Intelligent drying device and drying method for coating machine
By collecting drying process images and state parameters in real time, building sludge and drying optimization models, combining dust removal brushes and exhaust fans, the problems of drum dirt accumulation and harmful gases are solved, and uniform drying of paints and health protection for staff are achieved. .
Patent Information
- Application Number
- CN202311859544.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-12-30
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2043-12-30
AI Technical Summary
During use, dirt accumulates on the drum, resulting in uneven drying of the paint, low working efficiency, and the drying process cannot be monitored in real time, resulting in harmful gases that are harmful to the staff.
Real-time acquisition of drying process images and state parameters are used, and a sludge identification model and drying optimization model are constructed through a convolutional neural network, which realizes automatic sludge judgment and drying process optimization, and sets up dust removal brushes and exhaust fans for cleaning and purification.
Ensure that the paint is evenly dry, improve work efficiency, avoid harmful gases to the health of staff, and realize automation and efficient optimization of the drying process.
Smart Images

Figure CN118179880B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of drying devices, and in particular to an intelligent drying device for a coating machine and a drying method thereof. Background Art
[0002] The intelligent drying device for coating machines can quickly and evenly dry the coated materials after the coating machine is finished working. The intelligent drying device for coating machines can provide efficient and precise drying processing according to different coating materials and requirements, thereby improving production efficiency and product quality.
[0003] However, when the intelligent drying device in the prior art is used:
[0004] 1. Adhesive, dust or other dirt may accumulate on the roller, causing the paint to be not dried evenly. However, cleaning the roller requires suspending the work of the intelligent drying device for the coating machine, which affects the working efficiency of the intelligent drying device for the coating machine.
[0005] 2. The drying process of the drying device cannot be monitored in real time, and harmful gases will be generated during the drying process, which will harm the workers;
[0006] Therefore, there is a need for an intelligent drying device for a coating machine and a drying method thereof. Summary of the invention
[0007] The purpose of the present invention is to provide an intelligent drying device for a coating machine, which solves the problem in the prior art that dirt on a cleaning roller reduces the working efficiency of the intelligent drying device for a coating machine.
[0008] In order to achieve the above object, the present invention adopts the following technical solution: an intelligent drying device for a coating machine, comprising the following steps:
[0009] Step 1: Real-time acquisition of images and state parameters of the coating machine during the drying process, including temperature, air quality parameters and moisture content of the material during the drying stage;
[0010] Step 2: Analyze the images of the drying stage, extract the color and texture features of the material, and build a sludge recognition model to automatically identify the sludge in the drying stage;
[0011] Preferably, the specific process of constructing the sludge identification model is as follows:
[0012] A. Obtain the image of the drying stage within the preset time, perform fuzzy filtering, and then perform multi-channel color feature extraction and high-dimensional material texture feature extraction on the image of the drying stage. The specific process is as follows:
[0013] A1, divide the denoised image equally into multiple regions, where each region image includes height h, width w and channel number features; then define a 3*3 convolution layer to perform convolution operation on the three channels of the region image to obtain the color feature value Sc;
[0014] A2, obtain (R, G, B) channels, deepen the network depth in each channel, extract the material texture features, and obtain the material texture feature value Wl;
[0015] A3, then combine the color feature value Sc with the material texture feature value Wl to generate the feature factor S j ;
[0016] B. Real-time labeling of the drying degree of the image in the drying stage: matching the image in the drying stage with the preset drying degree image, and dividing the drying state according to the drying degree, the drying state is sequentially divided into initial stage, gradual drying stage, color fixing stage and color forming stage, wherein the initial stage indicates that the drying has not been done, the gradual drying stage indicates a low drying degree, the color fixing stage indicates a medium drying degree, and the color forming stage indicates a high drying degree; at the same time, the drying state is sequentially labeled with the drying degree value, which is divided into 1, 2, 3 and 4, wherein 1 corresponds to the initial stage, 2 corresponds to the gradual drying stage, 3 corresponds to the color fixing stage, and 4 corresponds to the color forming stage, and then the drying state is combined with the drying degree value to generate the drying information;
[0017] C, combining the drying information with the characteristic factor and performing weighted summation to obtain the material factor Wn;
[0018] D. Obtain the material factor Wn within the preset time, and compare the material factor Wn with the preset material factor WN, set the gradient difference, and each image area corresponds to a gradient difference, then each area corresponds to a sludge rate; wherein, the error interval of the preset gradient difference is ±wc%, when the gradient difference is not within the error interval, it means that the sludge rate exceeds the standard, then targeted optimization and improvement operations are performed.
[0019] Step 3: Combine the sludge rate and state parameters to build a drying optimization model, realize the judgment and prediction of the drying state, and optimize the drying process;
[0020] Preferably, the specific process of establishing the drying optimization model is as follows:
[0021] S1, obtaining drying information and drying state parameters within a preset time, building a training set, and initializing a neural network structure;
[0022] S2, training the neural network according to the training set, and determining the output value of the objective function of the trained neural network, where the output value of the objective function includes the output value of the hidden layer and the output value of the output layer; the specific process is:
[0023] Sa, takes air quality parameters, temperature, moisture and sludge rate as the network input layer, takes drum speed as the network output layer, and sets the nodes and weights between network layers;
[0024] Sb, sets the error coefficient E, and updates the weights and thresholds between each node in the network structure in real time;
[0025] Sc, generates the output value of the objective function;
[0026] S3, when the training process is stable, the predicted drum speed is obtained in real time, that is, the node corresponding to the output value of the output layer; and the curve change of the input parameter-drum speed during the training process is established, and then normalized to generate the variable limit value JM;
[0027] S4, obtaining the variable limit values JM of the preset time, and grouping them into a set J = {JM1, JM2, JM3, ..., JM D}, where D is a natural number greater than 1, and the maximum variable limit value in the set is obtained, and the time period where the maximum variable limit value is located is recorded as the optimal drying time period, and finally the optimization operation in the drying process is achieved.
[0028] Preferably, the intelligent drying device for the coating machine includes a machine main body, wherein rollers are arranged in sequence inside the machine main body, a plurality of groups of heating ports are fixedly installed inside the machine main body, and the rollers are located below the heating ports, a protective box is fixedly installed on the outer wall of the machine main body, an electric push rod is fixedly installed in the protective box, holes are provided on the protective box, and the protective box is connected with the interior of the machine main body through the holes, one end of the electric push rod is passed through the interior of the machine main body through the hole, a slide groove is provided inside the machine main body, and the slide groove is located below the roller, a moving block is slidably connected in the slide groove, and one end of the electric push rod is fixedly connected to the moving block, a support rod is fixedly connected to the moving block, a bearing plate is fixedly installed on the top of the support rod, a dust removal brush is provided on the bearing plate, and the dust removal brush is in contact with the roller.
[0029] Preferably, a silicone ring is provided in the hole, and one end of the electric push rod passes through the interior of the silicone ring.
[0030] Preferably, a storage box is installed on the top of the machine body, an exhaust fan is installed inside the storage box, an exhaust pipe is connected to the exhaust fan, and one end of the exhaust pipe is passed through the inside of the machine body, an air supply pipe is connected to the exhaust fan, a purification box is fixedly installed on the top of the machine body, and one end of the air supply pipe is connected to the purification box, an iron frame is provided in the purification box, a filter is detachably connected to the iron frame, and an exhaust pipe is connected to the purification box.
[0031] Preferably, a handle is fixedly connected to one end of the iron frame, an opening is provided on the purification box, and the iron frame is slidably connected to the opening via the handle.
[0032] Preferably, a groove is provided inside the purification box, a magnet is provided in the groove, and an iron frame is provided in the groove.
[0033] Preferably, an inspection door is provided on the machine body, and an observation window is provided on the inspection door.
[0034] Preferably, a feed port and a discharge port are symmetrically arranged at both ends of the machine body.
[0035] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:
[0036] 1. By setting a dust removal brush, when cleaning the dust and impurities attached to the roller, first start the electric push rod to push the moving block to move in the slide slot. When the moving block moves, the dust removal brush on the support rod is used to clean the impurities on the rolling roller to prevent the dirt on the roller from affecting the coating and drying, ensuring that the coating can be dried evenly. Cleaning the roller when there is no need to pause the drying device also improves the working efficiency of the intelligent drying device for the coating machine.
[0037] 2. By setting up an exhaust fan and a filter, when filtering the harmful gas generated by high temperature inside the drying device, first start the exhaust fan to make the exhaust pipe extract the gas in the machine body, and then enter the purification box through the air duct, so that the filter in the purification box filters the harmful gas, and the filtered gas is discharged to the outside through the exhaust pipe, avoiding the harmful gas generated by the intelligent drying device of the coating machine during drying and coating to cause harm to the health of the staff, thereby ensuring the health of the staff.
[0038] 3. By collecting images and state parameters of the drying stage, a sludge judgment algorithm based on convolutional neural network is proposed according to the color and material texture change characteristics of the coating machine in the drying stage, so as to realize the automatic judgment of the sludge degree in the drying stage, so as to perform cleaning operations in the process; in this process, the color characteristics of the material are extracted by multi-channel and the high-dimensional texture characteristics of the material are extracted by deepening the network depth, and the sludge recognition model is obtained through algorithm model design and simulation training, and then the trained model is used to extract features of the drying process of the coating material in real time through the image acquisition equipment; in addition, a drying optimization algorithm based on convolutional neural network is proposed according to the drying state of the coating machine, so as to realize the analysis of the drying state of the coating material ..., and the drying state of the coating machine is analyzed, so as to perform cleaning operations in the process; in this process, the color characteristics of the material are extracted by multi-channel and the high-dimensional texture characteristics of the material are extracted by deepening the network depth, and the high-dimensional texture characteristics of the material are extracted by deepening the network depth, and the high-dimensional texture characteristics of the material are extracted by deepening the network depth, and the high- The drying process is optimized; the heat temperature of the intelligent drying device for the coater, the initial moisture content of the coater material, the sludge rate, and the air quality value in the dryer are taken as input control quantities, and the drum speed collected with high precision is taken as the output quantity. Through the system identification method, a discrete time transfer function model of the optimization control system of the drying device for the coater is established; secondly, the batches with excellent drying in the drying process are recorded, and the changes in the drum speed of the batch are plotted as a standard reference. Combined with the functional relationship between the standard curve and each input parameter, the variable limit value is obtained, thereby deducing the appropriate temperature, moisture content, and air quality required in the drying process. As the input of the trained model, the predicted drum speed is obtained, and the drying process is optimized to improve efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] Figure 1 This is a schematic diagram of the overall structure of an intelligent drying device for a coating machine proposed by the present invention;
[0040] Figure 2 This is a schematic cross-sectional structure diagram of an intelligent drying device for a coating machine proposed by the present invention;
[0041] Figure 3 for Figure 2 A magnified schematic diagram of the structure in the middle;
[0042] Figure 4 This is a schematic diagram of the electric push rod structure of an intelligent drying device for a coating machine proposed by the present invention;
[0043] Figure 5 Based Figure 2 A magnified schematic diagram of structure B in the middle;
[0044] Figure 6 It is the system flow chart.
[0045] In the figure: 1-machine body; 2-drum; 3-heating port; 4-protection box; 5-electric push rod; 6-hole; 7-silicone ring; 8-moving block; 9-support rod; 10-carrying plate; 11-dust brush; 12-chute; 13-storage box; 14-exhaust fan; 15-exhaust pipe; 16-air supply pipe; 17-iron frame; 18-filter; 19-opening; 21-handle; 22-exhaust pipe; 23-feeding port; 24-discharging port; 25-inspection door; 26-observation window; 27-purification box; 28-magnet; 29-groove. DETAILED DESCRIPTION
[0046] 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.
[0047] Example 1: Figure 1 - Figure 4 As shown, an intelligent drying device for a coating machine in the figure comprises a machine body 1, wherein rollers 2 are arranged in sequence inside the machine body 1, a plurality of groups of heating ports 3 are fixedly installed inside the machine body 1, and the rollers 2 are located below the heating ports 3, a protective box 4 is fixedly installed on the outer wall of the machine body 1, an electric push rod 5 is fixedly installed inside the protective box 4, a hole 6 is provided on the protective box 4, and the protective box 4 is connected to the inside of the machine body 1 through the hole 6, one end of the electric push rod 5 is passed through the inside of the machine body 1 through the hole 6, a slide groove 12 is provided inside the machine body 1, and the slide groove 12 is located below the roller 2, a moving block 8 is slidably connected inside the slide groove 12, and one end of the electric push rod 5 is fixedly connected to the moving block 8, a support rod 9 is fixedly connected to the moving block 8, and the support rod 9 A carrying plate 10 is fixedly installed on the top, and a dust removal brush 11 is arranged on the carrying plate 10, and the dust removal brush 11 is in contact with the roller 2. The coating is transmitted in the machine body 1 through the roller 2. During the transmission, the heat energy provided by the heating system is used to dry the transmitted coating through the heating port 3. In the process of transmitting the coating, the roller 2 will inevitably be contaminated with dust and impurities. When cleaning the dust and impurities adhering to the roller 2, firstly, the electric push rod 5 is started to push the moving block 8 to move in the slide groove 12. When the moving block 8 moves, the dust removal brush 11 on the support rod 9 is used to clean the rolling roller 2 of impurities to prevent the dirt on the roller 2 from affecting the drying of the coating, thereby ensuring that the coating can be dried evenly. Cleaning the roller 2 when there is no need to pause the drying device also improves the working efficiency of the intelligent drying device for the coating machine.
[0048] Among them, Figure 4 As shown, a silicone ring 7 is arranged in the hole 6, and one end of the electric push rod 5 passes through the inside of the silicone ring 7, and the silicone ring 7 is used to prevent the hole 6 from causing wear to the electric push rod 5.
[0049] Embodiment 2, as Figure 1 , Figure 2 and Figure 5 As shown, this embodiment further illustrates Example 1, a storage box 13 is installed on the top of the machine body 1, an exhaust fan 14 is installed inside the storage box 13, an exhaust pipe 15 is connected to the exhaust fan 14, and one end of the exhaust pipe 15 is passed through the inside of the machine body 1, an air supply pipe 16 is connected to the exhaust fan 14, a purification box 27 is fixedly installed on the top of the machine body 1, and one end of the air supply pipe 16 is connected to the purification box 27, an iron frame 17 is arranged inside the purification box 27, and a filter screen is detachably connected to the iron frame 17 18. The purification box 27 is connected with an exhaust pipe 22. When filtering harmful gases generated by high temperature inside the drying device, the exhaust fan 14 is first started to allow the exhaust pipe 15 to extract the gas in the machine body 1, and then the gas enters the purification box 27 through the air supply pipe 16, so that the filter 18 in the purification box 27 filters the harmful gases, and the filtered gases are discharged to the outside through the exhaust pipe 22, thereby avoiding harmful gases generated by the intelligent drying device for the coating machine during drying and coating to cause harm to the health of the staff, thereby ensuring the health of the staff.
[0050] Among them, Figure 5 As shown, a handle 21 is fixedly connected to one end of the iron frame 17, an opening 19 is provided on the purification box 27, and the iron frame 17 is slidably connected to the opening 19 through the handle 21. The iron frame 17 is pulled out of the purification box 27 by the handle 21, and the filter 18 is regularly cleaned and replaced.
[0051] Among them, Figure 5 As shown, a groove 29 is provided inside the purification box 27, a magnet 28 is provided in the groove 29, and the iron frame 17 is arranged in the groove 29. The magnet 28 can quickly place the iron frame 17 in the groove 29 through the attraction of the iron frame 17 when placing the iron frame 17, thereby limiting the position of the iron frame 17.
[0052] Among them, Figure 1As shown, the machine body 1 is provided with an inspection door 25, and the inspection door 25 is provided with an observation window 26. The two ends of the machine body 1 are symmetrically provided with a feed port 23 and a discharge port 24. When the drying device fails, the inspection door 25 is convenient for maintenance personnel to repair the inside of the drying device, and the observation window 26 is convenient for the staff to observe the drying condition of the coating inside the drying device so as to make adjustments through the controller in time. The coating that needs to be dried enters the drying device through the feed port 23 for drying, moves through the drum 2, and then flows out from the discharge port 24.
[0053] During use: First, the coating is transmitted in the machine body 1 through the roller 2. During the transmission, the heat energy provided by the heating system is used to dry the transmitted coating through the heating port 3. During the transmission of the coating, the roller 2 will inevitably be contaminated with dust and impurities. When cleaning the dust and impurities adhering to the roller 2, first start the electric push rod 5 to push the moving block 8 to move in the slide groove 12. When the moving block 8 moves, the dust removal brush 11 on the support rod 9 is used to clean the rolling roller 2 from impurities to prevent the dirt on the roller 2 from affecting the coating drying, thereby ensuring that the coating can be dried evenly. Cleaning the roller 2 when there is no need to pause the drying device also improves the working efficiency of the intelligent drying device for the coating machine.
[0054] Finally, when filtering the harmful gas generated by high temperature inside the drying device, first start the exhaust fan 14 to allow the exhaust pipe 15 to extract the gas in the machine body 1, and then enter the purification box 27 through the air supply pipe 16, so that the filter 18 in the purification box 27 filters the harmful gas, and the filtered gas is discharged to the outside through the exhaust pipe 22, thereby avoiding the harmful gas generated by the intelligent drying device of the coating machine during drying and coating to cause harm to the health of the staff, thereby ensuring the health of the staff.
[0055] Embodiment three, as Figure 6 As shown, an intelligent drying device for a coating machine and a drying method thereof include the following steps:
[0056] Step 1: Real-time acquisition of images and state parameters of the coating machine during the drying process, including temperature, air quality parameters and moisture content of the material during the drying stage;
[0057] Among them, a temperature sensor and an air quality detector are arranged inside the drying device to detect the heat temperature of the drying device and the air quality generated respectively; a weight sensor is arranged outside the drying device to collect the initial weight of the coating machine material and compare it with the standard drying material to calculate the moisture content of the material; and the sensor and the detector are installed adaptively;
[0058] By detecting the air quality, the impact of harmful gases on workers can be avoided, and the air can be purified intelligently. The temperature and moisture content of the material in the drying stage directly affect the drying process and parameter settings. By building the relationship between the above variables, the sludge model and drying optimization model can be constructed.
[0059] Step 2: Analyze the images of the drying stage, extract the color and texture features of the material, and build a sludge recognition model to automatically identify the sludge in the drying stage;
[0060] Among them, the specific process of constructing the sludge identification model is as follows:
[0061] A. Obtain the image of the drying stage within the preset time, perform fuzzy filtering, and then perform multi-channel color feature extraction and high-dimensional material texture feature extraction on the image of the drying stage. The specific process is as follows:
[0062] A1, divide the denoised image equally into multiple regions, where each region image includes height h, width w and channel number features; then define a 3*3 convolution layer, perform convolution operation on the three channels of the region image, and obtain the color feature value Sc, the formula is: Sc = ∑(R, G, B)*Wq i +ε i , where (R, G, B) are color channels, Wq i is the convolution kernel corresponding to each channel, i is the image pixel, ε i is the error coefficient corresponding to each channel. The setting of the error coefficient makes the final result closer to the true value;
[0063] A2, obtain (R, G, B) channels, deepen the network depth in each channel, and then extract the material texture features to obtain the material texture feature value Wl, the formula is: Where Wl is the material texture feature value, (R, G, B) is the color channel, ε j is the error coefficient corresponding to each channel, SE is an introduced layer structure, is the convolution kernel corresponding to each channel;
[0064] Among them, the specific process of deepening the network depth is as follows:
[0065] The output value of the first layer of convolution operation is incorporated into the SE structure, and the second layer is changed to a 5*5 convolution kernel, that is, a 1*1 dimension is introduced on the 3*3 to deepen the network depth; compared with the traditional 3*3 structure, the texture can be obtained more clearly. At the same time, the color features extracted under the 3*3 convolution layer are retained here to prevent color loss after deepening the network depth;
[0066] The input layer of the SE structure is the upper network, that is, the Xi feature of the network output layer after the 3*3 convolution operation. The Xi feature is expressed as c*h*w, where c is the channel. The Xi feature is then compressed through a global pooling layer and finally output as a 1*1*c result.
[0067] Among them, the specific operations in the global pooling layer are:
[0068] First, the size of the spatial dimension is reduced to 1*1*(c / r), where r is the compression rate, which is used to reduce the amount of computational complexity of the network. Then, the convolution is expanded through the defined activation function to increase the output size from c / r to c. The activation function is used to increase the nonlinearity of the network, thereby increasing the network's fitting ability.
[0069] A3, then combine the color feature value Sc with the material texture feature value Wl to generate the feature factor S j , the formula is: Where j represents the image, and j is a positive integer, α1 and α2 are the control coefficients of the color feature value and the material texture feature value respectively. The control coefficients are set to make the calculation results closer to the true value;
[0070] B. Real-time labeling of the drying degree of the image in the drying stage: matching the image in the drying stage with the preset drying degree image, and dividing the drying state according to the drying degree, the drying state is sequentially divided into initial stage, gradual drying stage, color fixing stage and color forming stage, wherein the initial stage indicates that the drying has not been done, the gradual drying stage indicates a low drying degree, the color fixing stage indicates a medium drying degree, and the color forming stage indicates a high drying degree; at the same time, the drying state is sequentially labeled with the drying degree value, which is divided into 1, 2, 3 and 4, wherein 1 corresponds to the initial stage, 2 corresponds to the gradual drying stage, 3 corresponds to the color fixing stage, and 4 corresponds to the color forming stage, and then the drying state is combined with the drying degree value to generate the drying information;
[0071] The degree of drying is divided into the determination of the degree of color difference between the image matches, and the formula for calculating the color difference is: Where: Where ΔE is the color difference, the sum of L0, a0 and b0 represents the chromaticity value of the initial drying, the sum of L, a and b represents the chromaticity value during the drying process, L is the brightness, and a and b represent the color values during the process;
[0072] C, combine the drying information with the characteristic factor and perform weighted summation to obtain the material factor Wn, the formula is: Where L j Indicates the label category, S j represents the characteristic factor, j is the image, and n represents the number of images under the same drying state;
[0073] D. Obtain the material factor Wn within the preset time, and compare the material factor Wn with the preset material factor WN to calculate whether the coating is uniform and whether there is sludge; set the gradient difference, and each image area corresponds to a gradient difference, then each area corresponds to a sludge rate, the formula is: Where Wn o is the material factor of the oth region, is the sludge factor of the entire area, k means that k areas are divided, where k = 1, 2, ..., p, and o < p; wherein the error interval of the preset gradient difference is ±wc%, when the gradient difference is not within the error interval, it means that the sludge rate exceeds the standard, and targeted optimization and improvement operations are performed;
[0074] The larger the sludge in an area, the larger the sludge rate, indicating uneven drying or dust impurities in the area block; and each drying state has a corresponding material factor, and each material factor has multiple areas. When the sludge rate exceeds the standard in a single area or the sum of the entire area, the dust removal brush is increased to clean the dust impurities so that the cleaning work is completed when the next drying state is obtained;
[0075] Step 3: Combine the sludge rate and state parameters to build a drying optimization model, realize the judgment and prediction of the drying state, and optimize the drying process;
[0076] The specific process of establishing the drying optimization model is as follows:
[0077] S1, obtaining drying information and drying state parameters within a preset time, building a training set, and initializing a neural network structure;
[0078] S2, training the neural network according to the training set, and determining the output value of the objective function of the trained neural network, where the output value of the objective function includes the output value of the hidden layer and the output value of the output layer; the specific process is:
[0079] Sa, takes air quality parameters, temperature, moisture and sludge rate as the network input layer, takes drum speed as the network output layer, and sets the nodes and weights between network layers;
[0080] According to the formula: Thus, the number of nodes can be determined, where I represents the number of hidden layer nodes, mi represents the number of output layer nodes, ni represents the number of input layer nodes, ai represents the adjustment constant, and ai∈[1-10];
[0081] Sb, sets the error coefficient E, and updates the weights and thresholds between each node in the network structure in real time;
[0082] Sb1, then the hidden layer to the output layer is:
[0083]
[0084] Where η1 and η2 are the learning rates of weight and threshold respectively, is the updated weight, is the weight at the previous time point; is the update threshold, is the threshold at the previous time point; is the partial derivative of the error function from the hidden layer to the output layer, is the partial derivative of the error function with respect to the threshold from the hidden layer to the output layer;
[0085] Sb2, then the input layer and the receiving layer to the hidden layer:
[0086]
[0087] Among them, η3 and η4 are the learning rates of the weights from the input layer to the hidden layer and from the receiving layer to the hidden layer, respectively, and η5 is the learning rate of the threshold; and are the update weights from the input layer to the hidden layer and from the receiving layer to the hidden layer, respectively. and are the weights from the input layer to the hidden layer at the previous time point and from the receiving layer to the hidden layer at the previous time point; is the update threshold, is the threshold at the previous time point; is the partial derivative of the error function from the input layer to the hidden layer, is the partial derivative of the error function from the receiving layer to the hidden layer, is the partial derivative of the error function with respect to the threshold of the input layer and the receiving layer to the hidden layer;
[0088] Sc, generates the output value of the objective function:
[0089] Sc1, then the output value of the hidden layer is:
[0090] H′ IC =H IC (t-1);
[0091] Among them, H IC is the hidden layer output value, Qz u is the weight between the input layer and the hidden layer, u is the node between the input layer and the hidden layer, and u=1, 2, …, r; Qz v is the weight between the receiving layer and the hidden layer, v is the node between the receiving layer and the hidden layer, and v = 1, 2, ..., rl; yb v is the threshold of the vth node, x uis the input parameter, t is the number of learning times, f(.) is the hidden layer activation function, The hidden layer output value of the vth node at the previous moment, H′ IC is the output value of the hidden layer at the previous moment, which is fed back by the receiving layer and used together with the input layer as the current output;
[0092] Sc2, the output value formula of the output layer is: Among them, H SC is the output value of the output layer, e is the node between the output layer and the hidden layer, and e=1, 2, …, rll; Qz e is the weight between the hidden layer and the output layer, yb e is the threshold of the e-th node, is the hidden layer output value of the e-th node, and g(·) is the activation function of the output layer;
[0093] S3, when the training process is stable, the predicted drum speed is obtained in real time, that is, the node corresponding to the output value of the output layer; and the curve change of the input parameter-drum speed during the training process is established, and then normalized to generate the variable limit value JM, the formula is: Where Tc is the temperature set by the drying device, ZL is the air quality value in the dryer, Sd is the material moisture, Wnl is the sludge rate, H SC is the output value of the output layer, that is, the drum speed; ψ1, ψ2, ψ3, ψ4 and ψ5 are the normalization coefficients of temperature, moisture, air quality value, sludge rate and output layer output value, respectively, and ψ1, ψ2, ψ3, ψ4 and ψ5 are all greater than 0;
[0094] When the temperature is higher, the variable limit value is larger; when the air quality is small, the variable limit value is larger; when the material moisture is small, the variable limit value is larger; when the sludge rate is smaller, the variable limit value is larger; when the predicted drum speed is smaller, the variable limit value is larger;
[0095] S4, obtaining the variable limit values JM of the preset time, and grouping them into a set J = {JM1, JM2, JM3, ..., JM D}, where D is a natural number greater than 1, and the maximum variable limit value in the set is obtained, and the time period where the maximum variable limit value is located is recorded as the optimal drying time period, and finally the optimization operation in the drying process is achieved.
[0096] In summary of the above technical solutions: the present invention determines the drying state by collecting the chromaticity of the drying process, and simultaneously extracts color features and high-dimensional texture features through multi-channels to construct a sludge recognition model, automatically judges the sludge in the process, and simultaneously drives the dust removal brush for cleaning; a drying optimization model is also constructed through a neural network to achieve the effect of intelligent drying; in this process, the variable limit value is obtained by recording the change curve, and the optimized environmental conditions can be inferred through analysis, thereby improving the drying efficiency and ensuring the safety of the drying process.
[0097] Those skilled in the art will appreciate that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by a combination of devices or computer software and electronic hardware; whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution.
[0098] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.
Claims
1. An intelligent drying method for a coating machine, characterized in that: The following steps are involved: Step 1: Real-time acquisition of images and state parameters of the coating machine during the drying process, including temperature, air quality parameters and moisture content of the material during the drying stage; Step 2: Analyze the images of the drying stage, extract the color and texture features of the material, and build a sludge recognition model to automatically identify the sludge in the drying stage; Among them, the specific process of constructing the sludge identification model is as follows: A. Obtain the image of the drying stage within the preset time, perform fuzzy filtering, and then perform multi-channel color feature extraction and high-dimensional material texture feature extraction on the image of the drying stage to generate feature factors ; Characteristic Factor The specific generation process is: A1, divide the denoised image equally into multiple regions, where each region image includes height h, width w and channel number features; then define a 3×3 convolution layer to perform convolution operations on the three channels of the region image to obtain the color feature value ; A2, obtain (R, G, B) channels, deepen the network depth in each channel, extract the material texture features, and obtain the material texture feature value ; The formula is: ,in is the material texture feature value, (R, G, B) is the color channel, is the error coefficient corresponding to each channel, SE is an introduced layer structure, is the convolution kernel corresponding to each channel; A3, and then the color feature value and material texture feature value Combine to generate characteristic factors ; The formula is: , where j represents the image and j is a positive integer, and are the control coefficients of color characteristic value and material texture characteristic value respectively; the formula is: , where (R, G, B) are color channels, is the convolution kernel corresponding to each channel, i is the image pixel, is the error coefficient corresponding to each channel. The setting of the error coefficient makes the final result closer to the true value; B. Real-time labeling of the drying degree of the image in the drying stage: matching the image in the drying stage with the preset drying degree image, and dividing the drying state according to the drying degree, the drying state is sequentially divided into initial stage, gradual drying stage, color fixing stage and color finishing stage, wherein the initial stage indicates that the drying has not been done, the gradual drying stage indicates a low drying degree, the color fixing stage indicates a medium drying degree, and the color finishing stage indicates a high drying degree; at the same time, the drying state is sequentially labeled with the drying degree value, which is divided into 1, 2, 3 and 4, wherein 1 corresponds to the initial stage, 2 corresponds to the gradual drying stage, 3 corresponds to the color fixing stage, and 4 corresponds to the color finishing stage, and then the drying state and the drying degree value are combined to generate the drying information; C. Combine the drying information with the characteristic factors and perform weighted summation to obtain the material factor ; The formula is: ,in Indicates the label category. represents the characteristic factor, j is the image, and n represents the number of images under the same drying state; D. Get the material factor within the preset time , and the material factor With preset material factors Compare and set the gradient difference, and each image area corresponds to a gradient difference, and each area corresponds to a sludge rate; the error range of the preset gradient difference is ±wc%. When the gradient difference is not within the error range, it means that the sludge rate exceeds the standard, and targeted optimization and improvement operations are performed; Step 3: Combine the sludge rate and state parameters to build a drying optimization model, realize the judgment and prediction of the drying state, and optimize the drying process; Among them, the specific process of establishing the drying optimization model is as follows: S1, obtaining drying information and drying state parameters within a preset time, building a training set, and initializing a neural network structure; S2, training the neural network according to the training set, and determining and analyzing the output value of the objective function of the trained neural network, wherein the output value of the objective function includes the output value of the hidden layer and the output value of the output layer; S3, when the training process is stable, the predicted drum speed is obtained in real time, that is, the node corresponding to the output value of the output layer; and the curve change of the input parameter-drum speed during the training process is established, and then normalized to generate the variable limit value JM; S4, obtain the variable limit value JM of the preset time and collect it into a set , where D is a natural number greater than 1, and the maximum variable limit value in the set is obtained, and the time period where the maximum variable limit value is located is recorded as the optimal drying time period, and finally the optimization operation in the drying process is achieved.
2. The intelligent drying method for a coating machine according to claim 1, characterized in that: The process of generating the output value of the objective function is as follows: Sa, takes air quality parameters, temperature, moisture and sludge rate as the network input layer, takes drum speed as the network output layer, and sets the nodes and weights between network layers; Sb, sets the error coefficient E, and updates the weights and thresholds between each node in the network structure in real time; Sc, generates the output value of the objective function.
3. An intelligent drying device for a coating machine, the device using an intelligent drying method for a coating machine as described in any one of claims 1-2, characterized in that: The invention comprises a machine body (1), wherein rollers (2) are arranged in sequence inside the machine body (1), and a plurality of groups of heating ports (3) are fixedly installed inside the machine body (1), and the rollers (2) are located below the heating ports (3). The invention is characterized in that a protective box (4) is fixedly installed on the outer wall of the machine body (1), an electric push rod (5) is fixedly installed inside the protective box (4), a hole (6) is provided on the protective box (4), and the protective box (4) is connected to the inside of the machine body (1) through the hole (6), and one end of the electric push rod (5) is connected to the inside of the machine body (1) through the hole (6). The hole (6) is formed inside the machine body (1), and a slide groove (12) is provided inside the machine body (1), and the slide groove (12) is located below the roller (2). A moving block (8) is slidably connected in the slide groove (12), and one end of the electric push rod (5) is fixedly connected to the moving block (8). A support rod (9) is fixedly connected to the moving block (8), and a bearing plate (10) is fixedly installed on the top of the support rod (9). A dust removal brush (11) is provided on the bearing plate (10), and the dust removal brush (11) is in contact with the roller (2).
4. The intelligent drying device using the intelligent drying method for a coating machine according to claim 3, characterized in that: A silicone ring (7) is arranged in the hole (6), and one end of the electric push rod (5) passes through the inside of the silicone ring (7); a storage box (13) is installed on the top of the machine body (1), and an exhaust fan (14) is installed inside the storage box (13); an exhaust pipe (15) is connected to the exhaust fan (14), and one end of the exhaust pipe (15) is passed through the inside of the machine body (1); an air supply pipe (16) is connected to the exhaust fan (14); a purification box (27) is fixedly installed on the top of the machine body (1), and one end of the air supply pipe (16) is connected to the purification box (27); an iron frame (17) is arranged inside the purification box (27), and a filter screen (18) is detachably connected to the iron frame (17); and an exhaust pipe (22) is connected to the purification box (27).
5. The intelligent drying device using the intelligent drying method for a coating machine according to claim 4, characterized in that: A handle (21) is fixedly connected to one end of the iron frame (17), an opening (19) is provided on the purification box (27), and the iron frame (17) is slidably connected to the opening (19) via the handle (21).
6. The intelligent drying device using the intelligent drying method for a coating machine according to claim 4, characterized in that: A groove (29) is provided inside the purification box (27), a magnet (28) is provided in the groove (29), and the iron frame (17) is arranged in the groove (29).
7. The intelligent drying device using the intelligent drying method for a coating machine according to claim 3, characterized in that: A feed inlet (23) and a discharge outlet (24) are symmetrically arranged at both ends of the machine body (1).
