A latex mattress preparation device and method

Abstract

The invention relates to the technical field of latex mattresses, and discloses a latex mattress preparation device and method thereof, which solve the problems of latex mattress preparation in the current market. The device comprises a processing base, the top of the processing base is fixedly connected with a side protection plate, and the top of the side protection plate is fixedly connected with a top support plate; by portable storage of liquid generated during dehydration, the workload of operators can be reduced, and the continuous contact between the liquid and the surface of the latex mattress is reduced, thereby improving the dehydration effect; by heating with a heating wire, the heat is blown into a ventilation slot with a ventilation fan, thereby improving the drying efficiency of the latex mattress; by using a hollow metal plate, the temperature of the latex mattress after drying can be evenly reduced, and by using a plurality of circulation pipes to circulate the liquid, the efficiency of cooling the latex mattress can be improved, and the wind force generated by the ventilation fan driven by a motor body will blow toward the latex mattress, which also accelerates the cooling efficiency of the latex mattress.

Description

Technical Field

[0001] The invention belongs to the technical field of latex mattresses, and specifically relates to a latex mattress preparation device and a method thereof. Background Art

[0002] A latex mattress is a mattress made of natural latex material. It has good elasticity and air permeability, can effectively relieve body pressure and improve sleep quality. The advantage of a latex mattress is that it can provide better support and comfort. It also has anti-bacterial, anti-mite and anti-allergic properties. The latex mattress is formed by evaporation molding and has countless pores, which have certain air permeability. Because the surface of the pores is smooth, mites cannot attach.

[0003] Existing products contain formaldehyde, which is harmful to human health. Cotton pads, spring mattresses, and palm mattresses have been gradually replaced by sponge pads and latex mattresses. The raw materials of natural latex mattresses are limited, and the raw materials of synthetic latex mattresses and sponge mattresses are petrochemical products, which are non-renewable resources. In addition, the existing process is too simple and the structure is too simple. The air permeability and moisture dissipation of the latex mattresses are poor. After a long time of use, it is very easy to breed bacteria and other microorganisms, emit odors, which is not good for human health and difficult to meet the needs of modernization, humanity and personalization.

[0004] In the prior art, the authorization publication number "CN107513189A" discloses a "method for preparing a sterilization and deodorization latex mattress"; the method comprises mixing the powdered Chinese medicine with the ultrasonically mixed mixture, performing foaming and compression molding in a foaming machine, and finally demolding the latex embryo, taking out the latex embryo from the mold, dehydrating and drying the latex embryo, dehydrating the obtained latex embryo, and then drying to obtain a finished product. The sterilization and deodorization latex mattress of the present invention has good mechanical strength and resilience, as well as good mildew and antibacterial effects, good air permeability, and longer working life, and its preparation method is simple and easy to industrialize, thereby solving the above-mentioned problems.

[0005] However, the above-mentioned "a method for preparing a sterilizing and deodorizing latex mattress" still has some disadvantages, for example: in the step of squeezing the latex mattress, when the latex mattress is squeezed, the liquid will penetrate into or adhere to the equipment or the floor, increasing the cleaning and maintenance work of the equipment, increasing the workload of the operator, and requiring additional drying equipment, increasing the equipment investment and operating costs, and the drying process is slow, which prolongs the overall production time, and after drying, it is often cooled naturally to achieve the purpose of uniform cooling, but the natural cooling efficiency is too low, affecting the efficiency of latex mattress production;

[0006] For this reason, a latex mattress preparation device and method are proposed here to solve the above-mentioned problems. Summary of the invention

[0007] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a latex mattress preparation device and method, which effectively solves the problem that when the latex mattress is squeezed, liquid will seep into or adhere to the equipment or the floor, increasing the cleaning and maintenance work of the equipment, and requiring additional drying equipment, increasing equipment investment and operating costs, and the natural cooling efficiency is too low, affecting the efficiency of latex mattress production.

[0008] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: a latex mattress preparation device, comprising a processing base, the top of the processing base is fixedly connected to a side protective plate, the top of the side protective plate is fixedly connected to a top supporting plate, one side of the side protective plate is provided with a squeezing table, the bottom of the top supporting plate is provided with a main electric push rod, the bottom of the main electric push rod is movably connected to a top reinforcement plate, the surface of the top reinforcement plate is provided with a top through frame, the bottom of the top supporting plate is movably connected to a secondary electric push rod, and the bottom of the secondary electric push rod is movably connected to a clamping plate.

[0009] Preferably: a drainage groove is provided on the surface of the squeezing table, a liquid storage box is fixedly connected to the bottom of the squeezing table, connecting holes are provided on both sides of the liquid storage box, a liquid guiding elbow is movably connected to one side of the connecting hole, one end of the liquid guiding elbow is movably connected to a liquid collecting box, and one side of the liquid collecting box is fixedly connected to one side of the liquid storage box.

[0010] Preferably: one side of the processing base is fixedly connected to an outer bellows, one side of the outer bellows is fixedly connected to a support platform, the top of the support platform is fixedly connected to a motor body, the output shaft of the motor body is movably connected to a ventilation fan, one side of the outer bellows is fixedly connected to a supporting inner frame, and one side of the supporting inner frame is fixedly connected to an outer protective frame.

[0011] Preferably: a drying box is fixedly connected to one side of the outer wind box, a temperature controller is fixedly connected to one side of the drying box, a connector is fixedly connected to the inside of the drying box, a transmission cable is movably connected to the inside of the connector, and a heating wire is movably connected to one end of the transmission cable.

[0012] Preferably, an air outlet box is fixedly connected to the other side of the outer air box, one side of the air outlet box is fixedly connected to one side of the squeezing table, and a ventilation slot is provided on the inner side of the squeezing table.

[0013] Preferably, there are multiple ventilation slots, and one side of each of the multiple ventilation slots is fixedly connected with a water baffle.

[0014] Preferably, one side of the liquid storage tank is movably connected to a circulation pipe, one end of the circulation pipe is movably connected to a water pump box, and one side of the water pump box is movably connected to a water pump body.

[0015] Preferably, a top heat dissipation groove is provided on the surface of the top reinforcement plate, a hollow metal plate is fixedly connected to the inner side of the top heat dissipation groove, a heat sink is fixedly connected to the top of the hollow metal plate, and the number of the heat sink is multiple.

[0016] Preferably: sliding blocks are fixedly connected to both sides of the top reinforcement plate, side sliding rails are provided on the surface of the side protection plate, and the inner side of the side sliding rails is slidably connected to one side of the sliding block.

[0017] A method for using a latex mattress preparation device comprises the following steps:

[0018] S1. First, place the latex mattress to be dehydrated on the surface of the squeezing table. After the mattress is placed on the surface of the squeezing table, the main electric push rod at the bottom of the top support plate is started, and the top reinforcement plate can be driven to move vertically downward by starting the two main electric push rods. When the top reinforcement plate moves downward, it will drive the sliding block to move synchronously. The sliding connection between the sliding block and the side sliding rail can improve the stability of the top reinforcement plate during the vertical movement. When the top reinforcement plate moves downward, it will contact the surface of the latex mattress;

[0019] S2. After the liquid in the latex mattress is squeezed, the liquid will flow into the surface of the squeeze table and into the drainage groove opened on the surface of the squeeze table. When the liquid flows along the inner side of the drainage groove, it will flow into the inner side of the liquid collection box. Through the connection of the liquid guide elbow, the liquid entering the liquid collection box will flow into the liquid storage box along the liquid guide elbow and be stored in the liquid storage box;

[0020] S3. While the main electric push rod drives the top reinforcement plate to descend, the auxiliary electric push rod synchronously drives the clamping plate to move downward, keeping the clamping plate and the top reinforcement plate descending synchronously, and the clamping plate is kept on the inner side of the top through frame at this time. After the top reinforcement plate is used for squeezing and dehydration, the main electric push rod drives the top reinforcement plate to move vertically upward along the side protection plate, and the auxiliary electric push rod maintains the position of the clamping plate. At this time, after the top reinforcement plate rises, it will continue to clamp the latex mattress on the surface of the squeezing table through four clamping plates;

[0021] S4. After the motor body is started, the motor body will drive the ventilation fan on the inner side of the supporting inner frame to rotate. The rotating ventilation fan will rotate along the inner side of the supporting inner frame and generate wind force in the outer wind box. The generated wind force will blow into the ventilation slot along the outlet wind box. After starting the temperature controller on one side of the drying box, the temperature controller drives the heating wire on one side of the connector to heat through the connection of the transmission cable. The heating wire can generate heat in the drying box. The heat will enter the outer wind boxes on both sides along the drying box and blow into the ventilation slot along the ventilation fan.

[0022] S5. When the dried latex mattress needs to be cooled, the water pump body on one side of the water pump box is first started, and the liquid in the liquid storage tank is driven by the water pump body to flow into the hollow metal plate along the water pump body. When the top reinforcement plate is driven down by the main electric push rod, the hollow metal plate inside the top heat dissipation groove is driven to descend and contact the surface of the latex mattress, and the liquid is circulated using multiple circulation pipes.

[0023] Compared with the prior art, the present invention has the following beneficial effects:

[0024] 1) During the operation of the latex mattress preparation device and method thereof, after the main electric push rod at the bottom of the top supporting plate is started, the top reinforcing plate can be driven to move vertically downward by starting the two main electric push rods, and the top reinforcing plate will drive the sliding block to move synchronously when moving downward, and the sliding connection between the sliding block and the side sliding rail can improve the stability of the top reinforcing plate during the vertical movement, and the top reinforcing plate will contact the surface of the latex mattress when moving downward. After continuous downward movement, the top reinforcing plate can be used to squeeze and dehydrate the latex mattress, and the portable main electric push rod can be used to push synchronously, which is suitable for latex mattresses of different thicknesses and improves the dehydration effect and portability of the latex mattress;

[0025] 2) During the operation of the latex mattress preparation device and method, the liquid will flow into the surface of the squeeze table during the squeezing process, and then flow into the drainage groove provided on the surface of the squeeze table. Through the connection of the liquid guide elbow, the liquid entering the liquid collection box will flow into the liquid storage box along the liquid guide elbow, and will be stored in the liquid storage box. By storing the liquid generated during dehydration in a portable manner, the workload of the operator can be reduced, and the continuous contact between the liquid and the surface of the latex mattress can be reduced, thereby improving the dehydration effect.

[0026] 3) During the operation of the latex mattress preparation device and method, the heating wire is used to generate heat in the drying box. The heat will enter the outer wind boxes on both sides of the drying box and be blown into the ventilation slots by the ventilation fans to dry the latex mattress, thereby improving the drying efficiency of the latex mattress. At the same time, the latex mattress is clamped by four clamping plates, so that most of the surface area of ​​the latex mattress can be exposed, thereby improving the gas flowability of the latex mattress during the drying process and improving the drying effect.

[0027] 4) When the latex mattress preparation device and method are working, after the liquid enters the hollow metal plate, the temperature of the bottom of the hollow metal plate will be reduced, and the heat sink on the top of the hollow metal plate is used to increase the heat dissipation area of ​​the hollow metal plate. The hollow metal plate can evenly reduce the temperature of the latex mattress after drying, and the liquid is circulated by multiple circulation pipes, which can improve the efficiency of cooling the latex mattress. At this time, after the temperature controller is turned off, the wind force generated by the motor body driving the ventilation fan will blow towards the latex mattress, which also speeds up the cooling efficiency of the latex mattress;

[0028] 5) During the dehydration process of the latex mattress, the moisture distribution of the latex mattress is monitored in real time using a reinforcement learning algorithm, and the pressure and moving speed of the top reinforcement plate 501 are dynamically adjusted according to the humidity changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0030] Figure 1 It is a schematic diagram of the overall appearance structure of the present invention;

[0031] Figure 2 It is a schematic diagram of the top support plate structure of the present invention;

[0032] Figure 3 It is a schematic diagram of the structure of the liquid collecting box of the present invention;

[0033] Figure 4 It is a schematic diagram of the structure of the auxiliary electric push rod of the present invention;

[0034] Figure 5 It is a schematic diagram of the structure of the water retaining plate of the present invention;

[0035] Figure 6 It is a schematic diagram of the structure of the liquid drain tank of the present invention;

[0036] Figure 7 It is a schematic diagram of the structure of the liquid storage tank of the present invention;

[0037] Figure 8 It is a schematic diagram of the circulation pipe structure of the present invention;

[0038] Fig. 9 It is a schematic diagram of the structure of the outer bellows of the present invention;

[0039] Fig.10 It is a schematic diagram of the drying box structure of the present invention;

[0040] Fig.11 The optimized algorithm flow chart of the dehydration process of the present invention;

[0041] In the figure: 1, processing base; 2, side protection plate; 3, top support plate; 4, squeeze dryer; 501, top reinforcement plate; 502, main electric push rod; 503, sliding block; 504, side sliding rail; 601, drainage tank; 602, liquid collection box; 603, connecting hole; 604, liquid storage tank; 605, circulation pipe; 606, water pump box; 607, water pump body; 608, hollow metal plate; 609, heat sink; 6010, top heat sink; 6011, top heat sink; 6012, top heat sink; 6013, top heat sink; 6014, top heat sink; 6015, top heat sink; 6016, top heat sink; 6017, top heat sink; 6018, top heat sink; 6019, top heat sink; 602, top heat sink; 603, top heat sink; 604, top heat sink; 605, top heat sink; 606, top heat sink; 607, top heat sink; 608, top heat sink; 609, top heat sink; 6010, top heat sink; 6011, top heat sink; 6012, top heat sink; 6013, top heat sink; 602, top heat sink; 603, top heat sink; 604, top heat sink; 605, top heat sink; 606, top heat sink; 607, top heat sink; 608, top heat sink; 609, top heat sink; 6010, top heat sink; 6011, top heat sink; 602, top heat sink; 603, top heat sink; 604, top heat sink; 605, top heat sink; 606, top heat sink; 606, top heat sink; 607, top heat sink; 011, liquid guiding bend; 701, top frame; 702, auxiliary electric push rod; 703, clamping plate; 801, outer bellows; 802, air outlet box; 803, ventilation slot; 804, water baffle; 805, support platform; 806, motor body; 807, outer protective frame; 808, supporting inner frame; 809, ventilation fan; 901, drying box; 902, temperature controller; 903, connector; 904, transmission cable; 905, heating wire. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] Embodiment 1: In this embodiment, Figure 1-Figure 10 Given, the present invention provides the following technical solutions:

[0044] A latex mattress preparation device includes a processing base 1, the top of the processing base 1 is fixedly connected to a side protection plate 2, the top of the side protection plate 2 is fixedly connected to a top support plate 3, one side of the side protection plate 2 is provided with a squeezing table 4, the bottom of the top support plate 3 is provided with a main electric push rod 502, the bottom of the main electric push rod 502 is movably connected to a top reinforcement plate 501, the surface of the top reinforcement plate 501 is provided with a top through frame 701, the bottom of the top support plate 3 is movably connected to an auxiliary electric push rod 702, and the bottom of the auxiliary electric push rod 702 is movably connected to a clamping plate 703.

[0045] It should be noted that by clamping the latex mattress through four clamping plates 703, most of the surface area of ​​the latex mattress can be exposed, thereby improving the gas flowability of the latex mattress during the drying process and improving the drying effect.

[0046] In this embodiment, a liquid drainage groove 601 is provided on the surface of the squeezing table 4, and a liquid storage box 604 is fixedly connected to the bottom of the squeezing table 4. Communication holes 603 are provided on both sides of the liquid storage box 604. A liquid guide bend 6011 is movably connected to one side of the communication hole 603, and one end of the liquid guide bend 6011 is movably connected to a liquid collecting box 602. One side of the liquid collecting box 602 is fixedly connected to one side of the liquid storage box 604.

[0047] It should be noted that when the liquid flows along the inner side of the drainage trough 601, it will flow into the inner side of the liquid collecting box 602. Through the connection of the liquid guiding bend 6011, the liquid entering the liquid collecting box 602 will enter the liquid storage box 604 along the liquid guiding bend 6011 and be stored in the liquid storage box 604. By storing the liquid generated during dehydration in a portable manner, the workload of the operator can be reduced.

[0048] In this embodiment, an outer bellows 801 is fixedly connected to one side of the processing base 1, a support platform 805 is fixedly connected to one side of the outer bellows 801, a motor body 806 is fixedly connected to the top of the support platform 805, a ventilation fan 809 is movably connected to the output shaft of the motor body 806, an inner support frame 808 is fixedly connected to one side of the outer bellows 801, and an outer protective frame 807 is fixedly connected to one side of the inner support frame 808.

[0049] It should be noted that the motor body 806 can be supported by the support platform 805 , and the outer protective frame 807 can reduce the contact between external debris and the ventilation fan 809 .

[0050] In this embodiment, a drying box 901 is fixedly connected to one side of the outer bellows 801, a temperature controller 902 is fixedly connected to one side of the drying box 901, a connector 903 is fixedly connected to the inner side of the drying box 901, a transmission cable 904 is movably connected to the inner side of the connector 903, and a heating wire 905 is movably connected to one end of the transmission cable 904.

[0051] It should be noted that heat can be generated in the drying box 901 by heating the heating wire 905. The heat will enter the outer wind boxes 801 on both sides of the drying box 901, and will be blown into the ventilation slot 803 along the ventilation fan 809, and will dry the latex mattress, thereby improving the drying efficiency of the latex mattress.

[0052] In this embodiment, the other side of the outer bellows 801 is fixedly connected to an outlet bellows 802 , one side of the outlet bellows 802 is fixedly connected to one side of the wringing table 4 , and a ventilation slot 803 is provided on the inner side of the wringing table 4 .

[0053] It should be noted that the rotating ventilation fan 809 will rotate along the inner side of the supporting inner frame 808 and generate wind in the outer bellows 801. The generated wind will blow into the ventilation slot 803 along the air outlet box 802. The wind will blow out along the drainage slot 601 on one side of the ventilation slot 803 and contact the latex mattress, thereby improving the heat dissipation or drying effect of the latex mattress.

[0054] In this embodiment, there are multiple ventilation slots 803 , and one side of each of the multiple ventilation slots 803 is fixedly connected with a water baffle 804 .

[0055] It should be noted that the water baffle 804 on one side of the ventilation slot 803 can reduce the situation where the liquid falls into the drainage slot 601 and enters the ventilation slot 803 .

[0056] In this embodiment, a circulation pipe 605 is movably connected to one side of the liquid storage tank 604 , one end of the circulation pipe 605 is movably connected to a water pump box 606 , and one side of the water pump box 606 is movably connected to a water pump body 607 .

[0057] It should be noted that after the water pump body 607 on one side of the water pump box 606 is started, the liquid in the liquid storage tank 604 will be driven by the water pump body 607 to flow into the hollow metal plate 608 along the water pump body 607. After the liquid enters the hollow metal plate 608, the temperature at the bottom of the hollow metal plate 608 will be lowered, and the liquid can be circulated by using multiple circulation pipes 605, and the circulating liquid improves the heat dissipation effect.

[0058] In this embodiment, a top heat dissipation groove 6010 is opened on the surface of the top reinforcement plate 501, a hollow metal plate 608 is fixedly connected to the inner side of the top heat dissipation groove 6010, a heat sink 609 is fixedly connected to the top of the hollow metal plate 608, and the number of heat sinks 609 is multiple.

[0059] It should be noted that after the liquid enters the hollow metal plate 608 , the temperature at the bottom of the hollow metal plate 608 will be lowered, and the heat sink 609 on the top of the hollow metal plate 608 will increase the heat dissipation area of ​​the hollow metal plate 608 and improve the heat dissipation effect of the hollow metal plate 608 .

[0060] In this embodiment, sliding blocks 503 are fixedly connected to both sides of the top reinforcement plate 501 , and side sliding rails 504 are provided on the surface of the side protection plate 2 , and the inner side of the side sliding rail 504 is slidably connected to one side of the sliding block 503 .

[0061] It should be noted that, through the sliding connection between the sliding block 503 and the side sliding rail 504 , the stability of the top reinforcement plate 501 during the vertical movement can be improved.

[0062] Embodiment 2: Embodiment 2 provides a method for using a latex mattress preparation device, which is used to further illustrate the working process or principle of the latex mattress preparation device provided in the above embodiment 1, and the details are as follows:

[0063] A latex mattress preparation device and method thereof, comprising the following steps:

[0064] S1. First, place the latex mattress to be dehydrated on the surface of the squeezing table 4. After placing the mattress on the surface of the squeezing table 4, the main electric push rod 502 at the bottom of the top support plate 3 is started. The top reinforcing plate 501 can be driven to move vertically downward by starting the two main electric push rods 502. When the top reinforcing plate 501 moves downward, it will drive the sliding block 503 to move synchronously. The sliding connection between the sliding block 503 and the side sliding rail 504 can improve the stability of the top reinforcing plate 501 during the vertical movement. When the top reinforcing plate 501 moves downward, it will contact the surface of the latex mattress. After continuous downward movement, the top reinforcing plate 501 can be used to squeeze and dehydrate the latex mattress.

[0065] S2. During the process of squeezing and dehydrating by using the top reinforcing plate 501, after the liquid in the latex mattress is squeezed, the liquid will flow into the surface of the squeezing table 4 and into the drainage groove 601 opened on the surface of the squeezing table 4. When the liquid flows along the inner side of the drainage groove 601, it will flow into the inner side of the liquid collecting box 602. Through the connection of the liquid guiding bend 6011, the liquid entering the liquid collecting box 602 will enter the liquid storage box 604 along the liquid guiding bend 6011 and be stored in the liquid storage box 604. By storing the liquid generated during dehydration in a portable manner, the workload of the operator can be reduced, and the continuous contact between the liquid and the surface of the latex mattress can be reduced, thereby improving the dehydration effect.

[0066] S3. While the main electric push rod 502 drives the top reinforcing plate 501 to descend, the auxiliary electric push rod 702 synchronously drives the clamping plate 703 to move downward, and keeps the clamping plate 703 and the top reinforcing plate 501 descending synchronously, and the clamping plate 703 is kept on the inner side of the top through frame 701 at this time. After the top reinforcing plate 501 is used for squeezing and dehydration, the main electric push rod 502 drives the top reinforcing plate 501 to move vertically upward along the side protection plate 2, and the auxiliary electric push rod 702 keeps the position of the clamping plate 703. At this time, after the top reinforcing plate 501 rises, the four clamping plates 703 will continue to clamp the latex mattress on the surface of the squeezing table 4;

[0067] S4, at this time, after the motor body 806 is started, the motor body 806 will drive the ventilation fan 809 on the inner side of the supporting inner frame 808 to rotate, and the rotating ventilation fan 809 will rotate along the inner side of the supporting inner frame 808, and generate wind force in the outer bellows 801, and the generated wind force will be blown into the ventilation slot 803 along the outlet box 802, and the water baffle 804 on one side of the ventilation slot 803 can reduce the situation that the liquid falls into the drainage slot 601 and enters the ventilation slot 803, and the wind force will be blown out along the drainage slot 601 on one side of the ventilation slot 803, and at the same time, after the temperature controller 902 on one side of the drying box 901 is started, the transmission The connection of the cable 904 can drive the heating wire 905 on one side of the connector 903 to heat through the temperature controller 902. The heating of the heating wire 905 can generate heat in the drying box 901. The heat will enter the outer wind boxes 801 on both sides along the drying box 901, and will be blown into the ventilation slot 803 along the ventilation fan 809, and will be directed to the latex mattress for drying, thereby improving the efficiency of drying the latex mattress. At the same time, the latex mattress is clamped by four clamping plates 703, so that most of the surface area of ​​the latex mattress can be exposed, thereby improving the air flow of the latex mattress during the drying process and improving the drying effect.

[0068] S5. When the latex mattress needs to be cooled after drying, the water pump body 607 on one side of the water pump box 606 is first started, and the liquid in the liquid storage tank 604 is driven by the water pump body 607 to flow into the hollow metal plate 608 along the water pump body 607. After the liquid enters the hollow metal plate 608, the temperature at the bottom of the hollow metal plate 608 is reduced, and the heat dissipation area of ​​the hollow metal plate 608 is increased by using the heat sink 609 on the top of the hollow metal plate 608. At this time, the main electric push rod 502 is driven to cool the latex mattress. When the dynamic top reinforcement plate 501 descends, it will drive the hollow metal plate 608 inside the top heat dissipation groove 6010 to descend and contact the surface of the latex mattress. The hollow metal plate 608 can evenly reduce the temperature of the latex mattress after drying, and the liquid can be circulated by multiple circulation pipes 605 to improve the efficiency of cooling the latex mattress. At this time, after turning off the temperature controller 902, the wind generated by the ventilation fan 809 driven by the motor body 806 will blow towards the latex mattress, which also speeds up the cooling efficiency of the latex mattress.

[0069] Embodiment 3: During the dehydration process of the latex mattress, the moisture distribution of the latex mattress is monitored in real time using a reinforcement learning algorithm, and the pressure and movement speed of the top reinforcement plate 501 are dynamically adjusted according to the humidity changes.

[0070] Step 1: Install humidity sensors on the surface of the top reinforcement plate 501 and the squeeze table 4 to form a humidity sensing network. The sensors collect humidity data of the latex mattress in real time and transmit it to the control system;

[0071] 1.1 Establishment of humidity sensing network

[0072] A number of humidity sensors are installed on the surface of the top reinforcing plate 501 and the squeezing table 4 respectively; these sensors are used to monitor the humidity distribution of the latex mattress in real time to ensure that the data covers all areas of the mattress; the specific steps are as follows:

[0073] Select high-precision humidity sensors; ensure that the sensors have fast response time and high sensitivity to reflect humidity changes in real time;

[0074] Evenly distributed humidity sensors are installed on the bottom of the top reinforcement plate (501) and the surface of the squeeze table (4). Here, 16 sensors are installed on each plate to form a 4x4 matrix. The position coordinates of each sensor are expressed as , where i is the sensor number;

[0075] The sensor is connected to the control system through a wire; the real-time collected humidity data is transmitted to the central processing unit (CPU) of the control system for data processing and analysis;

[0076] 1.2 Humidity data collection and processing

[0077] Humidity sensors collect data in real time to ensure the real-time and accuracy of the data;

[0078] The data format is a two-dimensional matrix, and the currently collected data is:

[0079] ;

[0080] in is the humidity data of the sensor in the i-th row and j-th column;

[0081] Step 2 preprocesses the data collected by the humidity sensor, including data cleaning and normalization operations; uses time series analysis methods to analyze the changing trend of humidity data;

[0082] 2.1 Data Preprocessing

[0083] 2.1.1 Data cleaning includes

[0084] Remove outliers: Filter the data collected by the humidity sensor and remove outliers. Outliers are detected by statistical methods such as Z-score (standard score). If the Z-score of a data point exceeds the set threshold, it is considered an outlier.

[0085] Calculate the Z-score for each data point:

[0086] ;

[0087] in is the humidity value, is the mean, is the standard deviation;

[0088] Interpolation processing: For detected outliers or missing values, linear interpolation is used to fill them to ensure the continuity of the data;

[0089] The linear interpolation formula is as follows:

[0090] ;

[0091] in is the humidity value, representing the humidity value at different locations;

[0092] 2.1.2 Data Normalization

[0093] Normalize the humidity data to between 0 and 1 to eliminate the dimension effect and improve the efficiency and effect of data processing; the normalization method here is Min-Max normalization:

[0094] The normalization formula is:

[0095] ;

[0096] in and are the minimum and maximum values ​​of the data set respectively; Represents the normalized humidity data, which represents the data of the sensor in the i-th row and j-th column. The normalized value is between 0 and 1. Represents the original humidity data, which represents the data of the sensor in the i-th row and j-th column, and is the unnormalized value;

[0097] 2.2 Time Series Analysis

[0098] Use the sliding average method to smooth the humidity data and reduce the impact of noise; sliding average formula:

[0099] ;

[0100] in is the smoothed humidity data, which means the value of the humidity data at time point t after sliding average; N is the size of the sliding window, that is, the number of humidity data points used to calculate the average value; is the original humidity data, indicating the humidity value at time point i; t is the current time point; is the summation operation from time point t-N+1 to time point t;

[0101] Step 3: Select a deep reinforcement learning algorithm. Here, we use the deep Q network (DQN) and the deep deterministic policy gradient (DDPG) algorithm. Establish a reinforcement learning environment and define the state: (humidity data); action: (pressure and moving speed of the top reinforcement plate) and reward (dehydration effect). Use historical data to train the reinforcement learning model and optimize the pressure and moving speed of the top reinforcement plate 501 to achieve the best dehydration effect.

[0102] 3.1 Choosing a Deep Reinforcement Learning Algorithm

[0103] Deep Q Network (DQN): DQN is a value function method that selects the best action by estimating the state-action value function (Q value); it is applicable to discrete action spaces;

[0104] Deep Deterministic Policy Gradient (DDPG): DDPG is a policy gradient method that combines the value function approximation of DQN and the continuous action selection ability of policy gradient, and is suitable for continuous action space;

[0105] 3.2 Establishing a reinforcement learning environment

[0106] 3.2.1 State (State,s):

[0107] Defined as the real-time humidity data of each position of the latex mattress; the data collected by the humidity sensor network is a two-dimensional matrix H:

[0108] ;

[0109] 3.2.2 Action (Action, a):

[0110] Defined as the pressure P and moving speed V of the top reinforcement plate 501, the action vector is assumed to be:

[0111] ;

[0112] 3.2.3 Reward (r):

[0113] Defined as the dehydration effect, expressed as the total amount of humidity reduction:

[0114] ;

[0115] Among them, r is the reward, which represents the dehydration effect, which is specifically expressed as the total amount of humidity reduction; is the humidity data of the sensor in row i and column j before dehydration is the humidity data of the sensor in the i-th row and j-th column after dehydration; m is the number of rows in the humidity data matrix; n is the number of columns in the humidity data matrix To sum the rows of the humidity data matrix To sum the columns of the humidity data matrix;

[0116] 3.3 Training reinforcement learning models using historical data

[0117] 3.3.1DQN model training

[0118] Q-value update formula: DQN approximates the state-action value function through a neural network ;

[0119] , and update the Q value using the Bellman equation:

[0120] ;

[0121] in For the current state Next, select Action Q value; is the current state, representing the humidity data matrix; is the current action, indicating the pressure and moving speed of the top reinforcement plate 501; is the learning rate, which controls the step size of Q value update; is the current reward, indicating the dehydration effect; is a discount factor used to measure the importance of future rewards, ranging from 0 to 1; For the next state Next, select Action The maximum Q value after For the next state Next, select Action The Q value at this time is determined by the Q network parameter It is calculated that For possible actions It is the weighted value of TD error (TemporalDifference Error), which is used to update the current Q value;

[0122] Neural network architecture: The input layer is the humidity data matrix H; the hidden layer is multiple convolutional layers and fully connected layers for feature extraction; the output layer is the Q value of different actions;

[0123] Experience replay mechanism: storing a quadruple of past states, actions, rewards, and next states ; Randomly extract samples from the experience pool and calculate the loss function ;

[0124] in is the loss function, which indicates that the model is under the current parameters The error below are the parameters of the current Q network; the weights and biases of the neural network;

[0125] is the expectation operator, which represents the expected value of the sample drawn from the experience pool D; D is the experience pool, which stores the four-tuple of past state, action, reward and next state;

[0126] is a quaternion, representing a sample in the experience pool, including s is the current state, representing the humidity data matrix; a is the current action, representing the pressure and moving speed of the top reinforcement plate 501; r is the current reward, representing the dehydration effect, usually the total amount of humidity reduction; is the next state, which represents the humidity data matrix after the action is executed; is the target Q value, indicating that in the next state Next, select Action The maximum Q value after adding the current reward r; is a discount factor used to measure the importance of future rewards, ranging from 0 to 1; For the next state Next, use the target Q network parameters Calculate the maximum Q value; is the parameter of the target Q network, which is used to calculate the stable target Q value; is the current Q value, which means the Q value when action a is selected in the current state s, which is determined by the current Q network parameters Calculated; is the square of the TD error, which represents the difference between the target Q value and the current Q value; the smaller the TD error, the more accurate the model;

[0127] Gradient descent method updates network parameters:

[0128] ;

[0129] in represents the gradient;

[0130] 3.3.2DDPG model training

[0131] Policy Network and Q Network: Policy Network Output continuous action a; Q network Estimate state-action value function;

[0132] Q value update formula:

[0133] ;

[0134] in, and are the target policy network and target Q network parameters respectively, is the parameter of Q network, and the meanings of other symbols have been defined above;

[0135] Policy network update: Use the policy gradient method to update the policy network parameters ;

[0136] in is the policy network parameter The gradient of the objective function J on the policy network parameters The partial derivative of is the expectation operator, which indicates the The expected value of the state s extracted from ; To represent the state s from the experience pool Extraction; Experience Pool A 4-tuple storing the past state, action, reward, and next state; is the Q-value function of action a The gradient of represents the impact of action a on the Q value in state s; it is used to evaluate the performance of actions under the current strategy in the Q value function; Output actions for the policy network Policy network parameters The gradient of , which means that in state s, the action output by the policy network is proportional to the parameter The impact of

[0137] Experience replay mechanism: storing a quadruple of past states, actions, rewards, and next states ; Randomly extract samples from the experience pool, calculate the loss function of the Q network and the policy network, and update the parameters;

[0138] 3.3.3 Model Ensemble

[0139] Integration of DQN and DDPG: In the initial stage, the DQN model is used to perform rough optimization in the discrete action space to find the preliminary optimal pressure and moving speed range; the DDPG model is used to perform fine adjustment in the continuous action space to further optimize the pressure and moving speed of the top reinforcement plate (501);

[0140] 3.4 Improved DQN: Double Deep Q Network (Double DQN)

[0141] The standard DQN algorithm has an overestimation problem when updating the Q value, that is, when calculating the maximum Q value of the next state, it will choose an overestimated action, resulting in inaccurate Q value update; Double DQN alleviates this problem by introducing two Q networks; the specific formula is derived as follows

[0142] Double Q network: Q network is used to select actions; target Q network is used to calculate target Q value;

[0143] Q value update formula: The target Q value update formula of the standard DQN is as follows:

[0144] ;

[0145] The target Q value update formula of dual DQN is as follows:

[0146] ;

[0147] in, are the parameters of the Q network, are the parameters of the target Q network; is the current reward, indicating the dehydration effect; is a discount factor used to measure the importance of future rewards;

[0148] 3.5 Improvement (DDPG): Soft Updates

[0149] In DDPG, the update of target network parameters is usually performed through hard copy, which will cause the target network to be unstable; soft update maintains the stability of the target network by gradually adjusting the target network parameters; the specific formula is derived as follows:

[0150] ;

[0151] in, is the step size of soft update, are the parameters of the Q network, are the parameters of the target Q network, are the parameters of the policy network, are the parameters of the target policy network;

[0152] Step 4: Acquire the data of the humidity sensor in real time, input it into the trained reinforcement learning model, and control the action of the main electric push rod 502 according to the optimal pressure and moving speed output by the model to achieve precise squeezing and dehydration;

[0153] Model input: Input the preprocessed humidity data matrix H into the trained reinforcement learning model;

[0154] State definition: The humidity data matrix H is input into the model as the state s; the model outputs the optimal pressure P and moving speed V of the top reinforcement plate 501: ;

[0155] Action execution: The control system controls the action of the main electric push rod 502 according to the optimal pressure and moving speed output by the model;

[0156] Control process: set the initial position and state of the main electric push rod 502 to ensure that the equipment is in a safe working state; set the working parameters of the main electric push rod 502 according to the optimal pressure P and moving speed V output by the model; start the main electric push rod 502 to drive the top reinforcement plate 501 to move vertically downward and apply the calculated pressure; control the top reinforcement plate 501 to move at a speed of 5 mm / s to ensure a smooth and uniform extrusion process;

[0157] Real-time monitoring and feedback: During the squeezing and dehydration process, the humidity sensor continues to monitor the humidity data of the latex mattress in real time and transmits it to the control system. The control system adjusts the pressure and moving speed of the top reinforcement plate 501 according to the real-time humidity data to ensure the best dehydration effect;

[0158] It should be noted that the main electric push rod 502, the auxiliary electric push rod 702, the water pump body 607, the ventilation fan 809, the temperature controller 902, the transmission cable 904 and the heating wire 905 in the present invention are all prior arts, and the corresponding models can be selected according to actual needs. The internal structure and operating principle of the above parts are also common knowledge of technicians in this field, and no further elaboration will be made on this.

[0159] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A latex mattress preparation device, comprising a processing base (1), characterized in that: The top of the processing base (1) is fixedly connected to a side protection plate (2), the top of the side protection plate (2) is fixedly connected to a top support plate (3), one side of the side protection plate (2) is provided with a squeeze table (4), the bottom of the top support plate (3) is provided with a main electric push rod (502), the bottom of the main electric push rod (502) is movably connected to a top reinforcement plate (501), the surface of the top reinforcement plate (501) is provided with a top through frame (701), the bottom of the top support plate (3) is movably connected to a secondary electric push rod (702), and the bottom of the secondary electric push rod (702) is movably connected to a clamping plate (703); A liquid drainage groove (601) is provided on the surface of the squeezing table (4); a liquid storage box (604) is fixedly connected to the bottom of the squeezing table (4); communication holes (603) are provided on both sides of the liquid storage box (604); a liquid guide bend (6011) is movably connected to one side of the communication hole (603); one end of the liquid guide bend (6011) is movably connected to a liquid collection box (602); and one side of the liquid collection box (602) is fixedly connected to one side of the liquid storage box (604); One side of the processing base (1) is fixedly connected to an external bellows (801), one side of the external bellows (801) is fixedly connected to a support platform (805), the top of the support platform (805) is fixedly connected to a motor body (806), an output shaft of the motor body (806) is movably connected to a ventilation fan (809), one side of the external bellows (801) is fixedly connected to an inner support frame (808), and one side of the inner support frame (808) is fixedly connected to an external protective frame (807); A drying box (901) is fixedly connected to one side of the outer wind box (801), a temperature controller (902) is fixedly connected to one side of the drying box (901), a connector (903) is fixedly connected to the inside of the drying box (901), a transmission cable (904) is movably connected to the inside of the connector (903), and a heating wire (905) is movably connected to one end of the transmission cable (904); The other side of the outer wind box (801) is fixedly connected to an outlet wind box (802), one side of the outlet wind box (802) is fixedly connected to one side of the wringing platform (4), and a ventilation slot (803) is provided on the inner side of the wringing platform (4); There are a plurality of ventilation slots (803), and one side of each of the plurality of ventilation slots (803) is fixedly connected to a water baffle plate (804); One side of the liquid storage tank (604) is movably connected to a circulation pipe (605), one end of the circulation pipe (605) is movably connected to a water pump box (606), and one side of the water pump box (606) is movably connected to a water pump body (607); A top heat dissipation groove (6010) is provided on the surface of the top reinforcement plate (501), a hollow metal plate (608) is fixedly connected to the inner side of the top heat dissipation groove (6010), a heat sink (609) is fixedly connected to the top of the hollow metal plate (608), and the number of the heat sink (609) is plural; Sliding blocks (503) are fixedly connected to both sides of the top reinforcing plate (501), and side sliding rails (504) are provided on the surface of the side protection plate (2), and the inner side of the side sliding rail (504) is slidably connected to one side of the sliding block (503).

2. A method for using a latex mattress preparation device, applied to the latex mattress preparation device according to claim 1, characterized in that: The steps include: S1. First, a latex mattress to be dehydrated is placed on the surface of the squeezing table (4). After the mattress is placed on the surface of the squeezing table (4), the main electric push rod (502) at the bottom of the top support plate (3) is started, and the top reinforcement plate (501) is driven to move vertically downward by starting the two main electric push rods (502). When the top reinforcement plate (501) moves downward, it will drive the sliding block (503) to move synchronously. The sliding connection between the sliding block (503) and the side sliding rail (504) can improve the stability of the top reinforcement plate (501) during the vertical movement. When the top reinforcement plate (501) moves downward, it will contact the surface of the latex mattress; S2. After the liquid in the latex mattress is squeezed, the liquid flows into the surface of the squeeze table (4) and into the drainage groove (601) provided on the surface of the squeeze table (4). When the liquid flows along the inner side of the drainage groove (601), the liquid flows into the inner side of the liquid collecting box (602). Through the connection of the liquid guiding bend (6011), the liquid entering the liquid collecting box (602) flows along the liquid guiding bend (6011) into the liquid storage box (604) and is stored in the liquid storage box (604); S3, while the main electric push rod (502) drives the top reinforcement plate (501) to descend, the auxiliary electric push rod (702) synchronously drives the clamping plate (703) to move downward, and keeps the clamping plate (703) and the top reinforcement plate (501) descending synchronously, and the clamping plate (703) is kept on the inner side of the top through frame (701). After the top reinforcement plate (501) is squeezed and dehydrated, the main electric push rod (502) drives the top reinforcement plate (501) to move vertically upward along the side protection plate (2), and the auxiliary electric push rod (702) maintains the position of the clamping plate (703). At this time, after the top reinforcement plate (501) rises, it will continue to clamp the latex mattress on the surface of the squeezing table (4) through the four clamping plates (703); S4, after the motor body (806) is started, the motor body (806) drives the ventilation fan (809) on the inner side of the supporting inner frame (808) to rotate, and the rotating ventilation fan (809) rotates along the inner side of the supporting inner frame (808) and generates wind force in the outer wind box (801), and the generated wind force blows into the ventilation slot (803) along the outlet wind box (802), and after the temperature controller (902) on one side of the drying box (901) is started, through the connection of the transmission cable (904), the temperature controller (902) drives the heating wire (905) on one side of the connector (903) to heat, and the heating wire (905) can generate heat in the drying box (901), and the heat enters the outer wind boxes (801) on both sides along the drying box (901), and blows into the ventilation slot (803) along the ventilation fan (809); S5. When the dried latex mattress needs to be cooled, the water pump body (607) on one side of the water pump box (606) is first started, and the liquid in the liquid storage tank (604) is driven by the water pump body (607) to flow into the hollow metal plate (608) along the water pump body (607). When the top reinforcing plate (501) is driven to descend by the main electric push rod (502), the hollow metal plate (608) on the inner side of the top heat dissipation groove (6010) is driven to descend and contact the surface of the latex mattress, and the liquid is circulated using multiple circulation pipes (605).

3. The method for using the latex mattress preparation device according to claim 2, characterized in that: In step S1, during the dehydration process of the latex mattress, the humidity distribution of the latex mattress is monitored in real time using a reinforcement learning algorithm, and the pressure and moving speed of the top reinforcement plate (501) are dynamically adjusted according to the humidity change; the specific steps are: Step 1: humidity sensors are installed on the surface of the top reinforcement plate (501) and the squeeze table (4) to form a humidity sensing network. The sensors collect humidity data of the latex mattress in real time and transmit it to the control system; 1.1 Establishment of humidity sensing network A plurality of humidity sensors are respectively installed on the surface of the top reinforcing plate (501) and the squeezing table (4); these sensors are used to monitor the humidity distribution of the latex mattress in real time to ensure that the data covers all areas of the mattress; the specific steps are as follows: Select high-precision humidity sensors; ensure that the sensors have fast response time and high sensitivity to reflect humidity changes in real time; Evenly distributed humidity sensors are installed on the bottom of the top reinforcement plate (501) and the surface of the squeeze table (4). Here, 16 sensors are installed on each plate to form a 4x4 matrix. The position coordinates of each sensor are expressed as , where i is the sensor number; The sensor is connected to the control system through a wire; the real-time collected humidity data is transmitted to the central processing unit (CPU) of the control system for data processing and analysis; 1.2 Humidity data collection and processing Humidity sensors collect data in real time to ensure the real-time and accuracy of the data; The data format is a two-dimensional matrix, and the currently collected data is: in is the humidity data of the sensor in the i-th row and j-th column; Step 2 preprocesses the data collected by the humidity sensor, including data cleaning and normalization operations; uses time series analysis methods to analyze the changing trend of humidity data; 2.1 Data Preprocessing 2.1.1 Data cleaning includes Remove outliers: Filter the data collected by the humidity sensor and remove outliers. Outliers are detected by statistical methods. If the standard score Z-score of a data point exceeds the set threshold, it is considered an outlier. Calculate the Z-score for each data point: in is the humidity value, μ is the mean, and σ is the standard deviation; Interpolation processing: For detected outliers or missing values, linear interpolation is used to fill them to ensure the continuity of the data; The linear interpolation formula is as follows in is the humidity value, representing the humidity value at different locations; 2.1.2 Data Normalization Normalize the humidity data to between 0 and 1 to eliminate the dimension effect and improve the efficiency and effect of data processing; the normalization method here is Min-Max normalization: The normalized formula is in and are the minimum and maximum values ​​of the data set respectively; Represents the normalized humidity data, which represents the data of the sensor in the i-th row and j-th column. The normalized value is between 0 and 1. Represents the original humidity data, which represents the data of the sensor in the i-th row and j-th column, and is the unnormalized value; 2.2 Time Series Analysis Use the sliding average method to smooth the humidity data and reduce the impact of noise; sliding average formula: in is the smoothed humidity data, which means the value of the humidity data at time point t after sliding average; N is the size of the sliding window, that is, the number of humidity data points used to calculate the average value; is the original humidity data, indicating the humidity value at time point i; t is the current time point; From the time point The summation operation up to time point t; Step 3: Select a deep reinforcement learning algorithm. Here we use the deep Q network DQN and the deep deterministic policy gradient DDPG algorithm. Establish a reinforcement learning environment. define state: humidity data; Action: pressure and moving speed of the top reinforcement plate; Reward: dehydration effect; Use historical data to train the reinforcement learning model to optimize the pressure and moving speed of the top reinforcement plate (501) to achieve the best dehydration effect; 3.1 Choosing a Deep Reinforcement Learning Algorithm Deep Q Network DQN: DQN is a value function method that selects the best action by estimating the state-action value function Q value; it is applicable to discrete action space; Deep Deterministic Policy Gradient DDPG: DDPG is a policy gradient method that combines the value function approximation of DQN and the continuous action selection ability of policy gradient, and is suitable for continuous action space; 3.2 Establishing a reinforcement learning environment 3.2.1 State (State,s): Defined as the real-time humidity data of each position of the latex mattress; the data collected by the humidity sensor network is a two-dimensional matrix H: 3.2.2 Action (Action, a): Defined as the pressure P and moving speed V of the top reinforcement plate (501), the action vector is assumed to be: 3.2.3 Reward (r): Defined as the dehydration effect, expressed as the total amount of humidity reduction: Where r is the reward, which represents the dehydration effect, specifically the total amount of humidity reduction is the humidity data of the sensor in row i and column j before dehydration is the humidity data of the sensor in the i-th row and j-th column after dehydration; m is the number of rows in the humidity data matrix; n is the number of columns in the humidity data matrix To sum the rows of the humidity data matrix To sum the columns of the humidity data matrix; 3.3 Training reinforcement learning models using historical data 3.3.1DQN model training Q-value update formula: DQN approximates the state-action value function through a neural network ; , and update the Q value using the Bellman equation: in For the current state Next, select Action Q value is the current state, representing the humidity data matrix is the current action, indicating the pressure and moving speed of the top reinforcement plate (501); is the learning rate, which controls the step size of Q value update is the current reward, indicating the dehydration effect; γ is a discount factor, which is used to measure the importance of future rewards and ranges from 0 to 1 For the next state Next, select the maximum Q value after action a′ For the next state The Q value when action a′ is selected is calculated by the Q network parameter θ, where a′ is a possible action. It is the weighted value of TD error (TemporalDifference Error), which is used to update the current Q value; Neural network architecture: The input layer is the humidity data matrix H; the hidden layer is multiple convolutional layers and fully connected layers for feature extraction; The output layer is the Q value of different actions; Experience replay mechanism: store the past state, action, reward and next state of the four-tuple (s, a, r, s ′); randomly extract samples from the experience pool and calculate the loss function in is the loss function, which represents the error of the model under the current parameter θ; θ is the parameter of the current Q network; the weight and bias of the neural network; is the expectation operator, which represents the expected value of the sample drawn from the experience pool D; D is the experience pool, which stores the four-tuple of past state, action, reward and next state; (s, a, r, s′) is a four-tuple, representing a sample in the experience pool, including s is the current state, representing the humidity data matrix; a is the current action, representing the pressure and movement speed of the top reinforcement plate (501); r is the current reward, representing the dehydration effect, usually the total amount of humidity reduction; s′ is the next state, which represents the humidity data matrix after the action is executed; is the target Q value, which means the maximum Q value after selecting action a′ in the next state s′ plus the current reward r; γ is the discount factor, which is used to measure the importance of future rewards and ranges from 0 to 1; To use the target Q network parameters in the next state s′ The maximum Q value calculated; θ- is the parameter of the target Q network, which is used to calculate the stable target Q value; is the current Q value, which means the Q value when action a is selected in the current state s, which is calculated by the current Q network parameter θ; is the square of the TD error, which represents the difference between the target Q value and the current Q value; the smaller the TD error, the more accurate the model; Gradient descent method updates network parameters: in represents the gradient; 3.3.2DDPG model training Policy Network and Q Network: Policy Network Output continuous action a; Q network Estimate state-action value function; Q value update formula: in, and are the target policy network and target Q network parameters respectively, is the parameter of Q network, and the meanings of other symbols have been defined above; For expectations; Policy network update: Use the policy gradient method to update the policy network parameters in is the gradient of the policy network parameter θ, which represents the effect of the objective function J on the policy network parameter The partial derivative of is the expectation operator, which indicates the The expected value of the state s extracted from ; To represent the state s from the experience pool Extraction; Experience Pool A 4-tuple storing the past state, action, reward, and next state; is the Q-value function of action a The gradient of represents the impact of action a on the Q value in state s; it is used to evaluate the performance of actions under the current strategy in the Q value function; Output actions for the policy network Policy network parameters The gradient of , which means that in state s, the action output by the policy network is proportional to the parameter The impact of Experience replay mechanism: stores the past state, action, reward, and next state quadruple (s, a, r, s′); randomly extracts samples from the experience pool, calculates the loss function of the Q network and the policy network, and updates the parameters; 3.3.3 Model Ensemble Integration of DQN and DDPG: In the initial stage, the DQN model is used to perform rough optimization in the discrete action space to find the preliminary optimal pressure and moving speed range; the DDPG model is used to perform fine adjustment in the continuous action space to further optimize the pressure and moving speed of the top reinforcement plate (501); 3.4 Improved DQN: Double Deep Q Network (Double DQN) The standard DQN algorithm has an overestimation problem when updating the Q value, that is, when calculating the maximum Q value of the next state, it will choose an overestimated action, resulting in inaccurate Q value update; Double DQN alleviates this problem by introducing two Q networks; the specific formula is derived as follows Double Q network: Q network is used to select actions; target Q network is used to calculate target Q value; Q value update formula: The target Q value update formula of standard DQN is as follows The target Q value update formula of dual DQN is as follows Where θ is the parameter of the Q network, are the parameters of the target Q network; is the current reward, indicating the dehydration effect; γ is the discount factor, which is used to measure the importance of future rewards; 3.5 Improving DDPG: Soft Updates In DDPG, the update of target network parameters is usually performed through hard copy, which will cause the target network to be unstable; soft update gradually adjusts the target network parameters to maintain the stability of the target network; the specific formula is derived as follows Among them, τ is the step size of soft update, are the parameters of the Q network, are the parameters of the target Q network, are the parameters of the policy network, are the parameters of the target policy network; Step 4: acquiring data from the humidity sensor in real time, inputting the data into the trained reinforcement learning model, and controlling the movement of the main electric push rod (502) according to the optimal pressure and moving speed output by the model to achieve precise squeezing and dehydration; Model input: Input the preprocessed humidity data matrix H into the trained reinforcement learning model; State definition: The humidity data matrix H is input into the model as the state s; the model outputs the optimal pressure P and moving speed V of the top reinforcement plate (501): Action execution: The control system controls the action of the main electric push rod according to the optimal pressure and movement speed output by the model; Control process: Set the initial position and state of the main electric push rod to ensure that the equipment is in a safe working state; set the working parameters of the main electric push rod according to the optimal pressure P and moving speed V output by the model; start the main electric push rod to drive the top reinforcement plate to move vertically downward and apply the calculated pressure; control the top reinforcement plate to move at a speed of 5 mm / s to ensure a smooth and uniform extrusion process; Real-time monitoring and feedback: During the squeezing and dehydration process, the humidity sensor continues to monitor the humidity data of the latex mattress in real time and transmits it to the control system. The control system adjusts the pressure and moving speed of the top reinforcement plate according to the real-time humidity data to ensure the best dehydration effect.

Images