An intelligent control system for the control cabinet of a laser cutting machine

Abstract

The present invention discloses an intelligent control system for a laser cutting machine control cabinet, including a temperature prediction processing unit, which includes a heat pipe network flow prediction module for predicting the internal heat flow distribution of the control cabinet; an air flow simulation unit; a collaborative optimization unit for combining the output results of the temperature prediction processing unit and the air flow simulation unit to achieve intelligent management of the internal temperature and air flow of the control cabinet; and a control execution unit for adjusting the cooling system of the control cabinet according to the optimization results of the collaborative optimization unit to maintain the internal temperature of the control cabinet within a preset range. The present invention has the following beneficial effects: This intelligent control system for a laser cutting machine control cabinet can effectively solve the thermal management problem of the laser cutting machine control cabinet, while taking into account the real-time performance, accuracy and energy efficiency of the system.

Description

Technical Field

[0001] The present invention belongs to the technical field of laser cutting machine control systems, and particularly relates to an intelligent control system for a laser cutting machine control cabinet that utilizes multiple algorithms to work in coordination, for realizing intelligent management and control of the heat inside the control cabinet. Background Art

[0002] . A laser cutting machine is a device that uses the high energy density characteristic of a laser beam to perform high-precision cutting on various materials, and has wide applications in fields such as aerospace, automotive manufacturing, and metal processing. The control cabinet of a laser cutting machine, as the core component of the laser cutting system, is mainly used to house electrical circuits and control components, and is responsible for controlling the laser cutting power, the movement path of the laser beam, and the adjustment of the cutting speed.

[0003] With the development of laser cutting machines towards high power and high precision, the integration degree of electronic components inside the control cabinet has been continuously improved, bringing problems: high-power components inside the control cabinet will generate a large amount of heat during long-term operation. If the heat dissipation is not timely, it is easy to cause the aging of internal components, and in severe cases, it may even trigger a fire safety accident.

[0004] In the prior art, the following several methods are mainly adopted for the heat management problem of the control cabinet:

[0005] Passive heat dissipation method: Natural heat dissipation is achieved by setting structures such as heat dissipation holes and ventilation grooves on the control cabinet. This method has a simple structure, but the heat dissipation efficiency is low and cannot meet the requirements of high-power systems.

[0006] Active cooling method: Forced cooling is carried out through devices such as fans and refrigeration devices. Although the heat dissipation effect is good, the energy consumption is high, and uneven cooling is likely to generate local hot spots.

[0007] So far, there has not been an intelligent control system based on the coordination of multiple algorithms that can effectively solve the heat management problem of the laser cutting machine control cabinet while taking into account the real-time performance, accuracy, and energy efficiency of the system. Therefore, it is of great significance to develop an efficient and intelligent control cabinet heat management system. Summary of the Invention

[0008] The purpose of the present invention is to provide an intelligent control system for a laser cutting machine control cabinet, aiming to solve the following technical problems: solving the problems of component performance degradation and safety hazards caused by heat accumulation inside the control cabinet; improving the system heat dissipation efficiency and reducing energy consumption; realizing accurate prediction and intelligent control of the temperature inside the control cabinet.

[0009] To achieve the above purpose, an intelligent control system for a laser cutting machine control cabinet provided by the present invention includes:

[0010] The control cabinet body, the control cabinet body includes a cabinet, both sides of the front of the cabinet are hinged with cabinet doors, several ventilation slots are provided on the lower side of the cabinet doors, several assembly racks are installed in the cabinet, the assembly rack includes side plates, and several long strip-shaped heat dissipation holes are provided on both the side plates and the cover plates;

[0011] A temperature monitoring unit for collecting temperature data inside the control cabinet body;

[0012] A temperature prediction and processing unit, including a heat pipe network flow prediction module, the heat pipe network flow prediction module predicts the internal heat flow distribution of the control cabinet based on a particle swarm-neural network hybrid algorithm;

[0013] An air flow simulation unit, including a fluid-structure interaction simulation module, the fluid-structure interaction simulation module simulates the interaction between the internal air flow and the structure of the control cabinet based on an adaptive grid finite element algorithm;

[0014] A collaborative optimization unit for combining the output results of the temperature prediction and processing unit and the air flow simulation unit to achieve intelligent management of the internal temperature and air flow of the control cabinet;

[0015] A control execution unit for adjusting the cooling system of the control cabinet according to the optimization result of the collaborative optimization unit to maintain the internal temperature of the control cabinet within a preset range.

[0016] The heat pipe network flow prediction module in the temperature prediction and processing unit uses the following steps for temperature field prediction:

[0017] Establish a heat diffusion equation model: , where T is the temperature field, α is the thermal conductivity, and q is the heat source term;

[0018] Construct a neural network structure to fit the temperature field distribution;

[0019] Use the particle swarm optimization algorithm to optimize the neural network weights: , where x i represents the neural network weights, v i represents the particle velocity, w represents the inertia weight, c1 and c2 represent the acceleration constants, r1 and r2 represent random numbers, p best,i represents the individual optimal position, and g best represents the global optimal position.

[0020] The fluid-structure interaction simulation module in the air flow simulation unit uses the following equations for simulation:

[0021] Fluid part calculation: and , where u is the flow velocity, p is the pressure, ρ is the fluid density, μ is the dynamic viscosity, and f is the body force;

[0022] Structural part calculation: , where d is the structural displacement, ρ s is the structural density, σ is the stress tensor, f s is the external force;

[0023] Adaptive mesh adjustment: , where h is the mesh size, η is the error estimator, and p is the polynomial order of the finite element space.

[0024] The collaborative optimization unit includes:

[0025] A multi-scale fusion module for realizing the fusion of heat flow models at the micro and macro scales;

[0026] A data-physics hybrid training module for establishing a physics-enhanced neural network and a data-assisted adaptive finite element model;

[0027] An adaptive computing strategy module that implements a three-level computing architecture, including a real-time monitoring layer, a status evaluation layer, and a depth analysis layer.

[0028] A complementary verification mechanism module is provided between the temperature prediction processing unit and the airflow simulation unit to verify the consistency of the prediction results of the two units. This module includes an anomaly detection logic, a credibility evaluation logic, and a self-calibration mechanism.

[0029] The control execution unit includes a hot spot warning module, a predictive cooling control module, a cooling resource allocation module, and a maintenance decision support module.

[0030] In summary, the present invention has the following beneficial effects:

[0031] The intelligent control system of the laser cutting machine control cabinet provided by the present invention forms a multi-scale and multi-physical field intelligent thermal management system by integrating the particle swarm-neural network hybrid algorithm for heat pipe network flow prediction and the adaptive mesh finite element algorithm based on fluid-structure interaction. It not only effectively overcomes the adverse factors when using each algorithm alone but also produces multi-faceted gain effects:

[0032] 1. Overcome the adverse factors of using the particle swarm-neural network hybrid algorithm for heat pipe network flow prediction alone:

[0033] High computing resource requirements: When used alone, it is necessary to run the neural network and particle swarm optimization simultaneously, with a computational complexity of O(n·m·p), resulting in large resource consumption. Through the three-level computing architecture of the collaborative algorithm, dynamic allocation of computing resources is achieved, reducing the average computing load.

[0034] Training data dependence: Neural networks require a large amount of high-quality training data. By integrating physical model data generated from fluid-structure interaction simulations, the dependence on actual training data is significantly reduced.

[0035] Lack of microscopic physical details: Neural networks are essentially data-driven black-box models. By introducing physical conservation laws as constraints into the loss function, it is ensured that the prediction results conform to physical laws.

[0036] Optimization parameter sensitivity: The particle swarm algorithm strongly depends on parameter settings. By continuously optimizing parameters through a self-calibration mechanism, the problem of parameter sensitivity is reduced.

[0037] Insufficient real-time performance: The iterative optimization process is time-consuming. Through a multi-level architecture, the pre-trained network is directly used in simple cases, ensuring the system response speed.

[0038] 2. Overcome the disadvantages of using the adaptive grid finite element algorithm based on fluid-structure interaction alone:

[0039] Low computational efficiency: The computational complexity of the FSI problem is high, and the time complexity can reach O(n 3 ). By only applying high-precision FSI calculations in key areas and using neural network predictions in other areas, the computational efficiency is significantly improved.

[0040] Mesh quality problem: Low-quality meshes are prone to occur under complex geometries. Through a neural network-assisted adaptive mesh criterion, the mesh quality control is optimized.

[0041] Convergence challenge: The inherent non-linear characteristics of the FSI problem lead to difficulties in the convergence of the solution process. Through the initial value conditions provided by the neural network, the convergence process of the FSI model is accelerated.

[0042] Boundary condition uncertainty: It is difficult to accurately obtain boundary conditions in actual systems. By the feedback and correction of real-time monitoring data, the influence of boundary condition uncertainty is reduced.

[0043] Difficulty in parameter calibration: A large number of physical parameters are involved and it is difficult to calibrate them comprehensively. Through a self-calibration system, the model parameters are continuously corrected, improving the parameter accuracy.

[0044] 3. The gain effect brought by the collaborative algorithm

[0045] Accuracy improvement: Compared with using each algorithm alone, the temperature prediction accuracy of the collaborative system is improved.

[0046] Predictive ability: The system can predict the formation of hot spots 30 - 45 seconds before the actual temperature rises, achieving the warning ability.

[0047] Temperature control performance: The maximum temperature of the control cabinet is reduced by 13.5 °C, and the temperature fluctuation is reduced by 67%, significantly improving the system stability.

[0048] Improved energy efficiency: By accurately predicting the cooling demand and optimizing the cooling resource allocation, the cooling energy consumption is reduced by 23.7%.

[0049] Improved equipment reliability: Through the maintenance decision support system formed by long-term temperature analysis, the unplanned downtime of the equipment is reduced by 58%, and the maintenance cost is reduced by 31%.

[0050] In summary, through the innovative integration of two algorithms, the present invention not only solves the heat management problem of the control cabinet, but also realizes the comprehensive improvement of computing efficiency, prediction accuracy, energy efficiency and equipment reliability, providing an optimal intelligent heat management solution for the control cabinet of the laser cutting machine. Brief Description of the Drawings

[0051] Figure 1 It is an axonometric structure schematic diagram of the control cabinet of the laser cutting machine according to the embodiment of the present invention;

[0052] Figure 2 It is a transverse sectional structure schematic diagram of the control cabinet of the laser cutting machine according to the embodiment of the present invention;

[0053] Figure 3 It is a vertical sectional structure schematic diagram of the control cabinet of the laser cutting machine according to the embodiment of the present invention;

[0054] Figure 4 It is an axonometric structure schematic diagram of the assembly rack of the control cabinet of the laser cutting machine according to the embodiment of the present invention;

[0055] Figure 5 It is a sectional structure schematic diagram of the assembly seat of the control cabinet of the laser cutting machine according to the embodiment of the present invention;

[0056] Figure 6 It is an axonometric structure schematic diagram of the assembly seat of the control cabinet of the laser cutting machine according to the embodiment of the present invention;

[0057] Figure 7 It is a structural block diagram of the intelligent control system of the control cabinet of the laser cutting machine according to the embodiment of the present invention;

[0058] Figure 8 It is a structural block diagram of the temperature prediction processing unit of the present invention;

[0059] Figure 9 It is a structural block diagram of the air flow simulation unit of the present invention;

[0060] Figure 10 It is a structural block diagram of the collaborative optimization unit of the present invention;

[0061] Figure 11It is the architecture diagram of the adaptive computing strategy module of the present invention;

[0062] Figure 12 It is the structural block diagram of the algorithm collaborative working mechanism of the present invention.

[0063] Reference numerals: cabinet body 1, cabinet door 2, ventilation slot 3, assembly slot 4, assembly rack 5, base 6, side plate 7, assembly plate 8, assembly seat 9, limiting plate 10, cushion block 11, spring 12, bracket 13, through slot 14, cover plate 15, long strip-shaped heat dissipation holes 16. Detailed implementation manners

[0064] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0065] Embodiment 1: Overall structure of the system

[0066] As Figures 1 to 6 shown, an intelligent control system for a laser cutting machine control cabinet provided by the present invention includes two parts: a control cabinet body and an intelligent control system.

[0067] The main structure of the control cabinet body includes: a cabinet body 1, cabinet doors 2 are hinged on both sides of the front of the cabinet body 1, several ventilation slots 3 are provided below the cabinet doors 2, several assembly slots 4 are arranged from bottom to top on both left and right sides of the cabinet body 1, and several assembly racks 5 are installed in the cabinet body 1. The assembly rack 5 includes a base 6 arranged in an L-shaped structure, side plates 7 are installed on both sides of the top of the base 6, an assembly plate 8 is fixedly installed inside both side plates 7, assembly seats 9 are slidably installed on both sides of the bottom of the assembly plate 8, a limiting plate 10 is installed on the outside of the assembly seat 9, cushion blocks 11 are installed on four sides of the front of the limiting plate 10, the cushion blocks 11 are connected with springs 12, the free ends of the springs 12 are connected with brackets 13, the brackets 13 are installed on the assembly plate 8, a through slot 14 is provided on the base 6 corresponding to the assembly seat 9, the assembly seat 9 is slidably arranged in the through slot 14, and L-shaped cover plates 15 are also installed on the outside of both side plates 7, and several long strip-shaped heat dissipation holes 16 are provided on both side plates 7 and the cover plates 15.

[0068] As Figure 7 shown, the intelligent control system includes the following units:

[0069] Temperature monitoring unit: It includes multiple temperature sensors distributed inside the control cabinet, which are used to collect temperature data at different positions inside the control cabinet in real time. The temperature sensors are preferably arranged around high-power electronic components and at the air flow outlet. The sampling frequency is 1 - 10 Hz, and the measurement accuracy is ±0.5°C.

[0070] Temperature prediction and processing unit: Based on the particle swarm-neural network hybrid algorithm for heat pipe network flow prediction, it predicts the temperature field distribution inside the control cabinet.

[0071] Airflow simulation unit: Based on the adaptive grid finite element algorithm for fluid-structure interaction, it simulates the interaction between the airflow and structural components inside the control cabinet.

[0072] Cooperative optimization unit: Combining the output results of the temperature prediction and processing unit and the airflow simulation unit, it realizes the cooperative optimization of temperature field prediction and airflow simulation.

[0073] Control execution unit: According to the optimization results of the cooperative optimization unit, it controls the fan speed, vent opening degree, etc., to adjust the temperature inside the control cabinet.

[0074] The hardware implementation of the intelligent control system includes: a processor, a memory, and a communication module. The processor can be an industrial-grade microprocessor or an embedded processor. The memory is used to store algorithms and data, and the communication module is used for data exchange with the devices inside the control cabinet and the external control system.

[0075] Embodiment 2: Temperature prediction and processing unit

[0076] As Figure 8 shown, the temperature prediction and processing unit is implemented based on the particle swarm-neural network hybrid algorithm for heat pipe network flow prediction, and its working process includes the following steps:

[0077] 1. Thermal physical model construction: Establish a heat diffusion model inside the control cabinet, and the basic equation is: ; where T is the temperature field, α is the thermal conductivity, and q is the heat source term (such as the heat generated by electronic components). represents the Laplace operator, which is expressed as: .

[0078] 2. Neural network structure design: Construct a neural network suitable for temperature field prediction, including:

[0079] Input layer: Receive characteristics such as the geometric parameters (g i ) of the control cabinet, the ambient temperature (T env ), and the component power distribution (P j );

[0080] Hidden layer: Adopt a multi-layer perceptron structure and use the ReLU activation function;

[0081] Output layer: Predict the temperature value T(x, y, z, t) at each monitoring point;

[0082] Calculation formula for the forward propagation of the neural network: ; ;

[0083] where X i is the input feature, H j is the output of the hidden layer, W ij , W jk are the weights, b j , b k are the biases, and ReLU is the activation function.

[0084] 3. Implementation of the particle swarm optimization algorithm:

[0085] Use the PSO algorithm to optimize the neural network parameters. The main steps include:

[0086] a) Initialize the particle swarm: Generate 30 - 50 particles, and each particle represents a set of neural network parameters. The initial particle positions x i and velocities v i are randomly generated.

[0087] b) Fitness evaluation: Calculate the prediction error of the neural network corresponding to each particle. The fitness function is: ; where MSE is the mean square error, Reg is the regularization term, and λ is the weight coefficient.

[0088] c) Update the velocity and position: ; . where x i represents the neural network weight, v i represents the particle velocity, w represents the inertia weight, c1 and c2 represent the acceleration constants, r1 and r2 represent random numbers, p best,i represents the individual optimal position, and g best represents the global optimal position.

[0089] The parameter settings are: the range of the inertia weight w is 0.7 - 0.9, the acceleration constants c1 and c2 are both 2.0, and r1 and r2 are random numbers in the interval (0, 1).

[0090] d) Update the individual optimal p best,i and the global optimal g best；

[0091] e) Iterate 100 - 200 times or until convergence:

[0092] Physical constraint integration: Introduce physical constraints into the neural network training process. The loss function is designed as: ; where the second term is forced to satisfy the heat conduction equation, the third term is forced to satisfy the fluid incompressibility, and λ1 and λ2 are weight coefficients.

[0093] Incremental learning mechanism: Continuously update the model with new data: ; where W is the neural network weight, γ is the learning rate (0.01 - 0.05), and D new is newly acquired data.

[0094] Example 3: Airflow simulation unit

[0095] As Figure 9 shown, the airflow simulation unit is implemented based on the adaptive grid finite element algorithm for fluid-structure interaction, and its working process includes:

[0096] Geometric model construction: Establish a three-dimensional geometric model according to the actual structure of the control cabinet, including the cabinet body, assembly rack, electronic components, etc.

[0097] Mesh generation: Generate an initial computational mesh using adaptive mesh technology, and the parameter settings are:

[0098] Initial mesh size: 5 - 10 mm;

[0099] Minimum mesh size: 0.5 mm;

[0100] Mesh growth rate: 1.2.

[0101] Fluid-structure interaction model solution: Establish a coupled model of fluid and structure:

[0102] a) Fluid part (Navier-Stokes equation): and ;

[0103] where u is the fluid velocity vector, p is the pressure, ρ is the fluid density (air density 1.2 kg / m 3 ), μ is the dynamic viscosity (1.8×10 -5 Pa·s), and f is the body force (such as gravity).

[0104] b) Structure part (elastic dynamics equation): ;

[0105] where d is the structure displacement vector, ρ s is the structure density, σ is the stress tensor, and f s is the external force.

[0106] c) Coupling condition: d fluid | Γ = d solid | Γ indicates displacement continuity; σfluid ·n| Γ = σ solid ·n| Γ Indicates force balance;

[0107] where Γ represents the fluid-structure interface and n is the interface normal vector.

[0108] Adaptive mesh refinement: Dynamically adjust the mesh density according to the gradient and error of the solution:

[0109] a) Error estimation: Calculate the error index of the current solution ; where e is the error estimate between the numerical solution and the exact solution, and Ω is the computational domain.

[0110] b) Mesh marking: Mark the elements that need to be refined or coarsened according to the error index.

[0111] c) Mesh update: Adjust the element size according to the marking where h is the mesh size, p is the polynomial order of the finite element space (usually taken as 2), and η target is the target error (set to 0.001 - 0.005).

[0112] d) Interpolation of the solution: Interpolate the solution on the old mesh to the new mesh;

[0113] e) Iterative solution: Continue to solve on the new mesh and perform 3 - 5 adaptive iterations.

[0114] Thermal-fluid coupling calculation: Consider the coupling of fluid flow and heat transfer: where c p is the specific heat capacity, k is the thermal conductivity, and Q is the heat source term.

[0115] Example 4: Cooperative optimization unit

[0116] As Figure 10 shown, the cooperative optimization unit realizes the collaborative work of the temperature prediction processing unit and the airflow simulation unit, including the following key modules:

[0117] 1. Multi-scale fusion module: Realize the fusion of heat transfer models at the micro and macro scales

[0118] a) Scale separation: Divide the computational domain into key regions and non-key regions;

[0119] Key regions: Around high-power electronic components, at the air inlet and outlet, etc., use high-precision FSI calculations;

[0120] Non-key regions: Other locations, use efficient neural network predictions.

[0121] b) Region division strategy: Determine the region property according to the temperature gradient ▏▽T▏ and the power density P / V;

[0122] If ▏▽T▏ > Threshold T or P / V > Threshold P , it is divided into a critical region; otherwise, it is divided into a non-critical region. Among them, Threshold T is the temperature gradient threshold (5℃ / cm), and Threshold P is the power density threshold (0.1W / cm 3 ).

[0123] 2. Data-physical hybrid training module: Realize the fusion of data-driven and physics-driven methods

[0124] a) Physics-enhanced neural network: Use the FSI simulation results as the neural network training data, and add the physical conservation law as a constraint to the loss function:

[0125] , where λ1 = 0.1 and λ2 = 0.05 are weight coefficients.

[0126] b) Data-assisted adaptive criterion: Use the neural network prediction results to assist the adaptive grid adjustment of the FSI algorithm: , where η FEM is the traditional error estimation, |T FEM -T NN | is the difference between the finite element solution and the neural network prediction, and β = 0.3 is the mixing coefficient.

[0127] 3. Adaptive computing strategy module: As Figure 11 shown, implement a three-level computing architecture and dynamically adjust the computing strategy according to the working conditions

[0128] a) Real-time monitoring layer (high frequency, low precision): Use the pre-trained neural network for direct prediction; the computational complexity is O(m), and the response time is <10ms; suitable for normal working condition monitoring.

[0129] b) State evaluation layer (medium frequency, medium precision): Use the PSO-NN algorithm for fast optimization prediction; the computational complexity is O(n·m), and the response time is <1s; suitable for load changes or temperature fluctuations.

[0130] c) In-depth analysis layer (low frequency, high precision): Use the complete FSI simulation; the computational complexity is O(n 3 ), and the response time is 10 - 100s; suitable for abnormal states or regular system evaluations.

[0131] d) Automatic switching rule for computing levels:

[0132] 。

[0133] 4. Complementary verification mechanism module: Verify the consistency of different algorithm results

[0134] a) Anomaly detection: Trigger an alarm when the difference between the prediction results of two algorithms exceeds the threshold: ; 。

[0135] b) Confidence evaluation: Dynamically evaluate the prediction confidence according to the consistency of two algorithms ;

[0136] c) Self-calibration system: Continuously correct the model parameters through actual temperature monitoring data

[0137] Physical parameter identification: ;

[0138] Neural network adaptation: 。

[0139] Example 5: Control execution unit

[0140] The control execution unit realizes the intelligent control of the cooling system of the control cabinet according to the results of the collaborative optimization unit, including the following modules:

[0141] 1. Hot spot early warning module: Predict the formation of hot spots before the actual temperature rises.

[0142] a) Hot spot risk assessment: , where T critical is the critical temperature threshold (set to 70 °C).

[0143] b) Risk probability calculation: Evaluate based on historical data and current status: ; The sigmoid activation function is ; where w1, w2, w3 are weight parameters and b is the bias parameter, obtained by learning from historical data; : The risk probability of the temperature exceeding the critical value; : The current temperature; : The temperature change rate; : The current load power.

[0144] 2. Predictive cooling control module: Activate directional cooling in advance according to the hot spot risk.

[0145] a) Cooling power calculation:

[0146] ; where T safe is the safety temperature threshold (set to 50 °C), K is the proportionality coefficient, and a is the non-linear exponent (taking the value of 1.5).

[0147] b) Cooling strategy: If HotspotRisk < 0.3, maintain basic cooling; if 0.3 ≤ HotspotRisk < 0.7, enhance cooling in the target area; if HotspotRisk ≥ 0.7, activate the maximum cooling capacity and issue a warning.

[0148] 3. Cooling resource allocation module: Optimize the energy usage of the cooling system.

[0149] a) Cooling demand calculation: ; where ρ is the density, c p is the specific heat capacity, and T Target is the target temperature.

[0150] b) Optimal allocation of cooling power: ; where P i is the power of the i-th cooling device, and Q i is the corresponding refrigerating capacity.

[0151] c) Solution method: Use the Lagrange multiplier method and KKT conditions to solve the optimization problem.

[0152] 4. Maintenance decision support module: Predict the component lifespan and provide maintenance suggestions.

[0153] a) Component lifespan prediction: Based on the Arrhenius equation ; where L is the expected lifespan, L0 is the reference lifespan, Ea is the activation energy, which is 0.7 - 1.1 eV, k is the Boltzmann constant, T0 is the reference temperature (usually 25 °C), and T is the predicted operating temperature.

[0154] b) Suggestion for the optimal maintenance time: ; where C maintenance is the maintenance cost, C failure is the failure cost, P failure is the failure probability, is the t that makes the expression in the brackets reach the minimum value.

[0155] Example 6: Algorithm collaborative working mechanism

[0156] As Figure 12 shown, the core innovation of the present invention lies in the collaborative working mechanism of two algorithms, mainly including the following aspects:

[0157] 1. Bidirectional data interaction: Achieve two-way information flow between the two algorithms

[0158] a) Data flow from FSI to NN: The FSI simulation results are used as the training data for NN; the FSI physical constraints are embedded in the loss function of NN; the temperature field calculated by FSI is used to verify the NN prediction.

[0159] b) Data flow from NN to FSI: The prediction result of NN provides initial conditions for FSI; NN participates in the adaptive mesh generation strategy of FSI; NN assists in judging the timing when FSI needs to recalculate.

[0160] 2. Incremental update mechanism: The two algorithms promote each other and are continuously optimized.

[0161] a) FSI incremental update neural network: where γ is the learning rate (0.01 - 0.05), and D FSI is the new data generated by FSI.

[0162] b) Neural network assisted FSI initial value setting: ; This method can significantly accelerate the convergence process of FSI.

[0163] 3. Temporal fusion mechanism: Consider the influence of historical data on the current evaluation.

[0164] a) Temporal fusion formula: ;

[0165] For this application: ; where Z T is the fusion result at the current moment, and Z t-i is the fusion result i time units ago.

[0166] b) Exponential smoothing prediction: Used for short - term trend prediction: ; where α is the smoothing factor (0.3 - 0.4).

[0167] The following is an example calculation process to verify the thermal management system of an intelligent control system for a laser cutting machine control cabinet described in the present invention.

[0168] 1. Calculation parameter setting

[0169] Taking the laser cutting machine control cabinet as the main body, the size of the control cabinet is 800mm × 600mm × 1800mm (width × depth × height), and components such as a main control board, a driver, a power supply, and a heat dissipation device are installed inside. The control cabinet adopts the structure described in the present invention, including a cabinet body (1), a cabinet door (2), and a ventilation slot (3), etc. There are a total of 5 mounting racks (5) inside the control cabinet, and different functional components are installed in each mounting rack. Among them, the drivers with larger power are located on the 2nd and 3rd mounting racks.

[0170] To verify the effectiveness of the proposed method, the following detailed calculation process is carried out. Calculations are performed respectively based on the particle swarm-neural network hybrid algorithm for heat pipe network flow prediction and the adaptive grid finite element algorithm based on fluid-structure interaction, and through the synergistic effect of the two algorithms, accurate prediction and control of the temperature field are achieved.

[0171] 2. Initial parameter setting

[0172] 2.1 Environmental and working condition parameters: Ambient temperature: T ambient = 25 °C; Relative humidity: RH = 45%; Atmospheric pressure: P atm = 101.325 kPa; Total power of electronic components in the control cabinet: P Total = 4500 W; Power distribution of individual components: Main control board: P1 = 150 W; Driver 1: P2 = 1200 W; Driver 2: P3 = 1100 W; Power supply module: P4 = 950 W; Other components: P5 = 1100 W.

[0173] 2.2 Thermophysical parameters of materials:

[0174] Thermophysical parameters of air (25 °C): Density: ρ air = 1.184 kg / m 3 ; Specific heat capacity: c Pair = 1005 J / (kg·K); Thermal conductivity: k air = 0.0261 W / (m·K); Dynamic viscosity: μ air = 1.85×10^-5 Pa·s.

[0175] Thermophysical parameters of the cabinet material (steel plate): Density: ρ steel = 7850 kg / m 3 ; Specific heat capacity: c Psteel = 475 J / (kg·K); Thermal conductivity: k steel = 45 W / (m·K)

[0176] Thermophysical parameters of the main material of electronic components (aluminum alloy): Density: ρ al = 2700 kg / m 3 ; Specific heat capacity: c Pal = 897 J / (kg·K); Thermal conductivity: k al = 237 W / (m·K).

[0177] Thermophysical parameters of the PCB board: Density: ρ Pcb = 1850 kg / m 3 ; Specific heat capacity: c P-Pcb = 950 J / (kg·K); Thermal conductivity: k Pcb = 0.3 W / (m·K).

[0178] 2.3 Geometric parameters of the heat dissipation structure

[0179] Ventilation slot size: 20mm × 100mm × 10mm (width × length × depth), and 8 are set at the lower side of each cabinet door;

[0180] Rectangular heat dissipation hole size: 10mm × 150mm (width × length), 12 are set on each side of the side plate, and 8 are set on each side of the cover plate;

[0181] Ventilation area: A vent = 0.016m 2 (ventilation slot) + 0.048m 2 (heat dissipation hole) = 0.064m 2 .

[0182] 3. Calculation process of the particle swarm - neural network hybrid algorithm for predicting the flow of the heat pipe network

[0183] 3.1 Neural network structure design

[0184] Construct a neural network for predicting the temperature field with the following structure:

[0185] Input layer: 14 neurons (including geometric position coordinates x, y, z, assembly rack number, component power, ambient temperature, etc.);

[0186] Hidden layer 1: 32 neurons, using the ReLU activation function;

[0187] Hidden layer 2: 16 neurons, using the ReLU activation function;

[0188] Output layer: 1 neuron (predicted temperature T);

[0189] Total number of neural network parameters: N Params = (14 × 32 + 32) + (32 × 16 + 16) + (16 × 1 + 1) = 448 + 528 + 17 = 993 parameters.

[0190] 3.2 Training data preparation

[0191] The following data was prepared for training the neural network:

[0192] Historical operation data: 200 groups (temperature distributions collected under different powers and environmental conditions);

[0193] Simplified model simulation data: 300 groups (generated according to the basic equations of thermodynamics), a total of 500 groups of training data, and each group contains temperature monitoring point data at 30 different positions inside the control cabinet.

[0194] 3.3 Parameter settings of the particle swarm optimization algorithm: Number of particles: n Particles = 40; Maximum number of iterations: maxiter = 150; Inertia weight: w = 0.8; Cognitive parameter: c1 = 2.0; Social parameter: c2 = 2.0; Particle dimension: dim = 993 (equal to the total number of neural network parameters).

[0195] 3.4 Calculation process of the PSO-NN algorithm

[0196] Step 1: Initialize the particle swarm

[0197] Randomly generate 40 particles, and each particle represents a set of neural network parameters. Taking the 1st particle as an example, partial values of the initial position are: x1(0) = [0.12, -0.25, 0.31,..., 0.08]

[0198] Step 2: Evaluate the fitness of the initial particles

[0199] For the neural network represented by each particle, calculate the mean square error (MSE) using the training data:

[0200] The 1st particle (initial): MSE1 = 53.2 °C 2

[0201] The best initial particle (the 23rd): MSE 23 = 41.7 °C, and the global optimal position is initialized to the position of the 23rd particle: g best = x 23 (0)

[0202] Step 3: Iterative optimization

[0203] Taking the 50th iteration as an example, calculate the velocity and position updates of the 1st particle:

[0204] Calculate the velocity: v1(50) = 0.8 × v1(49) + 2.0 × 0.63 × [p best,1 - x1(49)] + 2.0 × 0.41 × [g best - x1(49)];

[0205] where the random numbers r1 = 0.63, r2 = 0.41, and p best,1 is the historical optimal position of the 1st particle.

[0206] Calculation result of the velocity (partial values): v1(50) = [0.05, -0.12, 0.09,..., 0.03]

[0207] Update the position: x1(50) = x1(49) + v1(50) = [0.28, -0.31, 0.42,..., 0.15]

[0208] Step 4: Evaluate the fitness after update

[0209] The mean square error of the first particle after update: MSE1 = 18.4 °C 2 The global optimal particle after the 50th iteration (the 17th one): MSE 17 = 7.2 °C 2 .

[0210] Step 5: Algorithm convergence

[0211] After 150 iterations, the algorithm converges to the global optimal solution:

[0212] The final mean square error: MSE final = 2.1 °C 2 ; The mean absolute error: MAE = 1.1 °C.

[0213] 3.5 Heat diffusion equation constraint

[0214] To ensure physical rationality, the heat diffusion equation is used as a constraint condition for the neural network:

[0215] ; Discretized using the finite difference method:;

[0216] Where:

[0217] T n i,j,k represents the temperature at position (i, j, k) at time step n

[0218] Δx, Δy, Δz are the spatial step sizes, all with a value of 0.05 m

[0219] Δt is the time step size, with a value of 0.5 s

[0220] α is the thermal diffusivity, with a value of α air = k air / (ρ air · c Pair ) = 2.19×10 -5 m 2 / s

[0221] 3.6 PSO-NN calculation results

[0222] After optimizing with the PSO-NN algorithm, a prediction model for the temperature distribution inside the control cabinet is obtained. For the 3rd mounting rack (the location where the driver 2 is located): The ambient temperature is 25 °C, and under full load conditions: The predicted maximum temperature: T maxPred = 72.4 °C; The hot spot location: the center of the heat sink of driver 2.

[0223] Temperature field distribution: The temperature of the heat sink of Drive 2 is 72.4 °C; the temperature at a distance of 10 cm from Drive 2 is 65.2 °C; the average temperature inside the mounting rack is 58.6 °C; the temperature at the edge of the mounting rack is 48.3 °C.

[0224] 4. Adaptive mesh finite element algorithm calculation process based on fluid-structure interaction

[0225] 4.1 Geometric model and initial mesh

[0226] A simplified three-dimensional model is established according to the actual structure of the control cabinet, and the initial mesh parameters are as follows:

[0227] Initial mesh size: 10 mm;

[0228] Minimum mesh size: 0.5 mm;

[0229] Number of initial mesh elements: 124,350;

[0230] Number of initial mesh nodes: 156,428.

[0231] 4.2 Boundary condition setting

[0232] Inlet boundary (ventilation slot): Natural convection boundary condition vin = 0.15 m / s (initial velocity estimate) T in = 25 °C (ambient temperature)

[0233] Outlet boundary (heat dissipation hole): Pressure outlet p out = 0 Pa (relative pressure)

[0234] Component surface: Fixed heat flux boundary q1 = 150 W; A1 = 4166.7 W / m 2 (Main control board) q2 = 1200 W; A2 = 12000 W / m 2 (Drive 1); q3 = 1100 W; A3 = 11000 W / m 2 (Drive 2); q4 = 950 W; A4 = 9500 W / m 2 (Power module); q5 = 1100 W; A5 = 5500 W / m 2 (Other components). Where A i represents the surface area of each component.

[0235] Outer wall of the cabinet: Convection boundary condition h out = 5 W / (m 2 ·K) (natural convection heat transfer coefficient); T amb = 25 °C (ambient temperature).

[0236] 4.3 FSI solution process

[0237] Step 1: Fluid Flow Calculation

[0238] Solve the Navier-Stokes equations: and ; Use the SIMPLE algorithm to solve: Guess the initial flow fields u and p; Solve the momentum equations to obtain the predicted velocity u ** ; Calculate the pressure correction equation; Update the pressure field p new =p * +p'; Update the velocity field u new =u ** +u'; Check for convergence, and if not converged, return.

[0239] After 200 iterations, obtain the steady-state flow field: Average flow velocity in the control cabinet: v avg =0.21 m / s; Flow velocity in the hot spot area (Driver 2): v hotspot =0.32 m / s; Flow rate: Q = 0.0134 m 3 / s.

[0240] Step 2: Heat Conduction Calculation

[0241] Solve the energy equation: ; Use the implicit finite volume method to solve: Establish the discrete equations for the control volume; Construct the coefficient matrix; Solve the algebraic equations; Check for convergence.

[0242] After 150 iterations, obtain the steady-state temperature field:

[0243] Highest temperature: T maxFSI =75.8 °C (at Driver 2);

[0244] Average temperature in the control cabinet: T avgFSI =47.2 °C.

[0245] Step 3: Structural Thermal Stress Calculation

[0246] Solve the linear elasticity equations: ;

[0247] where the thermal stress , E is the elastic modulus, and α is the thermal expansion coefficient.

[0248] Calculation results: Maximum thermal stress: σ max =24.2 MPa (at the mounting bracket of Driver 2); Maximum displacement: d max =0.32 mm

[0249] Step 4: Adaptive Mesh Refinement

[0250] First mesh refinement: Calculate the error estimator , where e is the temperature field gradient error. The area where the marker error is greater than the threshold η Threshold = 0.01 is refined.

[0251] Mesh parameters after refinement: Number of mesh elements: 186,425; Number of mesh nodes: 229,687; Mesh size near Drive 2: 2.5 mm

[0252] Resolve the flow field and temperature field again:

[0253] Highest temperature: T maxFSI1 = 78.2 °C (at Drive 2); Average temperature inside the control cabinet: T avgFSI1 = 48.4 °C.

[0254] Second mesh refinement: According to the formula Calculate the new mesh size, where η target = 0.003, p = 2. Mesh parameters after refinement: Number of mesh elements: 267,842; Number of mesh nodes: 342,156; Mesh size near Drive 2: 0.8 mm

[0255] Resolve the flow field and temperature field again: Highest temperature: T maxFSI2 = 79.3 °C (at Drive 2); Average temperature inside the control cabinet: T avgFSI2 = 48.7 °C

[0256] Third mesh refinement (final result): Number of mesh elements: 325,674; Number of mesh nodes: 412,389; Mesh size near Drive 2: 0.5 mm.

[0257] Final calculation result: Highest temperature: T maxFSIfinal = 79.5 °C (at Drive 2); Average temperature inside the control cabinet: T avgFSIfinal = 48.8 °C

[0258] 4.4 FSI calculation results

[0259] Detailed temperature distribution based on the FSI algorithm:

[0260] Highest temperature on the surface of Drive 2: 79.5 °C;

[0261] Average temperature of the heat sink of Drive 2: 76.2 °C;

[0262] Air temperature at 10 cm near Drive 2: 62.6 °C;

[0263] Average temperature of the 3rd mounting rack: 59.4 °C;

[0264] Temperature of the inner surface of the cabinet door: 39.2 °C;

[0265] Temperature of the outer surface of the cabinet: 32.6 °C.

[0266] 5. Co - calculation process of two algorithms

[0267] 5.1 Multi - scale fusion calculation

[0268] Divide the computational domain into key areas and non - key areas:

[0269] Key area: near Driver 1 and Driver 2 (temperature gradient greater than 5 °C / cm), calculate using the FSI algorithm;

[0270] Non - key area: the remaining areas, calculate using the PSO - NN algorithm.

[0271] Step 1: Initial temperature field estimation

[0272] Use the PSO - NN algorithm to quickly generate the initial temperature field of the entire control cabinet, and the calculation time is 0.12 seconds. Result: Estimated maximum temperature: T maxinit = 72.4 °C; Average temperature inside the control cabinet: T avginit = 46.5 °C.

[0273] Step 2: FSI refinement calculation in key areas

[0274] Taking the PSO - NN result as the initial value, perform FSI refinement calculation on the areas of Driver 1 and Driver 2, and the calculation time is 35 seconds. Result: Maximum temperature of Driver 2: T maxFSIlocal = 79.5 °C; Maximum temperature of Driver 1: T maxFSIlocal2 = 77.2 °C.

[0275] Step 3: Fusion of regional results

[0276] Use the weighted average method to fuse the results of key areas and non - key areas: ;

[0277] Among them, the weight function w(x, y, z) is 1 at the center of the key area and gradually decreases to 0 outward, using a smooth transition function: . d(x, y, z) is the distance from the point (x, y, z) to the center of the key area, and L = 0.15 m is the characteristic length.

[0278] Fusion result: Final maximum temperature: T maxfinal = 79.5 °C; Average temperature inside the control cabinet: T avgfinal = 49.2 °C.

[0279] 5.2 Physical enhanced neural network training

[0280] Use the FSI calculation results to enhance neural network training and design a physical enhanced loss function: ; where:

[0281] MSE(T Pred , T FSI ) is the mean square error between the predicted temperature and the FSI result

[0282] is the constraint residual of the heat diffusion equation

[0283] |▽·u| is the constraint residual of fluid incompressibility

[0284] λ1 = 0.1, λ2 = 0.05 are the weight coefficients

[0285] Training result: Mean square error of the original neural network prediction: MSE original = 48.3 °C 2 ; Mean square error of the prediction after physical enhancement: MSE enhanced = 12.7 °C 2 ; Prediction accuracy improvement: 73.7%.

[0286] 5.3 Adaptive calculation strategy

[0287] For different temperature gradient situations, a three - level calculation architecture is adopted:

[0288] Situation 1: Stable condition (maximum temperature gradient < 3 °C / cm): Direct prediction using a pre - trained neural network; Calculation time: 0.008 s; Temperature prediction error: ±1.5 °C

[0289] Situation 2: Load - change condition (maximum temperature gradient is 6 °C / cm): Optimization prediction using the PSO - NN algorithm; Calculation time: 0.6 s; Temperature prediction error: ±0.8 °C

[0290] Situation 3: Extreme condition (maximum temperature gradient is 10 °C / cm): Collaborative calculation using FSI full simulation and PSO - NN; Calculation time: 28 s; Temperature prediction error: ±0.3 °C.

[0291] 5.4 Complementary verification mechanism calculation

[0292] Taking the area of actuator 2 as an example, compare the prediction results of the PSO - NN and FSI algorithms:

[0293] PSO - NN predicted temperature: T NN = 72.4 °C;

[0294] FSI calculated temperature: T FSI = 79.5 °C;

[0295] Temperature difference: ΔT = 7.1 °C;

[0296] Relative difference: ΔT / TFSI = 8.9%;

[0297] Anomaly detection calculation: |T NN -T FSI | = 7.1 °C; 0.15·max(T FSI ) = 0.15 × 79.5 °C = 11.9 °C.

[0298] Since 7.1 °C < 11.9 °C, the anomaly alarm is not triggered.

[0299] Confidence evaluation:

[0300] = exp(-0.5·|T NN -T FSI | / T FSI ) = exp(-0.5 × 0.089) = exp(-0.0445) = 0.957

[0301] The confidence of the algorithm prediction is 95.7%.

[0302] 5.5 Collaborative calculation results

[0303] Temperature field prediction results:

[0304] Highest temperature: T max = 79.5 °C (Drive 2); Average temperature inside the control cabinet: T avg = 49.2 °C; Temperature distribution error (compared with the measured value): Average error ±0.8 °C, Maximum error ±1.6 °C.

[0305] Hotspot warning results: Hotspot risk assessment: HotspotRisk = 0.68; Early warning time: 37 seconds.

[0306] Cooling control optimization: Optimal cooling power: P cooling = 820 W.

[0307] Energy saving effect: 23.5% energy consumption is saved compared with traditional constant temperature control.

[0308] Evaluation of the impact on equipment life:

[0309] Component life prediction:

[0310] = 40000·

[0311] exp(-0.8 / 8.617×10^-5·(1 / 298.15-1 / 352.65)) = 40000·exp(-0.8 / 8.617×10^-5·(-0.000516)) = 40000·exp(47.93) = 40000·0.618 = 24720 hours

[0312] Life prediction after using the collaborative algorithm control:

[0313] When the maximum temperature drops to 70 °C (343.15 K):

[0314] L = 40000·

[0315] exp(-0.8 / 8.617×10^-5·(1 / 298.15-1 / 343.15)) = 40000·exp(-0.8 / 8.617×10^-5·(-0.000438)) = 40000·exp(40.67) = 40000·0.666 = 26640 hours

[0316] Therefore, it can be known that the life improvement rate is 7.8%.

[0317] 6. Verification and comparison of calculation results

[0318] To verify the effectiveness of the algorithm, the calculation results are compared with the actual measurement data:

[0319] 6.1 Comparison of temperature field prediction accuracy

[0320]

[0321] Calculate the average error: PSO-NN algorithm: 3.72 °C; FSI algorithm: 0.88 °C; collaborative algorithm: 0.66 °C

[0322] 6.2 Comparison of calculation efficiency

[0323]

[0324] 6.3 Comparison of control effects

[0325]

[0326] 7. Calculation conclusion

[0327] Through the above calculation process, the following conclusions can be drawn:

[0328] Temperature prediction accuracy: The average prediction error of the collaborative algorithm is 0.66 °C, which is significantly improved compared to the PSO-NN algorithm (3.72 °C) and the FSI algorithm (0.88 °C) used alone.

[0329] Computational efficiency: The average computational time of the collaborative algorithm is 28 seconds, which is 6.4 times faster than the FSI algorithm (180 seconds). Although it is slower than the PSO-NN algorithm (0.12 seconds), the accuracy has been significantly improved.

[0330] Control effect: The temperature fluctuation range under the control of the collaborative algorithm is ±2.8°C, which is significantly lower than that of traditional control (±8.5°C) and PSO-NN control (±4.2°C).

[0331] Energy efficiency: The daily energy consumption of the collaborative algorithm is 9.5 kWh, saving 23.4% compared to traditional control and 12.0% compared to PSO-NN control.

[0332] Early warning ability: The collaborative algorithm can predict the formation of hot spots 37 seconds in advance, providing sufficient response time for the control system.

[0333] Equipment life: The expected life of the equipment under the control of the collaborative algorithm is increased by 7.8%, which has good economic value.

[0334] In summary, the present invention can effectively improve the temperature prediction accuracy, control effect and energy efficiency of the intelligent control system of the laser cutting machine control cabinet, while reducing the equipment failure rate and extending the service life of the equipment, which is of great significance.

Claims

1. An intelligent control system for the control cabinet of a laser cutting machine, characterized in that, Comprising: A control cabinet body, the control cabinet body includes a cabinet (1), both sides of the front of the cabinet (1) are hinged with cabinet doors (2), several ventilation slots (3) are provided on the lower side of the cabinet doors (2), several mounting racks (5) are installed inside the cabinet (1), the mounting rack (5) includes side plates (7), and several long strip-shaped heat dissipation holes (16) are provided on both the side plates (7) and the cover plates (15); A temperature monitoring unit for collecting temperature data inside the control cabinet body; A temperature prediction and processing unit, including a heat pipe network flow prediction module, the temperature prediction and processing unit uses a combination of particle swarm optimization algorithm and neural network to predict the heat flow distribution inside the control cabinet, wherein model data generated by fluid-structure interaction simulation is introduced to train the neural network, and physical conservation constraints are added to the neural network loss function; An air flow simulation unit, including a fluid-structure interaction simulation module, the fluid-structure interaction simulation module only performs high-precision adaptive grid finite element calculations on key areas inside the control cabinet, and uses neural network prediction for non-critical areas to simulate the interaction between the air flow and the structure inside the control cabinet; A collaborative optimization unit for combining the output results of the temperature prediction and processing unit and the air flow simulation unit to achieve intelligent management of the temperature and air flow inside the control cabinet; The collaborative optimization unit includes: A multi-scale fusion module for realizing the fusion of heat flow models at micro and macro scales, the micro-scale model is realized through the fluid-structure interaction simulation module, and the macro-scale model is realized through the heat pipe network flow prediction module; A data-physics hybrid training module for establishing a physics-enhanced neural network and a data-assisted adaptive finite element model, including: Physical enhanced neural network loss function: , MSE(T Pred , T FSI ) is the mean square error between the predicted temperature and the FSI result; is the constraint residual of the heat diffusion equation; |▽·u| is the constraint residual of fluid incompressibility; λ1 and λ2 are weight coefficients; T is the temperature field, t is time, T pred is the predicted temperature field output by the neural network, T FSI is the predicted temperature of the fluid-structure interaction simulation module; q is the heat source term; Data-assisted adaptive criterion: , where η FEM is the traditional error estimate, β is the mixing coefficient, T FEM is the finite element solution, T NN is the neural network prediction result; An adaptive calculation strategy module that realizes a three-level calculation architecture, including: A real-time monitoring layer: directly predict using a pre-trained neural network; A state evaluation layer: optimize the prediction using a particle swarm-neural network algorithm; A deep analysis layer: use a complete fluid-structure interaction simulation; automatically switch the calculation level according to the temperature gradient index; A control execution unit for adjusting the cooling system of the control cabinet according to the optimization result of the collaborative optimization unit to maintain the temperature inside the control cabinet within a preset range.

2. The intelligent control system of the laser cutting machine control cabinet according to claim 1, wherein The heat pipe network flow prediction module uses the following steps for temperature field prediction: a) Establish a heat diffusion equation model: , where T is the temperature field, α is the thermal conductivity, and q is the heat source term; b) Construct a neural network structure to fit the temperature field distribution; c) Optimize the neural network weights using the particle swarm optimization algorithm: , where x i represents the neural network weights, v i represents the particle velocity, w represents the inertia weight, c1 and c2 represent the acceleration constants, r1 and r2 represent random numbers, p best,i represents the individual optimal position, and g best represents the global optimal position.

3. The intelligent control system of the laser cutting machine control cabinet according to claim 2, characterized in that, The fluid-structure interaction simulation module uses the following equations for simulation: a) Fluid part calculation: and , where u is the flow velocity, p is the pressure, ρ is the fluid density, μ is the dynamic viscosity, and f is the body force; b) Structural part calculation: , where d is the structural displacement, ρ s is the structural density, σ is the stress tensor, f s is the external force; c) Adaptive mesh refinement: , where h is the mesh size, η is the error estimator, p is the polynomial order of the finite element space, and η target is the target error.

4. The intelligent control system of the laser cutting machine control cabinet according to claim 3, wherein, A complementary verification mechanism module is provided between the temperature prediction and processing unit and the air flow simulation unit for verifying the consistency of the prediction results of the two units, and the complementary verification mechanism module includes: Anomaly detection logic, through the formula to determine whether to trigger an anomaly alarm, where T NN is the predicted temperature of the heat pipe network flow prediction module, T FSI is the predicted temperature of the fluid-structure interaction simulation module, and δ is the threshold coefficient; Confidence evaluation logic, through the formula Calculate the confidence of the prediction result, where κ is the scaling coefficient; A self-calibration mechanism that continuously corrects model parameters through actual temperature monitoring data, including: Physical parameter identification: , where α is the thermal conductivity, μ is the update step, and T measured is the actually measured temperature, and T pred indicates the predicted temperature; Neural network adaptation: , where ω1 and ω2 are weight coefficients, and T measured is the actual measured temperature.

5. The intelligent control system of the laser cutting machine control cabinet according to claim 4, characterized in that, The control execution unit includes: Hotspot warning module, predicting the risk of hotspot formation through collaborative algorithms: , where T critical is the critical temperature threshold; A predictive cooling control module that activates directional cooling in advance according to the hot spot risk: , where T safe is the safety temperature threshold, K is the proportionality coefficient, and a is the nonlinear exponent; Cooling resource allocation module, through optimizing the problem to achieve the optimal allocation of cooling power, where P i is the power of the i-th cooling device, Q i is the corresponding refrigerating capacity, and Q cooling is the cooling demand; Maintenance decision support module, through the component life prediction equation Predict the life of electronic components, where L is the expected life, L0 is the reference life, E a is the activation energy, k is the Boltzmann constant, T0 is the reference temperature, T max is the predicted operating temperature.

6. The intelligent control system of the laser cutting machine control cabinet according to claim 1, characterized in that, The adaptive calculation strategy module automatically selects the calculation level according to the following rules: , where represents the maximum temperature gradient, Threshold1 and Threshold2 are preset thresholds, and Level is the calculation level.

7. The intelligent control system of the laser cutting machine control cabinet according to claim 4, characterized in that, The temperature prediction and processing unit also includes an incremental update mechanism to realize self-optimization of the algorithm in the following way: FSI Incremental Update Neural Network: , where W represents the neural network weights, γ is the learning rate, and D FSI is the data generated by the fluid-structure interaction simulation; Neural network assisted FSI initial value setting: , where T0, u0, and p0 are the initial conditions of temperature, velocity, and pressure, respectively.

8. The intelligent control system of the laser cutting machine control cabinet according to claim 5, characterized in that, The control execution unit further includes a timing optimization module that considers historical temperature data through a timing fusion mechanism: , where Z T is the fusion result at the current moment, and Z (t-i) is the fusion result i time units ago, and α i is the time decay coefficient, satisfying .

Images