[0008] Preferably, performing spectral analysis on the timing signal includes: performing a fast Fourier transform or wavelet transform on the timing signal within a rolling time window to obtain its power spectral density; setting the frequency range of the high-frequency band as a characteristic frequency range characterizing welding spatter or keyhole instability; and setting the frequency range of the low-frequency band as a characteristic frequency range characterizing stable conductive welding modes or stable flow of the molten pool.

[0009] Preferably, the result of the statistical analysis in step 105 is to calculate the long-term average value of the instantaneous adjustment amount within the second control cycle; in step 107, the correction is achieved through an integrator, the input of which is the deviation between the long-term average value of the instantaneous adjustment amount and the preset zero-point reference, and the output of which is used to correct the preset value of the target spectrum ratio metric.

[0010] Preferably, the method further includes: step 501, monitoring the operating status of the laser oscillator; step 502, when the laser oscillator is detected to be in a closed state, reusing the acquisition method of step 101 to acquire the background noise signal and determine its noise spectrum; and in step 102, processing the time-series signal, including: firstly performing spectral analysis on the time-series signal to obtain its time-series signal spectrum; before calculating the current spectral ratio metric, subtracting the component corresponding to the noise spectrum from the time-series signal spectrum to obtain the purified process signal spectrum; and determining the current spectral ratio metric based on the purified process signal spectrum.

[0011] Preferably, in step 104, the first control logic is a PID control algorithm, and the instantaneous adjustment amount is calculated based on the PID control algorithm; the method further includes: step 601: based on the current spectrum ratio metric determined in step 102, or the difference between the current spectrum ratio metric and the target spectrum ratio metric, dynamically schedule one or more controller gains of the PID control algorithm; the scheduling logic is configured to: increase the controller gain when the current spectrum ratio metric is greater than a first preset threshold; and decrease the controller gain when the current spectrum ratio metric is less than a second preset threshold.

[0012] Preferably, the operating parameters adjusted in step 104 include at least one of laser power, laser pulse frequency, laser pulse duty cycle, or laser beam oscillation mode.

[0013] Preferably, the method further includes: step 801, acquiring the preset motion path information of the laser oscillator or the workpiece in real time; step 802, based on the motion path information, identifying path features that may cause predictable disturbances to the stability of the interaction zone process, the path features being changes in path curvature or motion speed; step 803, generating feedforward adjustment parameters based on the identified path features; and in step 104, adjusting one or more operating parameters of the laser oscillator is performed jointly based on the instantaneous adjustment amount and the feedforward adjustment parameters, and the initial value of the target spectral ratio measurement in step 103 is obtained in advance by calibrating a stable laser processing process.

[0014] Preferably, the method further includes: while adjusting the operating parameters in step 104, additionally superimposing a preset active detection micro-perturbation signal with specific frequency characteristics onto the operating parameters of the laser oscillator; analyzing the optical signal acquired in step 101 in parallel to demodulate the response signal amplitude corresponding to the specific frequency characteristics of the active detection micro-perturbation signal in the optical signal; determining the stability margin index of the interaction region process based on the response signal amplitude; and adjusting the basic process parameters of the laser oscillator or outputting a warning signal when the stability margin index exceeds a preset margin threshold.

[0015] An intelligent control system for a laser oscillator includes: The signal acquisition module is used to acquire optical signals generated in real time in the interaction zone between the laser oscillator and the workpiece to obtain timing signals; The spectrum analysis module is used to process the time-series signal during the first control period to determine the current spectrum ratio metric used to characterize the stability of the interaction region process. The first control module is used to compare the current spectrum ratio measurement with the target spectrum ratio measurement, generate an instantaneous adjustment amount according to the first control logic, and adjust one or more operating parameters of the laser oscillator based on the instantaneous adjustment amount; The status monitoring module is used to monitor and statistically analyze the instantaneous adjustment quantity generated in step 104 during the second control cycle, and based on the results of the statistical analysis, determine whether there is a systematic deviation of the instantaneous adjustment quantity from the preset zero point reference. The target correction module is used to automatically and progressively correct the preset value of the target spectral ratio measure if a systematic deviation is determined. The magnitude and direction of the correction are controlled by the degree to which the statistical analysis results deviate from the preset zero-point reference, until the statistical analysis results return to the preset zero-point reference.

[0016] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention establishes a dual-dimensional control closed loop for signal purification and baseline self-calibration. First, during execution, the control system can synchronously sense the operating status of the laser oscillator. When it is in a closed state (program dead zone), the system uses this gap to sample the background noise signal of the same optical acquisition channel and determine its noise spectrum. When the laser is turned on and control is executed, the system subtracts the dynamically updated noise spectrum from the currently acquired total spectrum to obtain a purified time-series signal that more accurately reflects the stability of the physical process itself. Based on this, the method further establishes a slow-time-scale adaptive correction mechanism: the system continuously monitors... The system measures and statistically analyzes the regulation output on a fast timescale to maintain process stability. When a systematic deviation is detected in the long-term average value of this regulation, the system determines that the current operating conditions, such as changes in material batches, have rendered the original target spectral ratio measurement inapplicable. It then initiates a gradual correction procedure to automatically adjust the value of the target spectral ratio measurement until the long-term statistical deviation of the regulation returns to near zero. This dual mechanism of first purifying the input and then calibrating the target enables the control system to maintain the accuracy and robustness of the control logic when facing complex real industrial environments such as electromagnetic interference, optical crosstalk, and unavoidable production conditions such as material and gas changes.

[0017] 2. This invention constructs a composite control architecture that combines feedforward pre-compensation and real-time feedback adjustment. On the one hand, the system constructs a fast feedback control path by real-time acquisition of optical signals and analysis of their spectral ratio measurement to respond to and suppress unpredictable random disturbances during the processing. On the other hand, the system acquires the motion path information to be executed in parallel and identifies path characteristics that will cause predictable disturbances to process stability, such as changes in path curvature or motion speed. Before the disturbance actually occurs, the system adjusts the operating parameters of the laser oscillator based on the identified path characteristics. This method decouples known programmed disturbances from unknown random disturbances: the feedforward path actively cancels out the large energy fluctuations that are inevitable due to changes in geometric path, while the feedback path can focus more on handling the remaining small-range random fluctuations. The two work together to maintain a high degree of process consistency in the processing even on complex geometric paths (including straight lines and corners), avoiding the quality defects that are inevitable at path feature points due to control lag in single feedback control.

[0018] 3. This invention implements a self-scheduling control response characteristic based on the real-time state of the process. Traditional control methods typically employ a fixed set of controller gains, making it difficult to simultaneously achieve both disturbance rejection capability in highly stable states and rapid error correction capability in highly unstable states. This invention, while judging the current process stability through a spectrum ratio metric, also utilizes the current spectrum ratio metric value or its difference from the target value to dynamically schedule one or more controller gains in real time. The scheduling logic is configured as follows: when the process is judged to be in a highly unstable state, the controller gain is increased, causing the controller to make a faster and larger adjustment action to force the process back to the stable region; while when the process is judged to be in a highly stable state, the controller gain is decreased, making the controller insensitive to small signal fluctuations and avoiding unnecessary oscillations caused by overreaction in the control system. Attached Figure Description

[0019] Figure 1 This is a logic block diagram of an intelligent control method for a laser oscillator according to the present invention; Figure 2 This is a data comparison chart of the present invention when slow loop correction is lacking; Figure 3 This is a schematic diagram of the functional architecture and interaction of the intelligent control system of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be described in detail below with reference to the embodiments of this invention. It should be noted that the embodiments of this invention are for illustrating this invention and not for limiting the scope of protection of this invention. In the absence of conflict, the embodiments of this invention and the technical features in the embodiments can be combined arbitrarily.

[0021] This invention provides an intelligent control method and system for a laser oscillator. As an advanced process control system, its architecture includes a signal acquisition module, a spectrum analysis module, a first control module, a state monitoring module, and a target correction module. The signal acquisition module, physically implemented as a high-bandwidth photodiode, is used to coaxially or off-axis align with the interaction area between the laser oscillator and the workpiece, acquiring the optical radiation signal generated in this area in real time and outputting a timing signal. The spectrum analysis module is connected to the output of the signal acquisition module and is used to process and analyze the timing signal. The first control module is connected to the output of the spectrum analysis module and is used to generate instantaneous adjustment values ​​to rapidly adjust the operating parameters of the laser oscillator. The state monitoring module is connected to the output of the first control module and is used to monitor the instantaneous... The long-term statistical characteristics of the adjustment quantity; the target correction module is connected to the state monitoring module and the first control module, and is used to adaptively correct the control benchmark of the first control module according to the analysis results of the state monitoring module, thereby constructing an adaptive control architecture combining fast and slow loops; an intelligent control method for a laser oscillator is specifically implemented through the above system. In a preferred embodiment, the method is anchored in the high-speed laser sealing welding scenario of a power battery casing. This scenario has zero tolerance for defects such as spatter and holes, and faces the real challenges of material batch, gas flow rate and other operating condition drift; the execution of the method includes the following steps: in step 101, the signal acquisition module is started to perform high-frequency sampling of the optical signal emitted from the laser-workpiece interaction area, such as the weld pool or keyhole, to obtain the time-series signal characterizing the dynamics of the process. In step 102, the spectrum analysis module analyzes the acquired timing signal within the first control period, for example, a rolling time window of 5 milliseconds. Perform spectrum analysis to determine the current spectrum ratio metric; in its implementation, spectrum analysis uses the Fast Fourier Transform (FFT) algorithm to analyze the time-series signal. Transform to the frequency domain to obtain its power spectral density; to perform this step, at least two frequency bands need to be predefined, one high-frequency band and one high-frequency band. and low frequency band The frequency band is determined according to standardized calibration procedures, namely... The frequency range is set to characterize the frequency range of welding instability, such as high-frequency pulses generated by spatter and severe keyhole fluctuations, and its range can be set from 5kHz to 10kHz. The frequency range is set to the characteristic frequency range characterizing stable conductive welding or smooth flow of the molten pool, and its range can be set from 0 kHz to 1 kHz; subsequently, the spectrum analysis module calculates and outputs the current spectrum ratio metric. The calculation formula is as follows: ,in, This refers to the signal energy within a high-frequency band, such as the integral of the power spectral density in the 5kHz to 10kHz band. For the signal energy in the low-frequency band, such as the integral of the power spectral density in the 0kHz to 1kHz band, this As a quantitative indicator of real-time process stability, an increase in the value indicates that the process is tending towards instability.

[0022] In steps 103 and 104, the first control module executes fast feedback control logic, which in a preferred implementation is a PID control algorithm; the first control module first processes the output of the spectrum analysis module... Measured by the ratio of the target spectrum to the preset target spectrum In comparison, the The initial value is obtained in advance through a calibrated and stable laser processing process; the PID control algorithm is based on... and The difference between them generates an instantaneous adjustment value, and based on this instantaneous adjustment value, one or more operating parameters of the laser oscillator are dynamically fine-tuned in real time; for example, when Surges instantaneously and above At this time, the PID controller outputs a negative instantaneous adjustment, causing the laser power to decrease instantaneously or activating a high-frequency beam oscillation mode to suppress keyhole instability and force the process to return to a stable conductive welding mode. Adjustable operating parameters include, but are not limited to, at least one of laser power, laser pulse frequency, laser pulse duty cycle, or laser beam oscillation mode. In steps 105, 106, and 107, the status monitoring module and the target correction module execute slow adaptive correction logic in parallel to address condition drift caused by material batch changes or equipment aging. In step 105, the status monitoring module continuously monitors and statistically analyzes the instantaneous adjustment generated by the first control module (fast loop) within the second control cycle. The second control cycle is longer than the first control cycle in terms of time scale, for example, 1 minute or the processing time of a complete workpiece. Statistical analysis in step 106 specifically involves calculating the long-term average value of the instantaneous adjustment within the second control cycle. It also determines whether the long-term average value has a systematic deviation from the preset zero-point reference, which is usually set to 0. In a system matched to healthy operating conditions, the instantaneous adjustment should fluctuate symmetrically around the 0 value, and its long-term average value... It approaches 0 infinitely; however, when the operating conditions drift, the spectral ratio corresponding to the true optimal steady state changes. It has deviated from the original calibration. At that time, the first control module will maintain the outdated... However, the continuous output of biased instantaneous regulation leads to its long-term average value. Deviation from 0; In step 107, once the status monitoring module determines that a systematic deviation exists, the target correction module is activated and automatically and progressively corrects the deviation. The preset value; this correction is achieved through an integrator, whose input is the long-term average value of the instantaneous adjustment. The integrator's output is used to slowly and gradually adjust the deviation from the preset zero-point reference. The value, until When the statistical results regress to near the zero-point baseline, the system automatically finds the true optimal steady-state point corresponding to the current operating condition. The self-calibration of the control target was completed.

[0023] To enhance the system's anti-interference capability in real industrial environments, dynamic noise reduction logic is also integrated. The system monitors the operating status of the laser oscillator in parallel, i.e., its on or off state. When the laser oscillator is detected to be off, such as during a program dead zone during workpiece switching, the system immediately reuses the signal acquisition module to acquire the pure background noise signal. The spectrum analysis module for this Processing is performed to determine the noise spectrum characterizing the current electromagnetic interference or optical crosstalk. Subsequently, when the laser oscillator is turned on and processing is performed, the internal processing logic of the spectrum analysis module is enhanced during step 102: firstly, the acquired total signal is processed... Perform spectral analysis to obtain the spectrum of the time-series signal. In calculating the current spectral ratio metric Before that, let's start from Subtract the cached value from the middle The corresponding component, i.e. To obtain the purified process signal spectrum ; the follow-up The calculations and controls are all based on this purified [material / material]. This mechanism enables the control system to autonomously distinguish between real process fluctuations and irrelevant environmental noise. To improve the system's control efficiency for nonlinear processes, the system also integrates self-scheduling logic for control gain. When the first control module executes the PID control algorithm, one or more of its controller gains, such as... or It is no longer statically fixed, but dynamically scheduled; the scheduling logic is configured to reuse the logic determined in step 102. ,or and The difference, as a state variable, when Much larger When the process is in a dangerous or highly unstable state, the system automatically increases the controller gain, causing the controller to make faster and larger adjustments to force the process back to the stable region; while when In In the vicinity or lower safety zone, when the process is in a highly stable state, the system automatically reduces the controller gain, making the controller insensitive to small signal fluctuations.

[0024] To enhance the system's ability to handle foreseeable disturbances, compensation logic is integrated, constructing a composite control architecture that combines feedforward pre-compensation with real-time feedback adjustment. The system acquires the preset motion path information of the laser oscillator or workpiece in parallel, for example, by reading the G-code program from the CNC controller. Based on this path information, the system identifies path characteristics that will inevitably cause disturbances to process stability, such as abrupt changes in path curvature (e.g., a 90-degree turn) or programmed acceleration / deceleration of motion speed. Based on the identified path characteristics, a feedforward module within the system generates a feedforward adjustment parameter. Furthermore, in step 104, the final adjustment command issued to the laser oscillator is the instantaneous adjustment amount jointly generated based on the PID control logic. and feedforward adjustment parameters To execute, for example Thus, before the disturbance actually occurs, The system proactively and in advance adjusts operating parameters, such as reducing power, to counteract the foreseeable disturbance, while the feedback path can focus more on handling the remaining, small-range random fluctuations. To enable the system to achieve online diagnostic stability margin, an active detection logic is also integrated. This logic, while performing the feedback adjustment in step 104, additionally superimposes a preset parameter with specific frequency characteristics onto one of the operating parameters of the laser oscillator, such as laser power. Actively detects micro-disturbance signals; these signals have extremely small amplitudes, for example, accounting for 1% of the total power, and have a frequency of... The frequency band is deliberately chosen to be quiet within the process noise range, such as 1.3 kHz, to avoid interfering with the welding process. The system analyzes the optical signal acquired in step 101 in parallel, and demodulates the optical signal at a specific frequency using a digital phase-locked loop algorithm or FFT specific frequency analysis. Corresponding response signal amplitude ;based on This involves determining a stability margin index for the interaction region process; the underlying logic is that a system with a high margin and high robustness is insensitive to external disturbances and will suppress the interaction region process. Signal, leading to The margin is very low, and a low-margin, near-instability system is extremely sensitive to external disturbances, which will nonlinearly amplify the impact. Signal, leading to Very high; when When the margin threshold is exceeded, the system can adjust the basic process parameters or output an early warning signal.

[0025] Example 1: This example demonstrates a specific operational instance of the disclosed intelligent control method and system for a laser oscillator in a particular industrial scenario, showcasing the collaborative working mechanism between the first control module, the state monitoring module, and the target correction module. In a large-scale, long-cycle continuous welding task for a power battery casing, the intelligent control system is deployed to ensure process consistency. At the initial stage of the task, the system employs a set of offline calibrated process parameters, including a target spectral ratio metric obtained based on calibration of batch A aluminum alloy material. ,Should This corresponds to a stable conductive welding process window with no spatter and no voids. During the first two hours of the welding task, the signal acquisition module acquires optical signals in real time, and the spectrum analysis module calculates the current spectrum ratio within the first control cycle, e.g., 5 milliseconds. The first control module (fast loop) will and The parameters are compared, and an instantaneous adjustment is generated based on the PID control logic to suppress random disturbances. At this time, the state monitoring module performs statistical analysis on the instantaneous adjustment within the second control cycle, for example, 1 minute, and calculates its long-term average value. Stable fluctuations around the preset zero-point reference indicate that the system is in a stable state with matched operating conditions. In the third hour of the task, the production line automatically switched to batch B aluminum alloy material. Due to differences in trace element content, the physical properties of this batch B material under laser irradiation, including plasma spectral characteristics and molten pool fluidity, have undergone slight but permanent changes. This change results in the true optimal steady-state process window for batch B material, corresponding to a specific spectral ratio. It is no longer equal to the original specification. The system's response to this operating condition drift is as follows: The spectrum analysis module calculates... It began to deviate from the original in a continuous and systematic way. To eliminate this persistent steady-state error, the first control module's PID control logic continuously outputs instantaneous adjustments deviating from zero, for example, to... Forcibly pulled back to a lower level It continuously outputs a 3% power regulation amount; at this time, the first control module itself is still performing its fast feedback function, but its operating benchmark is no longer in line with physical reality.

[0026] Under this operating condition, the function of the condition monitoring module (slow loop) begins to emerge. During the second control cycle, it monitors and statistically analyzes the instantaneous adjustment output by the first control module. The statistical analysis results of the condition monitoring module (step 106) clearly show the long-term average value. The system no longer fluctuates around the zero-point reference but systematically deviates to -3%, indicating a systematic deviation. Based on this non-zero statistical deviation, the target correction module (step 107) initiates a gradual correction procedure. This procedure uses an integrator, taking the -3% deviation as input, and its output is used to automatically and slowly correct the deviation used by the first control module. The preset value; this correction process continues. The value is gradually and automatically adjusted until it approaches the true optimal steady-state point of batch B material. ;along with As the system becomes increasingly accurate, the steady-state error sensed by the first control module (fast loop) gradually decreases, and its instantaneous adjustment also decreases accordingly; this process eventually reaches a new equilibrium state, namely... Automatically calibrated to Meanwhile, the long-term average value monitored by the status monitoring module It also returns to the vicinity of the preset zero reference. In this way, the control output of the fast loop is used by the slow loop as the input signal for diagnosing the drift of the operating condition, while the correction output of the slow loop, in turn, calibrates the control reference of the fast loop. The two work together to enable the system to autonomously adapt to the drift of the operating condition caused by the change of material batches without human intervention.

[0027] Example 2: This example objectively verifies the advanced process control effect of an intelligent control method that combines dynamic noise reduction and control gain self-scheduling mechanisms when facing complex disturbances. The experiment was conducted on a laser welding test platform, which includes a 2kW fiber laser oscillator, a three-axis CNC motion platform, an Ar gas supply system, and an intelligent control system. The system's signal acquisition module uses a 50kHz bandwidth photodiode coaxially arranged inside the welding head to acquire optical signals. The control system is implemented based on a digital signal processor (DSP). The processor is used to execute spectrum analysis, first control (PID) logic, noise reduction logic, and gain scheduling logic. The experimental design is as follows: On 5mm thick aluminum alloy workpieces from the same batch, 100mm straight-line welding is performed at a speed of 20mm / s. To simulate the complex disturbance conditions of an industrial environment, two types of disturbance sources are introduced in the experiment: the first type is environmental noise simulation, which involves placing a signal generator near the signal transmission line of the photodiode and applying a 10kHz sinusoidal interference signal with constant amplitude to simulate high-frequency electromagnetic interference generated by a servo motor or frequency converter; the second type is physical process disturbance, which occurs when welding reaches... At a certain time, a solenoid valve instantly reduces the protective gas flow rate from 20L / min to 18L / min, a 10% flow rate reduction, to simulate a real under-protection condition.

[0028] To verify the synergistic effect of different technical features in the method of this invention, the following three experimental groups were set up: Control group: a control method containing only the first control logic (fast loop) was adopted. The first control logic used a set of fixed, compromise PID controller gains. The dynamic noise cancellation logic and gain scheduling logic were not enabled in this control group; Experimental group A: based on the control group, the dynamic noise cancellation logic was enabled. As in the specific implementation, this logic uses the dead time of the G-code program, i.e. the workpiece switching gap, to sample the 10kHz ambient noise spectrum and establish the noise spectrum. Perform during the welding process The purification treatment; the first control logic (fast loop) of test group A still uses the same fixed PID controller gain as the control group; the sample group of this invention: adopts a complete technical solution, that is, simultaneously enabling dynamic noise purification logic and control gain self-scheduling logic; wherein, the gain scheduling logic is as described in the specific implementation method, based on the purified The PID gain is dynamically adjusted when... near (When set to 0.05) a low gain is used, when When the gain increases (e.g., exceeding 0.3), switch to high gain. During the experiment, monitor and record the current spectral ratio in real time. The dynamic changes were observed, and the number of spatter particles on the 100mm weld was counted after welding. The test results are shown in Table 1.

[0029] Table 1: Comparison of Control Effects under Combined Disturbances

[0030] The analysis of the experimental data in Table 1 is as follows: The control group, because noise purification was not used, Readings at The system was immediately affected by 10kHz ambient noise, resulting in a high baseline of 0.85, which caused the PID controller to make spurious adjustments based on this erroneous signal; when When a physical disturbance occurs, its fixed gain cannot respond quickly to the actual instability, leading to The peak value spiked to 3.42, and the weld produced a large amount of spatter (12.3 spatters); Test Group A enabled dynamic noise reduction logic, in At that moment, the system sampled and identified the 10kHz noise spectrum during the program dead zone, and subtracted it from the total spectrum during the welding process to obtain the purified process signal spectrum. All are worth returning to the true low of 0.08, avoiding a false adjustment; when When physical disturbances occur, the controller can clearly identify the actual instability signal, but because its PID gain is still fixed and a compromise, the response speed is insufficient. The peak value still reached 1.95, while the number of splashes decreased to 4.1, showing an improvement in control performance. Based on experimental group A, the sample group of this invention further enabled the self-scheduling logic of the control gain. At that time, the system was based on a low 0.06 The mean value places the PID controller in a conservative, low-gain state, avoiding overreaction to small fluctuations; when Physical disturbance occurs, When attempting to spike, the gain scheduling logic detects... If the threshold is quickly exceeded, the PID controller is immediately switched to a high-gain, aggressive state, outputting a large instantaneous adjustment to forcibly pull back the unstable process. The peak value was suppressed to 0.41, and the number of weld spatters was close to zero (0.2). After the disturbance ended, As the gain drops, the gain scheduling logic adjusts the PID back to a low-gain state, and the system resumes stable monitoring. This experimental data confirms that the dynamic noise purification logic is a prerequisite for the effective operation of the control gain self-scheduling logic, while the gain scheduling logic improves the control efficiency of the purified signal. The synergistic effect of the two enables the method of this invention to maintain the stability and rapid response capability of the control system when facing the dual challenges of environmental noise and physical disturbances.

[0031] To further illustrate the necessity of the adaptive correction mechanism in steps 105 to 107 of the present invention for solving the operating condition drift problem, the following embodiments are provided.

[0032] Example 3: This example illustrates the control effect relying solely on the first control module (fast loop) when the target correction module is missing, i.e., steps 105 to 107 (slow loop) are not executed. The experimental conditions are exactly the same as the initial stage of Example 1, and the control system only executes steps 101 to 104, which is the target spectral ratio measurement it relies on. The value was fixed at 0.05 based on the calibration of batch A material, and no adaptive correction was performed during operation; during the first two hours of the test, the system welded batch A aluminum alloy material, and the first control module (fast loop) operated normally, with its instantaneous adjustment value being... The long-term average value fluctuates around the zero-point reference, and the weld section shows that its penetration depth is 5.0 mm, which meets the process requirements.

[0033] In the third hour of the experiment, the production line automatically switched to batch B aluminum alloy material. The physical properties of this material had drifted, and its corresponding true optimal steady-state spectrum ratio... It has changed to 0.15; at this point, the first control module (fast loop) of the control system detects the current spectral ratio measurement. (Its physical value is 0.15) higher than the fixed target value. (0.05), and judged it as a continuous steady-state error; to eliminate this error, the PID control logic of the first control module, especially its integral element, begins to accumulate the deviation and continuously outputs a negative instantaneous adjustment; after the system has been running for several minutes, the long-term average value of this instantaneous adjustment is... Stabilizing at a non-zero systematic deviation of -3.5%, this -3.5% adjustment, for example, manifests as a continuous decrease in laser power, forcibly affects batch B material. The input deviation of the PID controller is reduced from its optimal value of 0.15 to 0.05, thus bringing the input deviation back to zero; although the control system appears stable on the data at this point, i.e. equal However, its physical process has been artificially pushed away from the true optimal process window of batch B materials; cross-sectional analysis of the weld at this time revealed that due to the laser power being operated at a state that was 3.5% lower than the optimal value for a long time, the weld penetration depth was only 3.1mm, resulting in process defects such as incomplete penetration or false welding.

[0034] Example 4: This example combines Figures 1 to 3 A description of an intelligent control method and system for a laser oscillator, such as... Figure 1As shown, the signal acquired from the laser-workpiece interaction zone is processed by the signal acquisition module in step 101: acquiring timing signals, and then sent to the spectrum analysis module in step 102: determining the current spectrum ratio metric. This module also receives dynamic noise purification input from the background noise acquisition and analysis module, which determines the background noise spectrum when the laser oscillator is off. The output of the spectrum analysis module is sent to the fast loop of the first control module in steps 103 / 104: generating an instantaneous adjustment amount. This instantaneous adjustment amount directly adjusts the laser oscillator on one hand, and is also monitored by the slow loop of the status monitoring module in steps 105 / 106: monitoring long-term statistical deviations. When the status monitoring module detects a systematic deviation result, the result is sent to the slow loop of the target correction module in step 107: correcting the target spectrum ratio metric to correct the control reference of the fast loop of the first control module. At the same time, a parallel feedforward path starts from the preset motion path information, and the path feature recognition module identifies the curvature / velocity changes to generate feedforward adjustment parameters, which work together with the instantaneous adjustment amount to adjust one or more operating parameters of the laser oscillator.

[0035] like Figure 2 As shown in the figure, the horizontal axis represents time (in hours), the left vertical axis represents weld penetration (in mm), and the right vertical axis represents instantaneous adjustment (%). The solid line in the figure indicates that the weld penetration, at 1.5 hours, decreased from the calibration value of 5.0 mm to a defect value of 3.1 mm due to operating condition drift. Simultaneously, the dashed line represents the instantaneous adjustment, which systematically deviates from 0% to -3.5% to maintain a failed control target. Figure 3 As shown, the operator can perform initial parameter calibration and execute intelligent control. The intelligent control logic includes executing slow loop adaptive correction and acquiring and analyzing optical signals. The optical signal acquisition and analysis module receives input from an execution dynamic noise reduction module and provides its output to the execution fast feedback adjustment module and the dynamic scheduling controller gain module, respectively. The output of the dynamic scheduling controller gain module is also fed into the execution fast feedback adjustment module. At the same time, an external CNC controller provides input to the execution feedforward compensation module, and the output of the execution feedforward compensation module is also fed into the execution fast feedback adjustment module. Finally, the execution fast feedback adjustment module controls the laser oscillator in a unified manner.

[0036] Example 5: When an intelligent control system is first deployed on a new laser welding workstation, in order to enable it to perform advanced process control and adaptive correction, a standardized initial parameter calibration procedure must first be executed to determine the core control baseline of the system. In the engineering practice of advanced process control, this procedure is designed as a calibration process consisting of three consecutive stages. The first stage of this calibration procedure aims to determine the metric used to calculate the current spectral ratio. high frequency band With low frequency band The operator first runs an unstable welding process on the test workpiece by setting a combination of process parameters that would cause keyhole instability, such as deliberately reducing the shielding gas flow rate. The signal acquisition module acquires the optical signals of this process in real time, and the spectrum analysis module performs wideband spectrum analysis on it, for example, 0-25kHz. At this point, one or more energy peaks characterized by keyhole instability or spatter are found in the frequency band, for example, 5kHz to 10kHz. The system then identifies this frequency band (5kHz to 10kHz) as the high-frequency band. Subsequently, the operator adjusted the parameters, running a lower-power, more stable conductive welding mode, and performed the same wideband spectrum analysis on the acquired optical signal. It was found that the signal energy was mainly concentrated in the 0kHz to 1kHz frequency band, and the system thus identified this frequency band (0kHz to 1kHz) as the low-frequency band. .

[0037] The second phase of this calibration procedure aims to determine the target spectral ratio metric. The initial preset value, in and Once the frequency band is determined, operators conduct process experiments, such as combining metallographic section analysis or non-destructive testing of the weld, to find the parameter combination confirmed as the optimal process window. Within this window, the weld penetration depth is appropriate and there are no defects. The system then operates stably within this optimal process window and is instructed to perform target calibration actions. During this action, the spectrum analysis module continuously calculates in real-time (based on the determined parameters) within a time period of, for example, 30 seconds. and ) And calculate all within these 30 seconds The time average of the readings; if this average is 0.05, it is determined as the target spectral ratio measure. The initial preset value is stored as the control reference for the first control module (fast loop) and the target correction module (slow loop); the third stage of this calibration procedure aims to construct the control gain self-scheduling logic; in this embodiment, the scheduling logic is specifically implemented as a one-dimensional lookup table, whose input is... The output is the controller gain of the PID controller, such as the proportional gain. Based on the second phase already determined The threshold is set to 0.05. The operator defines several key state points: First, a threshold for a highly stable state is defined, i.e., a second preset threshold, for example, set to 0.2. When the gain is below 0.2, the system is in a safe zone, and the lookup table is set to output a lower controller gain, for example... =0.5• ,in The nominal gain is defined; secondly, a threshold for a highly unstable state is defined, i.e., a first preset threshold, for example, set to 0.4. When the value is above 0.4, the system is in the danger zone, and the lookup table is set to output a higher controller gain, for example... =3.0• ; in the range of 0.2 to 0.4 Within the interval, the gain values ​​in the lookup table are automatically filled by linear interpolation. After this procedure is completed, all core control parameters of the intelligent control system, including the frequency band of spectrum analysis, the control target of fast feedback, and the nonlinear gain scheduling logic table, are determined. At this point, the system has completed the initialization settings and has the technical basis to execute subsequent process control and adaptive correction.

[0038] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0039] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method of intelligent control of a laser oscillator, characterized in that, Comprising: Step 101, collecting optical signals generated by the interaction between the laser oscillator and the workpiece in real time to obtain a time series signal; Step 102, processing the time series signal in a first control cycle to determine a current spectral ratio metric for characterizing the process stability of the interaction region; Step 103, comparing the current spectral ratio metric with a target spectral ratio metric, which characterizes a pre-set steady-state process window; Step 104, generating an instantaneous adjustment amount according to the first control logic based on the comparison result of step 103, and adjusting one or more operating parameters of the laser oscillator based on the instantaneous adjustment amount; Step 105, monitoring and statistically analyzing the instantaneous adjustment amount generated in step 104 in a second control cycle, which is longer in time scale than the first control cycle; Step 106, determining whether there is a systematic deviation of the instantaneous adjustment amount from a pre-set zero reference based on the result of the statistical analysis of step 105; Step 107, if it is determined that there is a systematic deviation, automatically and gradually correcting the pre-set value of the target spectral ratio metric in step 103, the correction amplitude and direction being controlled by the degree of deviation of the result of the statistical analysis of step 105 from the pre-set zero reference, until the result of the statistical analysis of step 105 returns to the pre-set zero reference.

2. The intelligent control method of a laser oscillator according to claim 1, wherein, The determining of the current spectral ratio measure in step 102 comprises performing a spectral analysis on the time series signal to obtain a signal energy thereof in at least a high frequency band, and a signal energy in at least a low frequency band; wherein, is the current spectral ratio measure, is the signal energy in the high frequency band, is the signal energy in the low frequency band.

3. The intelligent control method of a laser oscillator according to claim 2, wherein, The spectral analysis of the time series signal includes: performing a fast Fourier transform or wavelet transform on the time series signal within a rolling time window to obtain its power spectral density; the frequency range of the high-frequency band is set to a characteristic frequency range representing welding spatter or keyhole instability; the frequency range of the low-frequency band is set to a characteristic frequency range representing stable conduction welding mode or stable flow of the molten pool.

4. The intelligent control method of a laser oscillator according to claim 1, wherein, The result of the statistical analysis in step 105 is the long-term average value of the instantaneous adjustment amount in the second control cycle; in step 107, the correction is realized by an integrator, the input of which is the deviation of the long-term average value of the instantaneous adjustment amount from the pre-set zero reference, and the output of which is used to correct the pre-set value of the target spectral ratio metric.

5. The intelligent control method of a laser oscillator according to claim 1, wherein, Further comprising: step 501, monitoring the operating state of the laser oscillator; step 502, when it is monitored that the laser oscillator is in the off state, multiplexing the collection means of step 101 to collect a background noise signal and determine its noise spectrum; and in step 102, processing the time series signal includes: first performing spectral analysis on the time series signal to obtain its time series signal spectrum; before calculating the current spectral ratio metric, subtracting the component corresponding to the noise spectrum from the time series signal spectrum to obtain a purified process signal spectrum; determining the current spectral ratio metric based on the purified process signal spectrum.

6. The intelligent control method of a laser oscillator according to claim 1, wherein, In step 104, the first control logic is a PID control algorithm, and the instantaneous adjustment amount is calculated according to the PID control algorithm; the method further comprises: step 601: dynamically scheduling one or more controller gains of the PID control algorithm based on the current spectral ratio metric determined in step 102 or the difference between the current spectral ratio metric and the target spectral ratio metric; the scheduling logic is configured to: increase the controller gain when the current spectral ratio metric is greater than a first preset threshold; and decrease the controller gain when the current spectral ratio metric is less than a second preset threshold.

7. The intelligent control method of a laser oscillator according to claim 1, wherein, The operating parameters adjusted in step 104 include at least one of laser power, laser pulse frequency, laser pulse duty cycle, or laser beam swing mode.

8. The intelligent control method of a laser oscillator according to claim 1, wherein Further comprising: Step 801: acquiring preset motion path information of the laser oscillator or the workpiece in real time; step 802: identifying path features that are likely to cause foreseeable disturbances to the process stability of the interaction region based on the motion path information, the path features being path curvature changes or motion speed changes; step 803: generating feedforward adjustment parameters based on the identified path features; and in step 104, adjusting one or more operating parameters of the laser oscillator is performed based on both the instantaneous adjustment amount and the feedforward adjustment parameters, and the initial value of the target spectral ratio metric in step 103 is obtained by calibrating a stable laser processing process in advance.

9. The intelligent control method of a laser oscillator according to claim 1, wherein Further comprising: While adjusting the operating parameters in step 104, an active probing micro-disturbance signal with specific frequency characteristics is additionally superimposed on the operating parameters of the laser oscillator; The optical signal collected in step 101 is analyzed in parallel to demodulate a response signal amplitude corresponding to the specific frequency characteristics of the active probing micro-disturbance signal in the optical signal; Based on the response signal amplitude, a stability margin indicator of the interaction region process is determined; and when the stability margin indicator exceeds a preset margin threshold, the basic process parameters of the laser oscillator are adjusted or an early warning signal is output.

10. An intelligent control system of a laser oscillator for implementing the intelligent control method of the laser oscillator according to claim 1, characterized in that, Comprising: A signal acquisition module for acquiring optical signals generated by the interaction region of the laser oscillator and the workpiece in real time to obtain time series signals; A spectral analysis module for processing the time series signals within a first control period to determine a current spectral ratio metric for characterizing the stability of the interaction region process; A first control module for comparing the current spectral ratio metric with a target spectral ratio metric, generating an instantaneous adjustment amount according to a first control logic, and adjusting one or more operating parameters of the laser oscillator based on the instantaneous adjustment amount; A state monitoring module for monitoring and statistically analyzing the instantaneous adjustment amount generated in step 104 within a second control period, and determining whether the instantaneous adjustment amount has a systematic deviation from a preset zero reference based on the results of the statistical analysis; A target correction module for automatically and gradually correcting the preset value of the target spectral ratio metric if a systematic deviation is determined, the correction amplitude and direction being controlled by the degree of deviation of the statistical analysis results from the preset zero reference, until the statistical analysis results return to the preset zero reference.

Images