Intelligent monitoring and process parameter adaptive regulation and control method and system

Abstract

The invention relates to the technical field of industrial robot precision machining, in particular to an intelligent monitoring and process parameter self-adaption regulation and control method and system which are suitable for metal plate precision cutting, composite material special-shaped part cutting, pipe three-dimensional contour cutting and other scenes. Real-time state monitoring, cut quality online evaluation, process parameter dynamic optimization and equipment health early warning in the cutting process can be achieved, and the problems that in traditional laser cutting, cut quality consistency is poor, the parameter debugging period is long, and the material utilization rate is low are solved.

Description

Technical Field

[0001] This invention relates to the field of precision machining technology for industrial robots, and in particular to an intelligent monitoring and adaptive control method and system for process parameters. Background Technology

[0002] Automated laser cutting using industrial robots is a key technology for achieving high-precision machining of complex parts in high-end manufacturing. Its processing quality and efficiency directly impact product performance and manufacturing costs. Currently, laser cutting process control faces four major technical bottlenecks: First, lagging cut quality inspection. Traditional methods rely on offline inspection, which cannot assess key indicators such as cut roughness and perpendicularity in real time, leading to batch quality problems. Second, poor parameter adaptability. Using fixed parameter templates makes it difficult to adapt to changes in material thickness (0.1-20mm), material type (carbon steel / aluminum alloy / titanium alloy), and cutting trajectory (straight / curved / corner), easily resulting in defects such as slag buildup and burn-off. Third, focal point drift. The laser focal point is affected by temperature and vibration, causing micro-displacement (±0.05mm), leading to decreased cutting accuracy. Fourth, low energy utilization. The lack of a precise matching mechanism between laser energy and material removal results in energy waste and equipment wear.

[0003] In existing technologies, some solutions improve cutting quality by increasing motion precision, but lack dynamic compensation mechanisms; some solutions employ simple parameter matching strategies, which can only adapt to single working conditions; and some solutions rely on manual experience for debugging, making it difficult to achieve flexible production of multiple varieties. Therefore, there is an urgent need to build a laser cutting control system that integrates multi-source sensing and intelligent decision-making. Summary of the Invention

[0004] This invention provides an intelligent monitoring and adaptive process parameter control method and system. Through a four-layer architecture of multi-sensor fusion perception, real-time kerf quality assessment, intelligent process parameter optimization, and dynamic focus compensation, it achieves high-precision, high-efficiency, and low-energy-consumption laser cutting. The core innovation lies in proposing four key algorithms: a real-time cutting status monitoring algorithm based on plasma characteristics, a multi-objective optimized adaptive laser process parameter decision-making algorithm, a focus drift prediction and dynamic compensation algorithm, and a cutting path energy efficiency optimization algorithm. These algorithms significantly improve cutting quality and efficiency while reducing energy costs.

[0005] A first aspect of this invention provides an intelligent monitoring and adaptive control method for process parameters, comprising the following steps:

[0006] Multi-source information acquisition during the cutting process: Deploy a vision-spectrum-force fusion sensing system to acquire plasma spectral signals, cut images, laser energy parameters, robot motion parameters, and material characteristic data in real time;

[0007] Real-time evaluation of cutting status: Based on the collected data, an online evaluation of cutting quality and defect prediction are achieved through a real-time monitoring algorithm for cutting status based on plasma characteristics;

[0008] Intelligent optimization of process parameters: Based on the cutting status evaluation results, a multi-objective optimization adaptive decision algorithm for laser process parameters is activated to achieve dynamic adjustment of parameters such as laser power and cutting speed;

[0009] Dynamic compensation and path optimization for focal position: The focal position is corrected by a focus drift prediction and dynamic compensation algorithm, and the optimal cutting trajectory is generated by combining the cutting path energy efficiency optimization algorithm.

[0010] Closed-loop optimization throughout the entire process: Based on historical data and real-time feedback, the algorithm model is continuously iterated and optimized to achieve a synergistic improvement in cutting quality, efficiency, and energy consumption.

[0011] A second aspect of the present invention provides a system for implementing the above method, comprising:

[0012] Multi-source sensing units: spectrometer (300-800nm, sampling rate 1kHz), high-speed camera (resolution 1280×960, frame rate ≥200fps), laser displacement sensor (accuracy ±0.001mm), temperature sensor, vibration sensor;

[0013] Data processing unit: FPGA high-speed signal processing module, GPU accelerated computing module, industrial computer (data fusion and algorithm execution);

[0014] Intelligent Decision Unit: Deploys four core algorithms to achieve segmentation status assessment, parameter optimization, focus compensation, and path planning;

[0015] Robot and laser control unit: six-axis industrial robot (repeat positioning accuracy ±0.02mm), high-power laser (1-10kW), real-time controller (control cycle ≤1ms);

[0016] Human-machine interaction unit: Touch screen and monitoring software, which displays cutting parameters, quality assessment results and system status in real time.

[0017] Beneficial effects

[0018] Compared with the prior art, the present invention has the following beneficial effects:

[0019] Cutting quality is significantly improved: the surface roughness Ra is reduced from 12μm to below 3.2μm, dimensional accuracy is improved by 40%, and the defect rate is reduced from 2.8% to 0.2%;

[0020] Production efficiency is greatly improved: through path optimization and parameter matching, the processing cycle of a single piece is shortened by 15-20%, and the equipment utilization rate is increased by 25%;

[0021] Energy consumption costs are significantly reduced: energy consumption per unit length of cutting is reduced by 18-25%, equipment maintenance costs are reduced by 25%, and overall production costs are reduced by 30%.

[0022] Enhanced flexibility: It can automatically adapt to the cutting needs of different materials (carbon steel, aluminum alloy, titanium alloy) from 0.1-20mm, and the changeover and debugging time is shortened from 6 hours to 30 minutes. Attached image description:

[0023] Figure 1 Flowchart of the algorithm implementation principle. Detailed Implementation

[0024] Example 1: Real-time monitoring algorithm for cutting status based on plasma characteristics. Algorithm principle: By acquiring plasma spectral signals (300-800nm) and high-speed visual images generated during laser cutting, feature parameters (electron density, excitation temperature, spectral intensity ratio, plasma morphology) are extracted, and a mapping relationship is established with cutting quality (cut roughness, perpendicularity, slag amount).

[0025] Key innovations:

[0026] An adaptive spectral filtering algorithm is proposed, which effectively suppresses arc light interference and improves the signal-to-noise ratio by 40%.

[0027] Design a lightweight CNN model (PlasmaNet) to complete the segmentation state evaluation within 10ms;

[0028] A cutting defect prediction mechanism is introduced to identify potential defects (such as slag that is about to form) 50ms in advance.

[0029] Principle of efficiency enhancement:

[0030] Quality control: Through real-time monitoring and prediction, the cutting defect rate was reduced from 2.8% to 0.2%, reducing the cost of subsequent grinding processes by 30%.

[0031] Efficiency improvement: Avoid rework due to quality issues, and reduce single-piece processing time by 15%;

[0032] Material savings: Reduce scrap rate and increase material utilization rate by 8-12%.

[0033] Innovation Point 2: Adaptive Decision Algorithm for Laser Process Parameters Based on Multi-Objective Optimization

[0034] Algorithm principle: A multi-objective optimization model is established with cutting quality (cutting roughness Ra), processing efficiency (cutting speed v) and energy consumption (laser power P) as objectives. Based on material properties (thickness, material) and cutting trajectory characteristics, key parameters such as laser power, cutting speed and auxiliary gas pressure are dynamically adjusted.

[0035] Key innovations:

[0036] An improved NSGA-III algorithm is proposed, which integrates expert experience to construct adaptive crossover and mutation operators, improving the convergence speed by 35%.

[0037] Design a parameter sensitivity analysis mechanism to identify the key parameters that have the greatest impact on quality (such as the laser power sensitivity of high carbon steel reaching 0.82);

[0038] Establish a rapid response model for parameter adjustment, with a system stabilization time of ≤20ms after parameter changes.

[0039] Principle of efficiency enhancement:

[0040] Quality improvement: The surface roughness Ra value decreased from 12 μm to below 3.2 μm, and the quality consistency (CPK) improved from 1.1 to 1.8;

[0041] Reduced energy consumption: Through precise matching of power and speed, energy consumption per unit length of cutting is reduced by 18-25%;

[0042] Efficiency Improvement: Optimal parameters are automatically matched for different materials, resulting in an average cutting speed increase of 12-18%.

[0043] Innovation Point 3: Focus Drift Prediction and Dynamic Compensation Algorithm

[0044] Algorithm principle: Based on the laser head temperature field, vibration signal and historical drift data, a focal position (Z-axis) drift prediction model is established, and the focal position is corrected in real time through micro-displacement compensation of the robot end effector.

[0045] Key innovations:

[0046] A prediction model integrating LSTM and attention mechanism is proposed, with a focus drift prediction error ≤0.01mm within 10s;

[0047] A dual-closed-loop compensation control strategy is designed, with the outer loop performing pre-compensation based on prediction results and the inner loop performing fine compensation based on real-time measurement.

[0048] Develop a lightweight compensation actuator with a response time of ≤5ms and a compensation range of ±0.5mm.

[0049] Principle of efficiency enhancement:

[0050] Improved precision: Focus position control precision improved from ±0.05mm to ±0.01mm, and cutting size precision improved by 40%;

[0051] Stable quality: Avoids cutting quality fluctuations caused by focus drift, improving consistency in mass production by 50%;

[0052] Equipment protection: Reduces laser head overload caused by focus shift, extending the lifespan of core components by 20%.

[0053] Innovation Point 4: Energy Efficiency Optimization Algorithm for Cutting Path

[0054] Algorithm principle: Under the premise of ensuring cutting accuracy, based on the contour features of the part and the material properties, the cutting path sequence and corner handling strategy are optimized to reduce idle strokes and speed changes, thereby achieving efficient energy utilization.

[0055] Key innovations:

[0056] An improved TSP algorithm that incorporates energy consumption factors is proposed, introducing energy consumption weight per unit length in path planning;

[0057] Design an adaptive corner speed planning strategy to dynamically adjust the deceleration ratio (v'=v×f(θ)) based on the corner angle (θ);

[0058] Develop a common-edge cutting identification and merging algorithm to reduce repeated cutting paths by 15-25%.

[0059] Principle of efficiency enhancement:

[0060] Efficiency improvement: Idle travel time is reduced by 20-30%, and the average processing cycle per piece is shortened by 15%;

[0061] Reduced energy consumption: Overall energy consumption is reduced by 12-18% by reducing speed abrupt changes and repetitive paths;

[0062] Equipment lifespan: Reduces the number of times the robot stops and starts suddenly, reduces joint wear by 30%, and reduces equipment maintenance costs by 25%.

[0063] Example 2: Application of a Real-time Monitoring Algorithm for Cutting Status Based on Plasma Characteristics

[0064] To address the issues of lagging cut quality inspection and difficulty in controlling batch defects in traditional laser cutting, this embodiment innovatively applies a "real-time monitoring algorithm for cutting status based on plasma characteristics." Through multi-source signal fusion and intelligent analysis, it achieves online quality assessment and defect prediction during the cutting process, controlling processing quality from the source.

[0065] (I) Multi-source signal acquisition and preprocessing

[0066] The algorithm first constructs a multi-source data acquisition system of "spectrum-vision-state": it acquires the plasma spectral signal generated during the laser cutting process in real time through a spectrometer (covering the wavelength range of 300-800nm) to capture the electronic transition characteristics of the plasma; it uses a high-speed camera (frame rate ≥200fps) to capture images of the cutting area and record the shape, size and distribution changes of the plasma; and it simultaneously acquires process parameters such as laser power and cutting speed and robot motion state data to form a multi-dimensional monitoring dataset.

[0067] To address issues such as arc interference and noise in the acquired signals, an adaptive spectral filtering algorithm was developed. By analyzing the characteristic peak positions and intensity distribution of the plasma spectrum, a feature template for the interference signal was established. An adaptive threshold segmentation combined with wavelet transform was used to accurately separate the effective spectral signal from the interference signal, significantly improving the signal-to-noise ratio. For the visual images, image enhancement and edge detection preprocessing techniques were employed to highlight the plasma region and cut contours, laying the foundation for subsequent feature extraction.

[0068] (II) Feature Extraction and State Assessment Model

[0069] Core feature parameters are extracted from the preprocessed multi-source data: In terms of spectral signals, parameters such as electron density, excitation temperature, and intensity ratio of characteristic spectral lines are calculated to reflect the energy state of the plasma; in terms of visual images, features such as plasma area, morphological complexity, and centroid offset are extracted to characterize the dynamic changes of the cutting area; and multi-dimensional feature vectors are formed by combining process parameters and motion data.

[0070] A lightweight CNN model (PlasmaNet) is designed to achieve rapid evaluation of cutting status. The model adopts a depthwise separable convolutional and residual connection structure, which significantly reduces computational complexity while ensuring evaluation accuracy, and can complete a status evaluation within 10ms. The model is trained using historical cutting data (covering quality results under different materials, thicknesses, and process parameters) to establish a mapping relationship between feature vectors and cutting quality (cut roughness, perpendicularity, and slag amount), and outputs real-time quality evaluation results.

[0071] (III) Defect Prediction and Early Warning Mechanism

[0072] The algorithm incorporates temporal feature analysis to predict cutting defects in advance: by analyzing the changing trends of characteristic parameters over multiple consecutive frames, it identifies precursory signals of quality anomalies. For example, a sustained increase in the intensity ratio of plasma characteristic spectral lines and a sudden change in morphological complexity may indicate an impending slag-forming defect; an increase in the frequency of electron density fluctuations may lead to excessive cut roughness. The algorithm sets multi-level warning thresholds, identifies potential defects 50ms in advance, and notifies operators via audible and visual alarms and system prompts, while simultaneously triggering subsequent parameter adjustment mechanisms.

[0073] Efficiency Enhancement Principle: This algorithm transforms cutting quality inspection from offline to online, enabling real-time detection and early prediction of quality issues, significantly reducing the cutting defect rate. By avoiding rework due to quality problems, the processing time per piece is significantly shortened, while scrap generation is reduced, resulting in a marked improvement in material utilization. Furthermore, real-time monitoring data provides precise status feedback for subsequent process parameter optimization, forming a closed loop in quality control and further enhancing the stability of cutting quality.

[0074] Example 2: Application of Adaptive Decision Algorithm for Multi-Objective Optimization of Laser Process Parameters

[0075] To address the issues of poor parameter adaptability and difficulty in balancing quality, efficiency, and energy consumption in traditional laser cutting, this embodiment innovatively applies a "multi-objective optimization laser process parameter adaptive decision-making algorithm." Through an intelligent optimization model and dynamic adjustment mechanism, it achieves the global optimal matching of process parameters under different operating conditions.

[0076] (I) Construction of Multi-Objective Optimization Model

[0077] The algorithm uses cutting quality (kerf roughness), processing efficiency (cutting speed), and energy consumption (laser power) as core optimization objectives, constructing a multi-objective optimization model. Laser power, cutting speed, auxiliary gas pressure, and focal point position are used as decision variables. Through experimental design and data fitting, correlation functions between each decision variable and the optimization objectives are established: the quality function, based on the quality assessment results of plasma characteristic monitoring, reflects the impact of parameter changes on kerf quality; the efficiency function directly correlates cutting speed and processing cycle; and the energy consumption function quantifies the relationship between parameters and energy consumption by collecting real-time energy consumption data from the laser equipment.

[0078] To achieve multi-objective collaborative optimization, an improved NSGA-III algorithm is adopted for solving the problem. The algorithm integrates the experience of experts in the field of laser cutting, designs adaptive crossover and mutation operators, and dynamically adjusts the operator parameters according to different material properties. For example, for high carbon steel, the mutation probability of laser power parameters is increased to improve the algorithm's adaptability to hard and brittle materials. Crowding ranking and elite retention strategies are introduced to enhance population diversity, avoid the algorithm from getting trapped in local optima, and significantly improve the convergence speed.

[0079] (II) Parameter Sensitivity Analysis and Rapid Response

[0080] A parameter sensitivity analysis mechanism was developed to identify the key parameters that have the greatest impact on cutting quality. The sensitivity coefficient of each parameter to the quality target was calculated using local derivative analysis. For example, high-carbon steel has a high laser power sensitivity coefficient, indicating that even small changes in laser power can significantly affect the cut quality. Aluminum alloys have a high auxiliary gas pressure sensitivity coefficient, requiring close control of this parameter. Based on the priority of sensitivity coefficients, high-sensitivity parameters are optimized first during parameter adjustments to improve adjustment efficiency and accuracy.

[0081] Establish a rapid response model for parameter adjustment to ensure rapid system stabilization after parameter changes: By analyzing the dynamic response process of parameter adjustment, a first-order inertial lag model is constructed to predict the system stabilization time after parameter adjustment; a feedforward-feedback composite control strategy is adopted to compensate for system lag in advance while adjusting parameters, which significantly shortens the system stabilization time after parameter changes and meets the requirements of real-time segmentation.

[0082] (III) Adaptive Adjustment Strategy for Operating Conditions

[0083] The algorithm possesses adaptive adjustment capabilities based on operating conditions: it collects real-time information on material properties (thickness, material composition) and cutting trajectory characteristics (straight lines, curves, corners). When operating conditions change (e.g., switching from carbon steel to aluminum alloy, or from straight-line cutting to corner cutting), it automatically calls upon the corresponding optimized model parameters, performs multi-objective optimization again, and generates a combination of process parameters adapted to the new operating conditions. For example, when cutting thick plates, it appropriately increases the laser power and assist gas pressure while reducing the cutting speed to ensure cutting quality; when cutting corner areas, it fine-tunes the cutting speed and focal point position to avoid burn-out or slag buildup.

[0084] Efficiency Enhancement Principle: The multi-objective optimization algorithm achieves a synergistic improvement in cutting quality, efficiency, and energy consumption. While ensuring a significant reduction in kerf roughness, precise parameter matching results in a significant decrease in energy consumption per unit length of cutting and a marked increase in average cutting speed. Parameter sensitivity analysis and a rapid response mechanism make parameter adjustment more efficient and precise. The adaptive adjustment strategy enhances the system's adaptability to different materials and cutting trajectories, significantly shortening changeover and debugging time and improving the flexibility of the production line.

[0085] Example 3: Application of Focus Drift Prediction and Dynamic Compensation Algorithm

[0086] To address the issue of reduced cutting accuracy caused by laser focus drift due to temperature and vibration, this embodiment innovatively applies a "focus drift prediction and dynamic compensation algorithm." Through a multi-factor prediction model and dual closed-loop compensation control, it achieves real-time and accurate correction of the focus position, ensuring the stability of cutting accuracy.

[0087] (I) Multi-factor drift prediction model

[0088] The algorithm comprehensively considers factors such as the laser head temperature field, vibration signals, and historical drift data to construct a focus drift prediction model that integrates LSTM and attention mechanisms: temperature change data of key components of the laser head (such as resonant cavity and focusing lens) are collected by temperature sensors to analyze the influence of temperature gradient on focus position; vibration signals generated by robot motion and equipment operation are obtained using vibration sensors to extract vibration frequency and amplitude features; and the focus drift amount and corresponding working parameters in the historical cutting process are combined to form the training dataset for the prediction model.

[0089] The model employs an LSTM network to capture long-term dependencies in time-series data and incorporates an attention mechanism to dynamically allocate weights to different influencing factors. For example, under high-power cutting conditions, the weight of temperature is increased; under high-speed cutting conditions, the weight of vibration is increased, making the prediction results more closely match actual working conditions. Through model training and optimization, accurate prediction of focus drift can be achieved within 10 seconds, with prediction errors controlled within a very small range.

[0090] (II) Dual Closed-Loop Compensation Control Strategy

[0091] A dual-loop compensation control strategy combining outer-loop predictive pre-compensation and inner-loop real-time fine compensation is designed: the outer loop pre-compensates for the laser head's Z-axis position based on the focus drift output from the predictive model, offsetting most of the foreseeable drift error; the inner loop measures the deviation between the actual and target focus positions in real time using a laser displacement sensor, inputs the deviation signal to the PID controller, and outputs the fine compensation value for secondary correction of the focus position. This dual-loop control achieves full-process compensation from prediction to pre-compensation to real-time correction, significantly improving the control accuracy of the focus position.

[0092] A lightweight compensation actuator was developed to ensure rapid compensation response: The actuator adopts a combination of a high-precision servo motor and a ball screw, featuring a compact structure, high transmission efficiency, extremely short response time, and a compensation range covering ±0.5mm, meeting the focus compensation requirements under various working conditions. The actuator is seamlessly integrated with the robot control system, enabling rapid issuance and execution of compensation commands.

[0093] Efficiency Enhancement Principle: The focus drift prediction and dynamic compensation algorithm significantly improves the precision of focus position control, effectively avoiding cutting quality fluctuations caused by focus drift and significantly improving dimensional consistency in mass production. Precise compensation reduces overload of the laser head due to focus deviation, extends the service life of core components, and lowers equipment maintenance costs. Simultaneously, a stable focus position provides a reliable foundation for subsequent process parameter optimization, further enhancing overall cutting quality.

[0094] Example 4: Application of energy efficiency optimization algorithm for cutting path

[0095] To address the problems of excessive idle travel, high energy consumption, and significant equipment wear in traditional laser cutting paths, this embodiment innovatively applies a "cutting path energy efficiency optimization algorithm." By integrating energy consumption factors into path planning and intelligent corner handling, it achieves high efficiency, energy saving, and low loss in the cutting process.

[0096] (I) Path planning that integrates energy consumption factors

[0097] Based on traditional path planning, the algorithm introduces an energy consumption factor to construct an improved TSP algorithm: First, by analyzing the length, material properties, and cutting parameters of different cutting path segments, the energy consumption weight per unit length is calculated. For example, the energy consumption weight of a cutting segment for thick plate material is greater than that of a thin plate segment, and the energy consumption weight of a cutting segment for curves is greater than that of a straight line segment. Then, the starting point and ending point of the cutting on the part contour are regarded as nodes. With "lowest total energy consumption + shortest total path" as the optimization objective, the optimal cutting path sequence is solved by the TSP algorithm to minimize the repeated round trips of high-energy-consumption paths.

[0098] A common-edge cutting identification and merging algorithm was developed to further optimize the cutting path: By analyzing the CAD model of the part, cutting patterns with common edges are identified, and the common edge of these patterns is used as a single cutting path to avoid repeated cutting. For example, for multiple adjacent rectangular parts, the cutting paths of their common edges are merged, reducing the number of cuts and the path length. The common-edge cutting strategy can reduce repeated cutting paths, significantly reducing energy consumption and processing time.

[0099] (II) Adaptive Cornering Velocity Planning

[0100] To address the issue of increased energy consumption and equipment wear caused by sudden speed changes in cutting corner areas, an adaptive corner speed planning strategy is designed: by analyzing the corner angle, the deceleration ratio of the cutting speed is dynamically adjusted. For example, when the corner angle is small (e.g., acute angle), the cutting direction changes drastically, so a larger deceleration ratio is used to reduce the cutting speed at the corner and avoid inertial impact; when the corner angle is large (e.g., obtuse angle), the deceleration ratio is appropriately reduced to minimize speed loss while ensuring cutting quality. The speed adjustment process uses a smooth transition curve to avoid sudden speed changes and achieve stable cutting in corner areas.

[0101] In addition, the algorithm also has an idle travel optimization function: by analyzing the idle travel segments (non-cutting movements) in the cutting path, it optimizes the robot's movement speed and acceleration, and runs at an economical speed without affecting cutting efficiency, thereby reducing idle travel energy consumption; at the same time, it plans the shortest idle travel path to avoid invalid movement and further improve the overall processing efficiency.

[0102] (III) Qualitative Explanation of Algorithm Enhancement

[0103] The cutting path energy efficiency optimization algorithm significantly reduces idle travel time and shortens the average processing cycle per piece by minimizing idle travel and repeated cutting paths. The path planning incorporating energy consumption factors and an adaptive corner speed strategy reduces energy consumption during the cutting process, while also reducing the number of robot abrupt stops and starts, decreasing joint wear, and lowering equipment maintenance costs. This algorithm not only improves processing efficiency and energy utilization efficiency but also extends equipment lifespan, bringing significant economic and social benefits to enterprises.

[0104] In summary, through the synergistic application of the four innovative algorithms mentioned above, a complete intelligent monitoring and adaptive control system for laser cutting process parameters has been constructed, realizing intelligent control of the entire cutting process from status monitoring, parameter optimization, focus compensation to path planning. This system effectively solves the technical bottlenecks in traditional laser cutting, such as lagging quality inspection, poor parameter adaptability, focus drift affecting accuracy, and low energy efficiency. It significantly improves cutting quality and production efficiency, reduces energy consumption and equipment maintenance costs, enhances the flexibility of the production line, and provides a practical technical solution for the intelligent upgrading of high-end laser cutting manufacturing.

Claims

1. A method for intelligent monitoring and adaptive control of process parameters, characterized in that, include: Collect plasma spectral signals, cut images, laser energy parameters, robot motion parameters, and material characteristic data during the cutting process; By using a real-time monitoring algorithm for cutting status based on plasma characteristics, online assessment of cutting quality and prediction of defects can be achieved. Based on a multi-objective optimization adaptive decision-making algorithm for laser process parameters, the laser power, cutting speed and other parameters are dynamically adjusted according to the cutting state. The focus position is corrected by a focus drift prediction and dynamic compensation algorithm, and the optimal cutting trajectory is generated by combining the cutting path energy efficiency optimization algorithm. Based on historical data and real-time feedback, the algorithm model is continuously iterated and optimized to form a closed-loop control that synergistically improves quality, efficiency, and energy consumption.

2. The method according to claim 1, characterized in that, The plasma-based real-time monitoring algorithm for cutting status acquires plasma spectral signals in the 300-800nm ​​band, extracts characteristic parameters such as electron density, excitation temperature, and spectral intensity ratio, and uses a lightweight CNN model to evaluate the cutting status with an evaluation delay of ≤10ms.

3. The method according to claim 1, characterized in that, The plasma-based real-time monitoring algorithm for cutting status includes a cutting defect prediction mechanism, which can identify potential defects such as slag buildup and burn-off 50ms in advance, with a defect identification accuracy of ≥97%.

4. The method according to claim 1, characterized in that, The multi-objective optimization adaptive decision-making algorithm for laser process parameters adopts an improved NSGA-III algorithm, with cutting quality, processing efficiency and energy consumption as optimization objectives, and the parameter adjustment response time is ≤20ms.

5. The method according to claim 1, characterized in that, The multi-objective optimized adaptive decision-making algorithm for laser process parameters includes a parameter sensitivity analysis mechanism, which dynamically adjusts the parameter optimization weights for different materials, reducing energy consumption per unit length of cutting by 18-25%.

6. The method according to claim 1, characterized in that, The focus drift prediction and dynamic compensation algorithm integrates LSTM and attention mechanisms, achieving a focus drift prediction error of ≤0.01mm within 10 seconds. Employing a dual closed-loop compensation control strategy, the focus position control accuracy reaches [percentage missing]. ±0.01mm.

7. The method according to claim 1, characterized in that, The focus drift prediction and dynamic compensation algorithm includes a lightweight compensation actuator with a response time of ≤5ms, a compensation range of ±0.5mm, and a 40% improvement in cutting size accuracy.

8. The method according to claim 1, characterized in that, The cutting path energy efficiency optimization algorithm integrates the improved TSP algorithm with energy consumption factor, introduces energy consumption weight per unit length, and reduces idle travel time by 20-30%.

9. The method according to claim 1, characterized in that, The energy efficiency optimization algorithm for the cutting path includes an adaptive corner speed planning strategy and a common edge cutting identification and merging algorithm, which reduces overall energy consumption by 12-18% and reduces repeated cutting paths by 15-25%.

10. The method according to claim 1, characterized in that, The data collected in step 1) is transmitted in real time via industrial Ethernet with a transmission delay of ≤1ms. A laser cutting process database is established to store historical process parameters, sensor data, quality inspection results and energy consumption indicators.

Images