Steel-wood fireproof door integrated deformation prevention process and matching equipment

Abstract

The application relates to the technical field of steel-wood fireproof doors, and discloses a steel-wood fireproof door integrated anti-deformation process and matching equipment, which solves the problem of the integrated anti-deformation of the steel-wood fireproof door on the market, and comprises a supporting frame, the inner side of the supporting frame is provided with a roller conveyor body, and the roller conveyor body is provided with a driving part on one side; the rotating transmission fan can generate wind power, dust and debris on the surface of the material can be blown away, and the accumulation of dust and debris or liquid on the surface of the material is reduced; meanwhile, the rotating cleaning brush can brush and clean the surface of the steel-wood fireproof door moving on the surface of the roller conveyor body, manual treatment is not needed, the portability of the pretreatment step of the steel-wood fireproof door is improved, and the cost of manual work is saved; and the dust removal mechanism and the cleaning mechanism are integrated into the top frame, so that the maintenance process of the equipment is simplified, and the practicability of the fireproof door integrated forming equipment is improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of steel-wood fireproof doors, and particularly relates to a steel-wood fireproof door integrated anti-deformation process and matching equipment. BACKGROUND

[0002] The steel-wood fireproof door is a fireproof door made of steel and non-combustible wood materials or non-combustible wood products for door frame, door leaf skeleton and door leaf panel, and the door leaf filling material is a non-toxic and harmless fireproof and heat insulation material for human body, and is matched with fireproof hardware fittings. The door has the advantages of steel fireproof doors and wood fireproof doors, has good fireproof stability and fire resistance, and has good decoration and appearance.

[0003] In the prior art, the authorized publication No. CN116852479A discloses a "steel-wood fireproof door integrated die casting forming equipment"; a glue spraying mechanism is installed above the steel door panel frame roller conveyor; a jacking positioning mechanism is installed below the steel door panel frame roller conveyor; a support is arranged outside the steel door panel frame roller conveyor and the fireproof wood core roller conveyor, the upper end of the support is provided with a top plate, the lower surface of the top plate is provided with a first electric guide rail, and a material moving and die casting mechanism is installed on the sliding block of the first electric guide rail.

[0004] The "steel-wood fireproof door integrated die casting forming equipment" still has some shortcomings, for example: in the integrated forming process of the existing steel-wood fireproof door, the operator needs to pretreat the material, and in the pretreatment process, the operator needs to manually pretreat the material, which is relatively cumbersome and increases the workload of the operator, and in the production process of the steel-wood fireproof door, deformation problems caused by different material thermal expansion coefficients, humidity changes and other factors seriously affect the fireproof performance and service life of the fireproof door.

[0005] Therefore, the steel-wood fireproof door integrated anti-deformation process and matching equipment are provided to solve the above problems. SUMMARY

[0006] In view of the above problems, the steel-wood fireproof door integrated anti-deformation process and matching equipment are provided to overcome the defects of the prior art, effectively solve the problem that the steel-wood fireproof door integrated forming device on the market needs to be manually pretreated by an operator, and the steps are relatively cumbersome and increase the workload of the operator.

[0007] In order to achieve the above object, the present application provides the following technical scheme: steel wood fireproof door integrated deformation prevention process and matching equipment, including support frame, the inner side of the support frame is provided with a roller conveyor body, one side of the roller conveyor body is provided with a driving part, the outer side of the support frame is provided with a dust removal mechanism, the dust removal mechanism includes a top frame, a motor body and a protective box, the top of the top frame is fixedly connected with the motor body, the output shaft of the motor body is provided with a transmission rod, the bottom end of the transmission rod is movably connected with a connecting shaft, one side of the connecting shaft is movably connected with a transmission fan.

[0008] Preferably, one side of the support frame is provided with a heat preservation box, one side of the heat preservation box is provided with a temperature control module, the top of the heat preservation box is provided with a heating part, the outer side of the support frame is provided with a die casting table, one side of the die casting table is movably connected with a control box, and the top of the die casting table is provided with a die casting body.

[0009] Preferably, one side of the support frame is provided with an adjusting mechanism, the adjusting mechanism includes a bottom sliding frame, a bottom adjusting table and a sliding groove, both sides of the support frame are provided with a bottom adjusting table, the surface of the bottom adjusting table is provided with a sliding groove, the bottom of the top frame is fixedly connected with a bottom sliding frame, and the bottom of the bottom sliding frame is slidably connected with the surface of the sliding groove.

[0010] Preferably, one side of the top frame is fixedly connected with a protective box, and both sides of the protective box are movably connected with side protective nets.

[0011] Preferably, the inner side of the top frame is provided with a cleaning mechanism, the cleaning mechanism includes a first pulley, a transmission belt and a second pulley, the surface of the transmission rod is movably connected with the first pulley, one side of the first pulley is movably connected with the transmission belt, one side of the transmission belt is movably connected with the second pulley, the inner side of the second pulley is movably connected with a connecting rod, the bottom end of the connecting rod is fixedly connected with a first bottom gear, one side of the first bottom gear is movably connected with a first side gear, the inner side of the first side gear is fixedly connected with a side rod, both ends of the side rod are fixedly connected with a second side gear, one side of the second side gear is movably connected with a third side gear, one side of the third side gear is movably connected with a movable rod, and one side of the movable rod is movably connected with a cleaning brush.

[0012] Preferably, one side of the top frame is fixedly connected with a side frame, one side of the side frame is fixedly connected with a threaded sleeve, the inner side of the threaded sleeve is threadedly connected with a threaded rod, the top of the threaded rod is fixedly connected with a grip disc, and the bottom of the threaded rod is fixedly connected with a rubber pad.

[0013] Preferably, the top end of the connecting rod is movably connected with a bearing disc, and one side of the bearing disc is movably connected with one side of the top frame.

[0014] Preferably, the inner side of the top frame is fixedly connected with a fixing frame, one side of the fixing frame is fixedly connected with a top plate, one side of the top plate is fixedly connected with a limiting frame, one side of the limiting frame is provided with a communication hole, and one side of the communication hole is movably connected with both ends of the movable rod.

[0015] Preferably, the protective box is provided with a communication frame on one side, and the communication frame is arranged on the outer side of the transmission belt.

[0016] The steel-wood fireproof door integrated deformation prevention process comprises the following steps:

[0017] Step S1, material selection: special treated steel and wood materials are selected to ensure that they have similar thermal expansion coefficients, so as to maintain stability in subsequent processing, and auxiliary materials such as fireproof paint, glue and sealant are prepared;

[0018] Step S2, material pretreatment: using cutting machinery to preliminarily cut the steel and wood materials, cutting according to the size and structure design of the fireproof door, and performing stress relief treatment to ensure uniform internal stress of the materials and reduce the possibility of deformation, and drying the materials to remove excess moisture to prevent swelling or shrinkage in subsequent processing;

[0019] Step S3, integrated forming: using integrated forming equipment, the pretreated steel and wood materials are accurately positioned according to the design requirements, and hot pressing is performed using hot pressing machinery, so that the two materials are tightly combined through high temperature and high pressure to form an integral whole;

[0020] Step S4, fireproof treatment: spraying fireproof paint on the surface of the door body to increase the fireproof performance, and adding fireproof sealing strips at key positions of the door body as needed to improve the sealing performance and fire resistance time;

[0021] Step S5, post-treatment: polishing and finishing the formed fireproof door to ensure smooth and flat surface, and performing quality inspection including size accuracy, fireproof performance, etc., to ensure that the product meets the standards;

[0022] Step S6, packaging and warehousing: using appropriate packaging materials for packaging to prevent damage during transportation and storage.

[0023] Further, in S2, the intelligent algorithm based on machine learning can optimize the cutting and stress relief process of the materials; by monitoring the stress distribution and deformation of the materials in real time, the algorithm can adjust the cutting and processing parameters to minimize the possibility of deformation and ensure the quality of the materials; the specific process is as follows:

[0024] Step 1, install sensor network to collect real-time data during frame assembly processing, including temperature sensors, humidity and pressure sensors; transmit sensor data to data acquisition module for processing and preparation;

[0025] Temperature data collection: select thermocouple temperature sensor suitable for high temperature environment, install temperature detector at key positions of processing equipment, monitor temperature change in real time during processing; set sampling frequency, collect temperature data every second, ensure timeliness and accuracy of data; record temperature value at each time point and store in database for subsequent analysis;

[0026] Humidity data collection: use humidity sensor to arrange humidity detector in processing environment, monitor humidity change in real time during processing; set sampling frequency, collect humidity data every second, ensure real-time and accuracy of humidity data; record humidity value at each time point and store in database for subsequent analysis;

[0027] Pressure data collection: select strain gauge pressure sensor that can withstand high pressure environment to monitor pressure applied by hot pressing machine; set sampling frequency, collect pressure data every second, ensure real-time and accuracy of pressure data; record pressure value at each time point and store in database for subsequent analysis;

[0028] Through data cleaning, outlier processing and data smoothing method, the collected data is preprocessed to ensure data quality and availability;

[0029] Step 2, establish deep Q network (DQN) model combining Q learning and deep neural network according to existing processing data of sensor and real-time feedback data; the following is the detailed implementation process:

[0030] 1. State representation, state (S) is represented as a three-dimensional vector, where S=(T, H, E), where:

[0031] T is the temperature of the processed material (unit: Celsius);

[0032] H is the humidity during processing (unit: relative humidity RH);

[0033] E is the equipment running state (0 represents the equipment is not running, 1 represents the equipment is running);

[0034] 2. Action representation, action (A) is defined as a three-dimensional vector, where A=(V, T', P), where:

[0035] V is the cutting speed (0-100, unit: m / s);

[0036] T' is the processing temperature (0-500 degrees Celsius) ;

[0037] P is the processing pressure (0-50, unit: MPa) ;

[0038] 3. Reward function design:

[0039] The design of the reward function is a key part of reinforcement learning, which directly affects the convergence and performance of the algorithm; R t represents the reward, R t is the reward obtained after taking action A t in state S t , the designed reward function is as follows:

[0040] (a) Deformation control reward:

[0041] If the material deformation is successfully reduced, the reward value R t +10;

[0042] If the material deformation remains within a small range, the reward value R t +5;

[0043] If the material deformation exceeds the threshold, the reward value R t -10;

[0044] (b) Production efficiency reward:

[0045] If the production efficiency is high, the reward value R t +8;

[0046] If the production efficiency is low, the reward value R t -5;

[0047] (c) Resource utilization reward:

[0048] If the material and energy are effectively utilized in the processing process, the reward value R t +6;

[0049] If there is resource waste or excessive energy consumption, the reward value R t -3;

[0050] (d) Safety reward:

[0051] If the safety state is maintained during processing, the reward value R t +7;

[0052] If there is a safety hazard or accident, the reward value R t -8;

[0053] 4. Neural network structure:

[0054] Convolutional layer 1:

[0055] Input: State vector, size 3x1, i.e. three features (temperature, humidity, device status);

[0056] Convolution kernel size: 3x1;

[0057] Activation function: ReLU;

[0058] Output size: 32, representing 32 feature maps Figure 2 ;

[0059] Convolutional layer 2:

[0060] Input: Output of convolutional layer 1, size 32x1;

[0061] Convolution kernel size: 3x1;

[0062] Activation function: ReLU;

[0063] Output size: 64, representing 64 feature maps

[0064] Fully connected layer 1:

[0065] Input: Output of convolutional layer 2, size 64x1;

[0066] Output size: 128;

[0067] Activation function: ReLU;

[0068] Fully connected layer 2:

[0069] Input: Output of fully connected layer 1, size 128;

[0070] Output size: 3, representing Q-values for three actions;

[0071] 5. In the Q-learning algorithm, the target Q-value is calculated as the current reward plus the maximum Q-value of the next state multiplied by a discount factor; therefore, the update formula for the Q-value is:

[0072] Q(S t , A t ; θ) = (1 - α) · Q(S t , A t ; θ) + α · (R t + γ · max A′ Q(S i+1 , A'; θ target ))

[0073] Q(S t , A tθ): Q-value function, representing the Q-value of taking action A in state S, where θ is the parameter of the neural network;

[0074] α: learning rate;

[0075] R t : reward obtained after taking action A in state S t ; t

[0076] γ: discount factor, measuring the importance of future rewards;

[0077] max A′ Q(S i+ 1 ， A′; θ) target : maximum Q-value when choosing action in the next state S i+1 1;

[0078] θ target : parameter of the target network;

[0079] 6. Update the weights and biases of the network using gradient descent method based on the gradient of the loss function; the specific parameter update method is as follows:

[0080]

[0081] where α is the learning rate, is the gradient of the loss function with respect to the parameter θ;

[0082] 7. Loss function definition, using mean square error (MSE) loss function to measure the gap between the predicted Q-value and the target Q-value; specifically, the loss function is defined as the average of the square of the difference between the predicted Q-value and the target Q-value;

[0083]

[0084] Loss: loss function, measuring the gap between the predicted Q-value and the target Q-value;

[0085] N: sample number, used to calculate the average of the loss function;

[0086] i: represents the i-th sample in the training set;

[0087] S: state vector, representing the current state of the environment, including temperature, humidity and device status information;

[0088] A: action vector, representing the actions that can be taken, including adjusting the cutting speed, temperature and pressure parameters;

[0089] θ: parameter of the neural network, including weights and biases;

[0090] Q(S, A; θ): Q-value function, representing the Q-value of taking action A in state S, where θ is the parameter of the neural network;

[0091] R: The reward obtained after taking action A in state S;

[0092] γ: Discount factor, used to measure the importance of future rewards;

[0093] S i+1 The next state, i.e., in state S i Take action A i The state after;

[0094] A′: Next state S i+1 One of the possible actions to choose from;

[0095] max A′ Q(S i+1 , A′; θ target ): In the next state S i+1 In the middle, the maximum Q value when selecting an action;

[0096] θ target : Parameters of the target network, used to calculate the target Q value;

[0097] Step 3: Optimize parameter adjustments during processing using a Deep Q-Network (DQN) to reduce material deformation and improve production efficiency; select the action that maximizes the Q-value as the action to be executed in the current state according to a greedy policy; specifically, select the action that satisfies the following conditions:

[0098] A t =argmax A Q(S, A; θ)

[0099] Among them, A t It is the action to be performed under the current state S; argmax A Q(S, A; θ) represents the action that maximizes the Q value among all possible actions A, including adjusting different cutting speeds, processing temperatures, and processing pressures.

[0100] Compared with the prior art, the beneficial effects of the present invention are:

[0101] 1. In the steel-wood fireproof door integrated forming anti-deformation process and matching equipment work, after the motor body is started, the rotating transmission fan can generate wind, the generated wind blows the material on the surface of the roller conveyor body, and dust and debris on the surface of the material can be blown off. At the same time, the transmission fan can be protected by the protective box, and the dust can be blocked by the side protective net on both sides of the protective box, reducing the dust entering the inside of the protective box when the transmission fan rotates. The wind generated by the transmission fan can reduce the accumulation of dust and debris or liquid on the surface of the material;

[0102] 2. In the steel-wood fireproof door integrated forming anti-deformation process and matching equipment work, the rotation of the movable rod can drive the movement of the multiple cleaning brushes on the surface of the movable rod. The rotating cleaning brushes can clean the surface of the steel-wood fireproof door moving on the surface of the roller conveyor body without manual processing by the operator, improving the portability of the steel-wood fireproof door preprocessing step and saving the cost of manual labor.

[0103] 3. In the steel-wood fireproof door integrated forming anti-deformation process and matching equipment work, the sliding connection of the bottom sliding frame and the sliding groove can drive the top frame to move horizontally along the outside of the support frame. The position of the top frame can be adjusted according to the different specifications of the fireproof door material. At the same time, by integrating the dust removal mechanism and the cleaning mechanism into the top frame, the maintenance process of the equipment can be simplified, and the practicality of the fireproof door integrated forming equipment is improved.

[0104] 4. In the steel-wood fireproof door integrated forming process, factors such as material temperature, humidity, and equipment state are involved. These factors may be continuous or have high-dimensional characteristics. Using deep neural networks can effectively model complex state spaces. Through the optimization of the Q value function, the action with the maximum Q value in each state can be selected, thereby optimizing and adjusting key parameters such as cutting speed, temperature, and pressure during the processing process to improve the production efficiency and quality of the steel-wood fireproof door. Using reinforcement learning methods, the optimal strategy can be learned automatically during interaction with the environment without explicitly formulating rules for the processing process. This makes the production process of the steel-wood fireproof door automated and intelligent, reducing labor costs and error rates. Through the training of deep neural networks, the model can learn the complex characteristics of the steel-wood fireproof door integrated forming process and can generalize to new unseen situations. Even when faced with different material properties, environmental conditions, or production requirements, reasonable decisions and optimization adjustments can be made. BRIEF DESCRIPTION OF DRAWINGS

[0105] The accompanying drawings are used to provide a further understanding of the present application, and constitute a part of the specification, together with the embodiments of the present application, to explain the present application, and do not constitute a limitation on the present application. In the drawings:

[0106] Figure 1 It is the schematic diagram of the overall appearance structure of the present application;

[0107] Figure 2 It is the schematic diagram of the dust removal mechanism structure of the present application;

[0108] Figure 3 It is the schematic diagram of the bottom sliding frame structure of the present application;

[0109] Figure 4 It is the schematic diagram of the protective box structure of the present application;

[0110] Figure 5 It is the schematic diagram of the top plate appearance structure of the present application;

[0111] Figure 6 It is the schematic diagram of the cleaning mechanism structure of the present application;

[0112] Figure 7 It is the schematic diagram of the adjusting mechanism structure of the present application;

[0113] In the figure: 1, support frame; 2, roller conveyor body; 3, driving part; 4, dust removal mechanism; 401, top frame; 402, motor body; 403, protective box; 404, side protective net; 405, transmission rod; 406, connecting shaft; 407, transmission fan; 5, cleaning mechanism; 501, first belt pulley; 502, transmission belt; 503, second belt pulley; 504, bearing disc; 505, connecting rod; 506, communication frame; 507, No. 1 bottom tooth disc; 508, No. 1 side tooth disc; 509, side rod; 5010, No. 2 side tooth disc; 5011, No. 3 side tooth disc; 5012, movable rod; 5013, cleaning brush; 5014, limiting frame; 5015, communication hole; 5016, top plate; 5017, fixed frame; 6, adjusting mechanism; 601, bottom sliding frame; 602, bottom adjusting table; 603, sliding groove; 604, side frame; 605, threaded sleeve; 606, threaded rod; 607, grip disc; 608, rubber pad; 7, heat preservation box; 8, temperature control module; 9, heating part; 10, die casting table; 11, die casting body; 12, control box. DETAILED DESCRIPTION

[0114] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work belong to the scope of protection of the present application.

[0115] Embodiment 1

[0116] In this embodiment,Figures 1-7 The present application provides the following technical solutions:

[0117] The steel-wood fireproof door integrated anti-deformation process and matching equipment comprises a support frame 1, a roller conveyor body 2 is arranged on the inner side of the support frame 1, a driving part 3 is arranged on one side of the roller conveyor body 2, a dust removal mechanism 4 is arranged on the outer side of the support frame 1, the dust removal mechanism 4 comprises a top frame 401, a motor body 402 and a protective box 403, the motor body 402 is fixedly connected to the top of the top frame 401, a transmission rod 405 is arranged on the output shaft of the motor body 402, a connecting shaft 406 is movably connected to the bottom end of the transmission rod 405, and a transmission fan 407 is movably connected to one side of the connecting shaft 406.

[0118] It should be noted that the rotating transmission rod 405 can drive the connecting shaft 406 to rotate, the connecting shaft 406 can drive the transmission fan 407 to rotate after rotating, and the rotating transmission fan 407 can generate wind power.

[0119] In this embodiment, a heat preservation box 7 is arranged on one side of the support frame 1, a temperature control module 8 is arranged on one side of the heat preservation box 7, a heating part 9 is arranged on the top of the heat preservation box 7, a die casting table 10 is arranged on the outer side of the support frame 1, a control box 12 is movably connected to one side of the die casting table 10, and a die casting body 11 is arranged on the top of the die casting table 10.

[0120] In this embodiment, an adjusting mechanism 6 is arranged on one side of the support frame 1, the adjusting mechanism 6 comprises a bottom sliding frame 601, a bottom adjusting table 602 and a sliding groove 603, the bottom adjusting tables 602 are arranged on the two sides of the support frame 1, the sliding grooves 603 are arranged on the surfaces of the bottom adjusting tables 602, the bottom sliding frame 601 is fixedly connected to the bottom of the top frame 401, and the bottom of the bottom sliding frame 601 is slidably connected to the surface of the sliding groove 603.

[0121] It should be noted that the sliding connection of the bottom sliding frame 601 and the sliding groove 603 can drive the top frame 401 to move horizontally along the outer side of the support frame 1.

[0122] In this embodiment, the protective box 403 is fixedly connected to one side of the top frame 401, and the side protective nets 404 are movably connected to the two sides of the protective box 403.

[0123] It should be noted that the two sides of the protective box 403 can block dust through the side protective nets 404.

[0124] In the embodiment, the inner side of the top frame 401 is provided with a cleaning mechanism 5, which comprises a first belt pulley 501, a transmission belt 502 and a second belt pulley 503. The surface of the transmission rod 405 is movably connected with the first belt pulley 501. One side of the first belt pulley 501 is movably connected with the transmission belt 502. One side of the transmission belt 502 is movably connected with the second belt pulley 503. The inner side of the second belt pulley 503 is movably connected with a connecting rod 505. The bottom end of the connecting rod 505 is fixedly connected with a first bottom gear 507. One side of the first bottom gear 507 is movably connected with a first side gear 508. The inner side of the first side gear 508 is fixedly connected with a side rod 509. The two ends of the side rod 509 are fixedly connected with a second side gear 5010. One side of the second side gear 5010 is movably connected with a third side gear 5011. One side of the third side gear 5011 is movably connected with a movable rod 5012. One side of the movable rod 5012 is movably connected with a cleaning brush 5013.

[0125] It should be noted that the rotation of the movable rod 5012 can drive the movement of the plurality of cleaning brushes 5013 on the surface of the movable rod 5012. The rotating cleaning brush 5013 can brush and clean the surface of the steel and wood fireproof door moving on the surface of the roller conveyor body 2.

[0126] In the embodiment, one side of the top frame 401 is fixedly connected with a side frame 604. One side of the side frame 604 is fixedly connected with a threaded sleeve 605. The inner side of the threaded sleeve 605 is threadedly connected with a threaded rod 606. The top of the threaded rod 606 is fixedly connected with a grip disc 607. The bottom of the threaded rod 606 is fixedly connected with a rubber pad 608.

[0127] It should be noted that by lowering the rubber pad 608, the contact between the rubber pad 608 and the sliding groove 603 can limit the top frame 401.

[0128] In the embodiment, the top end of the connecting rod 505 is movably connected with a bearing disc 504. One side of the bearing disc 504 is movably connected with one side of the top frame 401.

[0129] It should be noted that the connection of the bearing disc 504 can limit the connection of the connecting rod 505 by the bearing disc 504.

[0130] In the embodiment, the inner side of the top frame 401 is fixedly connected with a fixed frame 5017. One side of the fixed frame 5017 is fixedly connected with a top plate 5016. One side of the top plate 5016 is fixedly connected with a limiting frame 5014. One side of the limiting frame 5014 is provided with a communication hole 5015. One side of the communication hole 5015 is movably connected with the two ends of the movable rod 5012.

[0131] It should be noted that the two ends of the movable rod 5012 are limited and supported through the communication holes 5015 opened on the surface of the limiting frame 5014. By rotating the movable rod 5012, the movement of the plurality of cleaning brushes 5013 on the surface of the movable rod 5012 can be driven.

[0132] In this embodiment, a communication frame 506 is arranged on one side of the protection box 403 and outside the transmission belt 502.

[0133] It should be noted that the communication frame 506 arranged on one side of the protection box 403 enables the transmission belt 502 to move along the two sides of the protection box 403.

[0134] Embodiment 2

[0135] The embodiment 2 provides the working principle of the steel-wood fireproof door integrated deformation prevention process and the matching equipment, which is used for further explaining the working process or principle of the steel-wood fireproof door integrated deformation prevention process and the matching equipment provided in the above embodiment 1, and the specific process is as follows:

[0136] The steel-wood fireproof door integrated deformation prevention process comprises the following steps:

[0137] S1, by driving the driving part 3, the roller conveyor body 2 can be driven to move. By using the roller conveyor body 2 moving in the support frame 1, the steel-wood fireproof door material which needs to be integrated can be transported along the surface of the roller conveyor body 2. After the steel-wood fireproof door material moves to the inner side of the top frame 401, the motor body 402 is started, the transmission rod 405 is driven to rotate by the motor body 402, the connecting shaft 406 is driven to rotate by the rotating transmission rod 405, the transmission fan 407 is driven to rotate by the rotating connecting shaft 406, and the wind power is generated by the rotating transmission fan 407. The wind power generated by the rotating transmission fan 407 can blow the material on the surface of the roller conveyor body 2, so as to blow off the dust and debris on the surface of the material. At the same time, the transmission fan 407 can be protected by the protection box 403 when rotating, and the dust can be blocked by the side protection net 404 arranged on the two sides of the protection box 403, so as to reduce the dust entering the inner side of the protection box 403 when the transmission fan 407 rotates. By using the wind power generated by the transmission fan 407, the accumulation of dust and debris or liquid on the surface of the material can be reduced.

[0138] S2, while rotating the transmission rod 405, the rotation of the transmission rod 405 drives the first pulley 501 to rotate, and the transmission belt 502 is sleeved with the first pulley 501 and the second pulley 503, so that the first pulley 501 can drive the second pulley 503 to rotate when rotating, and the rotating second pulley 503 can drive the connecting rod 505 to rotate, and the connecting rod 505 is connected and limited by the bearing disc 504 when rotating, and the rotating connecting rod 505 drives the first bottom gear disc 507 to rotate, and the first bottom gear disc 507 is movably connected with the first side gear disc 508, so that the first bottom gear disc 507 can drive the first side gear disc 508 to rotate when rotating, and the rotation of the first side gear disc 508 drives the side rod 509 to rotate, and the rotation of the side rod 509 drives the two second side gear discs 5010 to rotate, and the rotation of the second side gear disc 5010 drives the third side gear disc 5011 to rotate, and the rotation of the third side gear disc 5011 drives the movable rod 5012 to rotate, and the two ends of the movable rod 5012 are limited and supported by the through holes 5015 formed in the surface of the limiting frame 5014, and the rotation of the movable rod 5012 drives the movement of the plurality of cleaning brushes 5013 on the surface of the movable rod 5012, and the rotating cleaning brushes 5013 can brush and clean the surface of the steel and wood fireproof door moving on the surface of the roller conveyor body 2, without manual operation, improving the portability of the steel and wood fireproof door preprocessing step, and saving the cost of manual operation;

[0139] S3, while the operator rotates the handle disc 607, the rotating handle disc 607 drives the threaded rod 606 to rotate, and the threaded rod 606 is connected with the threads in the threaded sleeve 605, so that the threaded rod 606 can be vertically adjusted along the inside of the threaded sleeve 605 after rotating, and the threaded rod 606 drives the rubber pad 608 to move vertically after vertical movement, and the rubber pad 608 is separated from the surface of the sliding groove 603 on the inside of the bottom adjusting table 602 after being lifted along one side of the side frame 604, and then the top frame 401 can be pushed, the bottom sliding frame 601 is slidably connected with the sliding groove 603, so that the top frame 401 can be horizontally moved along the outside of the support frame 1, and after moving, the rubber pad 608 is lowered, and the rubber pad 608 is in contact with the sliding groove 603, so that the top frame 401 can be limited, and the position of the top frame 401 can be adjusted according to different specifications of the fireproof door material, and the dust removal mechanism 4 and the cleaning mechanism 5 are integrated into the top frame 401, so that the maintenance process of the equipment is simplified, and the practicability of the fireproof door integrated forming equipment is improved;

[0140] S4, after the surface of the material is dusted and dried, the material enters the heat preservation box 7. At this time, after the parameters are adjusted by the temperature control module 8, the temperature in the heat preservation box 7 can be raised by starting the heating component 9, and the material can be dried. According to the characteristics of the material and the required drying degree, the temperature control module 8 can set appropriate drying temperature, humidity and time. After drying is completed, the temperature of the material is raised again, and after the temperature of the material is raised to a certain high temperature and maintained for a period of time, and then slowly cooled, the stress in the material can be released and homogenized, reducing the deformation of the material. After the material is pretreated, the material is placed on the surface of the die casting table 10. Through the control of the control box 12, the die casting machine body 11 can be used to die-cast the material on the surface of the die casting table 10;

[0141] In S2, the intelligent algorithm based on machine learning can optimize the cutting and stress relief process of the material; by monitoring the stress distribution and deformation of the material in real time, the algorithm can adjust the cutting and processing parameters to minimize the possibility of deformation and ensure the quality of the material; the specific process is as follows:

[0142] Step 1, install a sensor network to collect real-time data during the processing of the frame assembly, including temperature sensors, humidity and pressure sensors; transmit the sensor data to the data acquisition module for processing and preparation;

[0143] Temperature data collection: select a thermocouple temperature sensor suitable for high temperature environment, install temperature detectors at key positions of the processing equipment, and monitor temperature changes in real time during processing; set the sampling frequency to one per second to ensure the timeliness and accuracy of the data; record the temperature value at each time point and store it in the database for subsequent analysis;

[0144] Humidity data collection: use humidity sensors to place humidity detectors in the processing environment to monitor humidity changes in real time during processing; set the sampling frequency to one per second to ensure the real-time and accuracy of the humidity data; record the humidity value at each time point and store it in the database for subsequent analysis;

[0145] Pressure data collection: select a strain gauge pressure sensor that can withstand high pressure environment to monitor the pressure applied by the hot pressing machine; set the sampling frequency to one per second to ensure the real-time and accuracy of the pressure data; record the pressure value at each time point and store it in the database for subsequent analysis;

[0146] Through data cleaning, outlier processing and data smoothing methods, the collected data is preprocessed to ensure data quality and availability;

[0147] Step 2: Establish a Deep Q-Network (DQN) model combining Q-learning and deep neural networks based on existing process data and real-time feedback data from the sensor. The detailed implementation process is as follows:

[0148] 1. State representation: The state (S) is represented as a three-dimensional vector, where S = (T, H, E), where:

[0149] T is the temperature of the processed material (in Celsius);

[0150] H is the humidity during the processing process (in relative humidity RH);

[0151] E is the equipment operating state (0 represents the equipment is not running, 1 represents the equipment is running);

[0152] 2. Action representation: The action (A) is defined as a three-dimensional vector, where A = (V, T', P), where:

[0153] V is the cutting speed (0-100, in m / s);

[0154] T' is the processing temperature (0-500 degrees Celsius);

[0155] P is the processing pressure (0-50, in MPa);

[0156] 3. Reward function design:

[0157] The design of the reward function is a key part of reinforcement learning, which directly affects the convergence and performance of the algorithm; use R t to represent the reward, R t is the reward obtained after taking action A t in state S t , the designed reward function is as follows:

[0158] (a) Deformation control reward:

[0159] If the material deformation is successfully reduced, the reward value R t +10;

[0160] If the material deformation remains within a small range, the reward value R t +5;

[0161] If the material deformation exceeds the threshold, the reward value R t -10;

[0162] (b) Production efficiency reward:

[0163] If the production efficiency is high, the reward value R t+8;

[0164] If the production efficiency is low, the reward value R t -5;

[0165] (c) Resource utilization reward:

[0166] If the materials and energy are effectively utilized during the processing, the reward value R t +6;

[0167] If there is resource waste or excessive energy consumption, the reward value R t -3;

[0168] (d) Safety reward:

[0169] If the safety state is maintained during processing, the reward value R t +7;

[0170] If there is a safety hazard or an accident occurs, the reward value R t -8;

[0171] 4. Neural network structure:

[0172] Convolutional layer 1:

[0173] Input: State vector, size 3x1, i.e. three features (temperature, humidity, device status);

[0174] Convolution kernel size: 3x1;

[0175] Activation function: ReLU;

[0176] Output size: 32, representing 32 feature maps Figure 2 ;

[0177] Convolutional layer 2:

[0178] Input: Output of convolutional layer 1, size 32x1;

[0179] Convolution kernel size: 3x1;

[0180] Activation function: ReLU;

[0181] Output size: 64, representing 64 feature maps;

[0182] Fully connected layer 1:

[0183] Input: Output of convolutional layer 2, size 64x1;

[0184] Output size: 128;

[0185] Activation function: ReLU;

[0186] Fully connected layer 2:

[0187] Input: The output of fully connected layer 1, with a size of 128;

[0188] Output size: 3, representing the Q-values ​​of the three actions;

[0189] 5. In the Q-learning algorithm, the target Q-value is calculated by adding the current reward to the maximum Q-value of the next state multiplied by a discount factor; therefore, the Q-value update formula is:

[0190] Q(S t A t ;θ)=(1-α)·θ(S t A t ;θ)+α·(R t +γ·max A′ Q(S i+1 , A′; θ target ))

[0191] Q(S t A t ;θ): Q-value function, representing the Q-value of taking action A in state S, where θ is a parameter of the neural network;

[0192] α: Learning rate;

[0193] R t In state S t Take action A t The reward received later;

[0194] γ: Discount factor, used to measure the importance of future rewards;

[0195] max A′ Q(S i+1 , A′; θ target ): In the next state S i+1 In the middle, the maximum Q value when selecting an action;

[0196] θ target : Parameters of the target network;

[0197] 6. Based on the gradient of the loss function, use gradient descent to update the network weights and biases; the specific parameter update method is as follows:

[0198]

[0199] Where α is the learning rate. It is the gradient of the loss function with respect to the parameter θ;

[0200] 7. Loss function definition: The mean squared error (MSE) loss function is used to measure the difference between the model's predicted Q value and the target Q value; specifically, the loss function is defined as the average of the squares of the differences between the predicted Q value and the target Q value.

[0201]

[0202] Loss: The loss function measures the difference between the predicted Q-value and the target Q-value;

[0203] N: Number of samples, used to calculate the average value of the loss function;

[0204] i: represents the i-th sample in the training set;

[0205] S: State vector, representing the current state of the environment, including temperature, humidity and device status information;

[0206] A: Action vector, representing the actions that can be taken, including adjusting cutting speed, temperature, and pressure parameters;

[0207] θ: The parameters of the neural network, including weights and biases;

[0208] Q(S, A; θ): Q-value function, representing the Q-value of taking action A in state S, where θ is the parameter of the neural network;

[0209] R: The reward obtained after taking action A in state S;

[0210] γ: Discount factor, used to measure the importance of future rewards;

[0211] S i+1 The next state, i.e., in state S i Take action A i The state after;

[0212] A′: Next state S i+1 One of the possible actions to choose from;

[0213] max A′ Q(S i+1 , A′; θ target ): In the next state S i+1 In the middle, the maximum Q value when selecting an action;

[0214] θ target : Parameters of the target network, used to calculate the target Q value;

[0215] Step 3, according to the deep Q network (Deep Q-Network, DQN) to optimize the parameter adjustment in the processing process, so as to reduce the material deformation and improve the production efficiency; according to the greedy policy (Greedy Policy), the action that makes the Q value maximum is selected as the execution action under the current state; specifically, the action that satisfies the following conditions is selected:

[0216] A t = argmax A Q(S, A; θ)

[0217] Wherein, A t is the execution action selected under the current state S; argmax A Q(S, A; θ) represents selecting the action that makes the Q value maximum in all possible actions A, including adjusting different cutting speed, processing temperature and processing pressure.

[0218] It should be noted that: the roller conveyor body 2, the driving component 3, the motor body 402, the temperature control module 8, the heating component 9, the die casting table 10, the die casting body 11 and the control box 12 in the application are all prior art, and the corresponding models can be selected according to actual needs. The internal structure and operating principle of the above-mentioned parts also belong to the common knowledge of those skilled in the art, and will not be described in detail.

[0219] Finally, it should be pointed out that: the above-mentioned is only the preferred embodiment of the application, and is not used to limit the application, although the application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the application shall be included in the protection scope of the application.

Claims

1. A steel-wood fireproof door integrated deformation prevention process, applied to a matching equipment for the steel-wood fireproof door integrated deformation prevention process, the matching equipment comprising a support frame (1), a roller conveyor body (2) is arranged on the inner side of the support frame (1), a driving component (3) is arranged on one side of the roller conveyor body (2), a dust removal mechanism (4) is arranged on the outer side of the support frame (1), the dust removal mechanism (4) comprises a top frame (401), a motor body (402) and a protection box (403), the top frame (401) is fixedly connected with the motor body (402) at the top, a transmission rod (405) is arranged on the output shaft of the motor body (402), a connecting shaft (406) is movably connected with the bottom end of the transmission rod (405), a transmission fan (407) is movably connected with one side of the connecting shaft (406), and a cleaning mechanism (5) is arranged on the inner side of the top frame (401); characterized in that The deformation prevention process comprises the following steps: Step S1, material selection: special treated steel and wood materials are selected to ensure that they have similar thermal expansion coefficients, so as to maintain stability in subsequent processing, and fireproof paint, glue and sealant auxiliary materials are prepared; Step S2, material pretreatment: using cutting machinery to preliminarily cut the steel and wood materials, cutting according to the size and structure design of the fireproof door, and stress relief treatment is carried out at the same time to ensure that the internal stress of the material is uniform and the possibility of deformation is reduced, and the material is dried to remove excess moisture to prevent swelling or shrinkage in subsequent processing; Step S3, integrated forming: using integrated forming equipment, the pretreated steel and wood materials are accurately positioned according to the design requirements, and hot pressing is carried out using hot pressing machinery, so that the two materials are tightly combined through high temperature and high pressure to form an integral whole; Step S4, fireproof treatment: spraying fireproof paint on the surface of the door body to increase the fireproof performance, and adding fireproof sealing strips at the key parts of the door body as needed to improve the sealing performance and fire resistance; Step S5, post-treatment: polishing and finishing the formed fireproof door to ensure smooth and flat surface, and quality inspection including size accuracy and fireproof performance to ensure that the product meets the standards; Step S6, packaging and storage: using appropriate packaging materials to prevent damage during transportation and storage; In S2, the cutting and stress relief process of the material is optimized by intelligent algorithm; by monitoring the stress distribution and deformation of the material in real time, the algorithm can adjust the cutting and processing parameters to minimize the possibility of deformation and ensure the quality of the material; the specific process is as follows: Step 1, install a sensor network to collect real-time data during processing, including temperature sensors, humidity and pressure sensors; transmit the sensor data to the data acquisition module for processing and preparation; Temperature data collection: select a thermocouple temperature sensor suitable for high temperature environment, install temperature detectors at key positions of the processing equipment to monitor temperature changes in real time during processing; Set the sampling frequency, collect temperature data once every second, ensure the timeliness and accuracy of the data; record the temperature value at each time point and store it in the database for subsequent analysis; Humidity data collection: humidity sensors are placed in the processing environment to monitor humidity changes in real time; Set the sampling frequency, collect humidity data once every second, ensure the real-time and accuracy of the humidity data; record the humidity value at each time point and store it in the database for subsequent analysis; Pressure data collection: strain gauge pressure sensors that can withstand high pressure environments are selected to monitor the pressure applied by the hot pressing machine; Set the sampling frequency, collect pressure data once every second, ensure the real-time and accuracy of the pressure data; record the pressure value at each time point and store it in the database for subsequent analysis; Through data cleaning, outlier processing and data smoothing methods, the collected data is preprocessed to ensure data quality and availability; Step 2, according to the existing processing process data of the sensor and the real-time feedback data, a deep Q network model combining Q learning and deep neural network is established; the following is the detailed implementation process: 2-1. State representation, the representation of a state S is a three-dimensional vector, where wherein: T is the temperature of the processing material; H is the humidity during processing; E is the equipment running state: 0 represents the equipment is not running, 1 represents the equipment is running; 2-2, Action representation, the definition of an action A is a three-dimensional vector, where , wherein: V is the cutting speed: 0-100, unit: m / s; is the processing temperature: 0-500, in degrees Celsius; P is the processing pressure: 0-50, unit: MPa; 2-3, reward function design: The design of reward function is a key part in reinforcement learning, which directly influences the convergence and performance of the algorithm. In this paper, we propose a new reward function for the reinforcement learning algorithm. The reward function is designed as follows: The reward function is designed as follows: The reward function is designed as follows: The reward function is designed as follows: (a) deformation control reward: If the material deformation is successfully reduced, a reward value +10; If the deformation of the material remains within a small range, the reward value +5; If the deformation of the material exceeds the threshold, the reward value -10; (b) production efficiency reward: If production efficiency is high, reward value +8; If the production efficiency is low, the reward value -5; (c) resource utilization reward: If the material and energy are effectively utilized in the process, the reward value +6; If there is a case of resource waste or excessive energy consumption, the reward value -3; (d) safety reward: If the safety state is maintained during the process, the reward value +7; If there is a security risk or an accident occurs, the reward value -8; 2-4, neural network structure: Convolutional layer 1: Input: state vector, size 3 x 1, i.e. three features: temperature, humidity, equipment state; Convolution kernel size: 3 1; Activation function: ReLU; Output size: 32, representing 32 feature maps; Convolutional layer 2: input: output of convolutional layer 1, size 32 1; Convolution kernel size: 3 1; Activation function: ReLU; Output size: 64, representing 64 feature maps; Fully connected layer 1: input: output of convolutional layer 2, size 64 1; Output size: 128; Activation function: ReLU; Fully connected layer 2: Input: output of fully connected layer 1, size 128; Output size: 3, representing the Q value of three actions; 2-5, in the Q-learning algorithm, the target Q value is calculated as the current reward plus the maximum Q value of the next state multiplied by the discount factor; therefore, the Q value update formula is: ； : Q-value function representing the Q-value of taking action A in state S, where are parameters of the neural network; : learning rate; :in the state :take action :reward obtained after : discount factor, which measures the importance of future rewards; : In the next state , the maximum Q value when selecting an action; : parameters of the target network; 2-6、According to the gradient of the loss function, the weights and biases of the network are updated using the gradient descent method; the specific parameter updating method is as follows: ; wherein, is the learning rate, is the gradient of the loss function with respect to the parameters ; 2-7, loss function definition, using mean square error MSE loss function to measure the gap between the predicted Q value and the target Q value; specifically, the loss function is defined as the average of the square of the difference between the predicted Q value and the target Q value; ； : a loss function that measures the gap between the predicted Q-values and the target Q-values; : number of samples, used to compute the average of the loss function; : denotes the i-th sample in the training set; : denotes the i-th sample in the training set; : state vector, representing the current state of the environment, including temperature, humidity and device state information; : action vector, representing actions that can be taken, including adjusting cutting speed, temperature, and pressure parameters; : parameters of the neural network, including weights and biases; : Q-value function representing the Q-value of taking action A in state S, where are parameters of the neural network; : the reward obtained after taking action A in state S; : discount factor, which measures the importance of future rewards; : next state, i.e. the state action is taken after the state; : next state : some action that can be chosen in the state; : In the next state , the maximum Q value when selecting an action; : parameters of the target network, used to calculate the target Q value; Step 3, optimize the parameter adjustment in the machining process according to the deep Q network to reduce material deformation and improve production efficiency; according to the greedy strategy, the action that makes the Q value maximum is selected as the execution action under the current state; specifically, the action that satisfies the following conditions is selected: ; wherein, is the current state the selected execution action; represents the Q value of all possible actions Among them, the action that makes the Q value maximum is selected, including adjusting different cutting speed, processing temperature and processing pressure.

2. The steel-wood fireproof door integrated deformation prevention process according to claim 1, characterized in that: One side of the support frame (1) is provided with a heat preservation box (7), one side of the heat preservation box (7) is provided with a temperature control module (8), the top of the heat preservation box (7) is provided with a heating part (9), the outer side of the support frame (1) is provided with a die casting table (10), one side of the die casting table (10) is movably connected with a control box (12), the top of the die casting table (10) is provided with a die casting machine body (11).

3. The steel-wood fire door integrated deformation prevention process of claim 2, wherein: One side of the support frame (1) is provided with an adjusting mechanism (6), the adjusting mechanism (6) includes bottom sliding frame (601), bottom adjusting table (602) and sliding groove (603), both sides of the support frame (1) are provided with bottom adjusting table (602), the surface of the bottom adjusting table (602) is provided with sliding groove (603), the bottom of the top frame (401) is fixedly connected with bottom sliding frame (601), the bottom of the bottom sliding frame (601) is slidably connected with the surface of the sliding groove (603).

4. The steel-wood fire door integrated deformation prevention process of claim 1, wherein: One side of the top frame (401) is fixedly connected with a side frame (604), one side of the side frame (604) is fixedly connected with a threaded sleeve (605), the inner side of the threaded sleeve (605) is threadedly connected with a threaded rod (606), the top of the threaded rod (606) is fixedly connected with a handle disc (607), the bottom of the threaded rod (606) is fixedly connected with a rubber pad (608).

5. The steel-wood fire door integrated deformation prevention process of claim 4, wherein: The inner side of the top frame (401) is fixedly connected with a fixed frame (5017), one side of the fixed frame (5017) is fixedly connected with a top plate (5016), one side of the top plate (5016) is fixedly connected with a limiting frame (5014), one side of the limiting frame (5014) is provided with a communication hole (5015), one side of the communication hole (5015) is movably connected with both ends of the movable rod (5012).

6. The steel-wood fire door integrated deformation prevention process of claim 5, wherein: One side of the protection box (403) is provided with a communication frame (506), and the communication frame (506) is arranged on the outer side of the transmission belt (502).

Images