Intelligent control system and control method for monitoring needle felt production process

Abstract

The present invention relates to the field of intelligent manufacturing technology, and to an intelligent control system and a control method for monitoring the production process of needle felt. The control method includes: collecting the opening degree monitoring image and the mixing uniformity monitoring data in the production process of needle felt; preprocessing the opening degree monitoring image and the mixing uniformity monitoring data; inputting the preprocessed opening degree monitoring image into a pre-built opening degree adjustment parameter prediction model, and outputting the opening degree adjustment parameter and the opening abnormality mark; inputting the pre-processed mixing uniformity monitoring data into a pre-built mixing uniformity adjustment parameter prediction model, and outputting the mixing uniformity adjustment parameter. The present invention improves the automation level, realizes the real-time monitoring and intelligent regulation of the opening degree and mixing uniformity, thereby replacing the traditional manual adjustment, reducing the manpower input, and improving the stability and consistency of product quality.

Description

Technical Field

[0001] The present invention relates to the field of intelligent manufacturing technology, and more specifically, to an intelligent control system and a control method thereof for monitoring a needle felt production process. Background Art

[0002] Needle felt is a common filling material, widely used in furniture, mattress and other industries. With the popularization of automation technology today, automated production has become the main direction of needle felt production.

[0003] For example, the Chinese patent with publication number CN113391608A provides an intelligent production control system for glass wool felt, which includes a controller, the controller is connected with a series module and a parallel module, the series module and the parallel module respectively connect the bag opener, the loosening machine, the vibrating cotton feeder, the carding machine, the web laying machine, the reeling machine, the needle punching machine, and the trimming and rolling machine in series and in parallel; the cargo output ports of the bag opener, the loosening machine, the vibrating cotton feeder, the carding machine, the web laying machine, the reeling machine, the needle punching machine, and the trimming and rolling machine are respectively provided with a buffer area; the parameter acquisition sensor is used to collect the internal and external environmental parameters of each device in the system, and adjust the operating status of each device in the system according to the feedback data output of the parameter acquisition sensor. The various devices of the glass wool felt production line are managed in different modes to ensure the smooth and continuous operation of the production line, and improve the working efficiency and emergency response capabilities of the glass wool felt production line.

[0004] In the production process of needle felt, opening degree and mixing uniformity are two key indicators that affect the quality. In traditional needle felt production lines, important process parameters such as opening degree and mixing uniformity are mainly controlled by manual testing and adjustment. This manual testing and adjustment method is complicated to operate, inefficient, and easily affected by subjective factors, making it difficult to ensure product consistency.

[0005] With the popularization of modern automation technology, intelligent manufacturing technology will also emerge in light industry. How to use intelligent control in the production process to control automated machines for automated production of needle felt so that the output of products can be increased while ensuring quality stability and consistency has become a difficult problem that needs to be solved by current technology. Summary of the invention

[0006] In order to solve the above problems, an embodiment of the present invention provides an intelligent control method for monitoring a needle felt production process, the method comprising:

[0007] Collect the monitoring images of the degree of opening and the monitoring data of the mixing uniformity during the production process of needle felt;

[0008] Preprocessing the opening degree monitoring images and mixing uniformity monitoring data;

[0009] Input the preprocessed opening degree monitoring image into the pre-built opening degree adjustment parameter prediction model, and output the opening degree adjustment parameter and opening abnormality mark;

[0010] The preprocessed mixing uniformity monitoring data is input into a pre-built mixing uniformity adjustment parameter prediction model, and the mixing uniformity adjustment parameter is output;

[0011] Receive the opening degree adjustment parameters and the opening abnormality mark, control the opening equipment to open according to the opening degree adjustment parameters and the opening abnormality mark, obtain the raw material batching, and control the transportation line to start and transport the raw material batching to the mixing equipment;

[0012] Receive the mixing uniformity adjustment parameter, control the mixing equipment to mix the raw materials according to the mixing uniformity adjustment parameter to obtain uniform materials, and control the transportation line to start conveying the uniform materials to the combing machine.

[0013] Furthermore, the method for constructing the opening degree adjustment parameter prediction model includes:

[0014] The looseness adjustment parameter prediction model adopts the VGG model structure, which includes an input layer, D convolution blocks, 2 fully connected layers and an output layer. The image size of the input layer is uniformly planned to be 224×224×U, D and U are preset values, and the convolution blocks include the first convolution block, the second convolution block, the third convolution block, the fourth convolution block and the fifth convolution block; wherein, the convolution blocks all contain two 3×3 convolution layers, and the convolution kernel size in the convolution layer is 3×3. The pooling structure of the convolution block includes a pooling layer and a convergence layer. The pooling structures of the first convolution block and the second convolution block are both pooling layers, and the pooling structures of the third convolution block, the fourth convolution block and the fifth convolution block are all convergence layers. In the pooling structure of the convolution block, the pooling window size is 2×2, with a stride of 2; the convolution layer of the first convolution block has 64 output channels, the convolution layer of the second convolution block has 128 output channels, the convolution layer of the third convolution block has 256 output channels, the convolution layer of the fourth convolution block has 512 output channels, and the convolution layer of the fifth convolution block has 512 output channels; the fully connected layer includes the first fully connected layer and the second fully connected layer; the number of input nodes of the first fully connected layer is 25088, the number of output nodes is 4096, and the activation function is the ReLU function, the number of input nodes of the second fully connected layer is 4096, the number of output nodes is 4096, the activation function is the ReLU function, the number of input nodes of the output layer is 4096, the number of output nodes is 3, and there is no activation function.

[0015] Furthermore, the training method of the opening degree adjustment parameter prediction model includes:

[0016] P opening adjustment parameter prediction logs are extracted from the database. Each opening adjustment parameter prediction log includes one opening monitoring image and a set of corresponding opening adjustment parameters and opening abnormality marks. The p opening adjustment parameter prediction logs are divided into a training set and a validation set. The training set is H opening adjustment parameter prediction logs, and the validation set is z opening adjustment parameter prediction logs. The optimizer is the Adam optimizer. The training set is imported into the opening adjustment parameter prediction model to complete the forward calculation from the input layer to the output layer. The cross entropy is used as the loss function. The loss function loos formula is:

[0017]

[0018] In the formula, E i is the actual opening adjustment parameter and opening abnormality mark corresponding to the i-th opening monitoring image, Ea(Fh i ) is the predicted opening adjustment parameter and opening abnormality mark of the i-th opening monitoring image, η is a hyperparameter, is the regularization of the weights of the first fully connected layer and the second fully connected layer; i is the number of the looseness monitoring image;

[0019] According to the optimizer, the parameter values ​​of the parameter prediction model are adjusted according to the degree of relaxation updated. The learning rate is reduced by half after each round of training. The loss function on the validation set is calculated to observe whether the loss function converges. If it converges, the training is terminated.

[0020] Furthermore, the method for constructing the mixing uniformity adjustment parameter prediction model includes:

[0021] Initialize the mixed uniformity adjustment parameter prediction model structure. The mixed uniformity adjustment parameter prediction model structure adopts an MLP type multi-layer forward network structure. The mixed uniformity adjustment parameters include mixing speed, mixing time and mixing sequence. The mixing sequence includes fiber type, mixing order and mixing ratio. The MLP type multi-layer forward network structure includes an input layer, a hidden layer and an output layer. The input layer includes a first input layer, a second output layer and a third output layer, which correspond to color analysis data, component analysis data and quality data respectively. The hidden layer includes a first hidden layer and a second hidden layer. The number of nodes in the first hidden layer is 128R, with Re as the input layer. lu function is used as the activation function; the number of nodes in the second hidden layer of the second hidden layer is 64R, R is a preset value, and the Relu function is used as the activation function; the output layer includes the first output layer, the second output layer and the third output layer, the number of nodes in the first output layer of the first output layer is 1, and the applicable mixing speed is predicted; the number of nodes in the second output layer of the second output layer is 1, and the applicable mixing time is predicted; the third output layer nodes of the third output layer include node 1, node 2 and node 3, the number of nodes is [3, m+2], node 1 is the predicted fiber type, m is the number of mixed fiber types, node 2 is the predicted mixing order, and node 3 is the predicted mixing ratio;

[0022] After initializing the mixed uniformity adjustment parameter prediction model structure, the mixed uniformity adjustment parameter prediction model is trained using the color analysis data, component analysis data, quality data, and the mixed uniformity adjustment parameters corresponding to the color analysis data, component analysis data, and quality data. The optimizer is Adam, and the loss function is the MSE loss function. The training ends when the loss function converges.

[0023] Furthermore, the training method of the mixing uniformity adjustment parameter prediction model includes:

[0024] Converting the color analysis data, the component analysis data and the quality data into a corresponding set of first feature vectors;

[0025] The color analysis data, component analysis data and quality data are used as inputs of a mixing uniformity adjustment parameter prediction model. The mixing uniformity adjustment parameter prediction model takes a group of mixing uniformity adjustment parameters corresponding to each group of color analysis data, component analysis data and quality data as output, takes the mixing uniformity adjustment parameters actually corresponding to each group of color analysis data, component analysis data and quality data as prediction targets, and takes minimizing the loss function value of the mixing uniformity adjustment parameter prediction model as training target; and stops training when the loss function value of the mixing uniformity adjustment parameter prediction model is less than or equal to a preset target loss value.

[0026] Furthermore, the intelligent control method for monitoring the needle felt production process also includes:

[0027] Collect temperature and humidity;

[0028] Input temperature, humidity, opening degree adjustment parameters and mixing uniformity adjustment parameters into the pre-built final adjustment parameter model to output the final adjustment parameters.

[0029] Furthermore, the method for constructing the final adjustment parameter model includes:

[0030] Taking temperature, humidity, opening degree adjustment parameters and mixing uniformity adjustment parameters as analysis data, collecting final adjustment parameters corresponding to c groups of analysis data in advance, and converting the analysis data and the corresponding final adjustment parameters into a corresponding set of second eigenvectors;

[0031] Each group of second eigenvectors is used as the input of the final adjustment parameter model. The final adjustment parameter model takes a group of final adjustment parameters corresponding to each group of analysis data as output, and takes the actual final adjustment parameters corresponding to each group of analysis data as the prediction target. The actual final adjustment parameters are the final adjustment parameters corresponding to the analysis data collected in advance. The training target is to minimize the sum of the prediction errors of all analysis data. The calculation formula of the prediction error is Z K =(a K -μ K ) 2 , where Z K is the prediction error, K = 1, 2, ..., c, K is the group number of the second eigenvector corresponding to the analysis data, α K is the final adjustment parameter corresponding to the Kth group of analysis data, μ K is the actual final adjustment parameter corresponding to the Kth group of analysis data; the final adjustment parameter model is trained until the sum of the prediction errors reaches convergence and the training is stopped;

[0032] The final adjustment parameter model is a deep neural network model.

[0033] Intelligent control system for monitoring the needle felt production process, the system includes:

[0034] Collection module: used to collect temperature, humidity, looseness monitoring images and mixing uniformity monitoring data during the production process of needle felt;

[0035] Preprocessing module: used to preprocess the opening degree monitoring image and mixing uniformity monitoring data;

[0036] Opening adjustment module: used to input the preprocessed opening degree monitoring image into the pre-built opening degree adjustment parameter prediction model, and output the opening degree adjustment parameter and opening abnormality mark;

[0037] Mixing adjustment module: used to input the pre-processed mixing uniformity monitoring data into the pre-built mixing uniformity adjustment parameter prediction model, and output the mixing uniformity adjustment parameters;

[0038] Final adjustment module: used to input temperature, humidity, opening degree adjustment parameters and mixing uniformity adjustment parameters into the pre-built final adjustment parameter model, and output the final adjustment parameters;

[0039] Final control module: The air conditioner controls the temperature and humidity according to the final adjustment parameters; the opening equipment and mixing equipment are adjusted according to the final adjustment parameters.

[0040] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for dynamic evaluation of project performance based on adaptive multi-objective optimization is implemented.

[0041] A computer-readable storage medium stores a computer program, and when the computer program is executed, the method for dynamically evaluating project performance based on adaptive multi-objective optimization is implemented.

[0042] The technical effects and advantages of the intelligent control method and device for monitoring the needle felt production process provided by the present invention are as follows:

[0043] The present invention improves the automation level, realizes real-time monitoring and intelligent control of opening degree and mixing uniformity, thereby replacing traditional manual adjustment and reducing manpower input; collects multi-source and multi-modal data, evaluates opening degree and uniformity from multiple angles such as image, spectrum, quality, etc., and the information is richer and more complete; establishes a prediction model for opening degree, uniformity and final adjustment parameters, realizes end-to-end intelligent prediction and decision-making, and directly outputs the adjustment plan; considering environmental influencing factors such as temperature and humidity, the model can adaptively perform fine adjustment under different conditions, and the result is more reliable; preprocessing and feature extraction can effectively extract key information and improve the efficiency of model training and prediction; model training collects a large amount of real data, optimizes the structure, and the prediction result is more in line with actual production needs; the overall system is modularly designed, collection-preprocessing-prediction-output, and the workflow is clear and efficient; real-time monitoring and learning can continuously improve the model, so that the control level and product changes are synchronously improved; full-process quality monitoring and tracking are carried out, which is conducive to improving product quality stability and consistency. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a flow chart of the intelligent control method for monitoring the needle felt production process in Example 1;

[0045] Figure 2 is a flow chart of the intelligent control method for monitoring the needle felt production process in Example 2;

[0046] Figure 3 A schematic diagram of the connection of an intelligent control system for monitoring the needle felt production process;

[0047] Figure 4 is a schematic diagram of an electronic device of the present invention;

[0048] Figure 5 Schematic diagram of the storage medium of the present invention. DETAILED DESCRIPTION

[0049] 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.

[0050] Embodiment 1:

[0051] See also Figure 1 As shown, the intelligent control method for monitoring the needle felt production process described in this embodiment includes:

[0052] Collect the monitoring images of the degree of opening and the monitoring data of the mixing uniformity during the production process of needle felt;

[0053] Preprocessing the opening degree monitoring images and mixing uniformity monitoring data;

[0054] Input the preprocessed opening degree monitoring image into the pre-built opening degree adjustment parameter prediction model, and output the opening degree adjustment parameter and opening abnormality mark;

[0055] The preprocessed mixing uniformity monitoring data is input into a pre-built mixing uniformity adjustment parameter prediction model, and the mixing uniformity adjustment parameter is output;

[0056] Receive the opening degree adjustment parameters and the opening abnormality mark, control the opening equipment to open according to the opening degree adjustment parameters and the opening abnormality mark, obtain the raw material batching, and control the transportation line to start and transport the raw material batching to the mixing equipment;

[0057] Receive the mixing uniformity adjustment parameter, control the mixing equipment to mix the raw materials according to the mixing uniformity adjustment parameter to obtain uniform materials, and control the transportation line to start conveying the uniform materials to the combing machine.

[0058] The looseness monitoring image is captured by a high-resolution camera or optical sensor. During the needle felt production process, the fibers need to be fully loosened to avoid clumping. The mixing uniformity monitoring data includes color analysis data, component analysis data and quality data. The color analysis data can analyze the slight differences in the colors of different types of fibers and understand the uniformity of the mixing, using a color sensor for detection. The component analysis data uses infrared spectroscopy (such as near-infrared spectroscopy) to detect the fiber component ratio, and the fiber component ratio can understand whether the mixing is uniform. The quality data includes fiber weight and density. The weight and density of the mixed fiber samples are analyzed to obtain a relatively accurate mixing ratio as one of the reference values ​​for mixing uniformity.

[0059] The opening degree adjustment parameters include the speed of the opening device and the feed amount of the opening device. Adjusting the speed of the opening device is to increase or decrease the speed of the opening device from the current setting value, thereby increasing or decreasing the opening degree. Adjusting the feed amount of the opening device is to increase or decrease the feed amount to improve the opening degree of the fiber. The opening abnormality marking is to detect and mark the opening abnormality areas that may exist in the image. If fiber agglomeration is found in certain areas, the model will mark these areas in the output image.

[0060] The mixing uniformity adjustment parameters include mixing speed, mixing time and mixing sequence. The mixing sequence includes fiber type, mixing order and mixing ratio. The mixing equipment can adjust the parameters according to the mixing uniformity to make the fibers mix evenly to ensure the consistency of the product. The mixing speed is to increase or decrease the speed of the mixing equipment. The mixing time is to extend or shorten the mixing time. The mixing sequence is the order and ratio of addition, that is, adjusting the order and ratio of addition of different fiber types, for example, first adding 20% ​​polyester and then adding 30% polypropylene.

[0061] The pretreatment method comprises:

[0062] Normalize the color analysis data and extract the color distribution characteristics, such as the average color value and color distribution; reduce the noise of the spectral data and extract the absorption peak characteristics, such as the main absorption peak position and intensity, to obtain the component analysis data; pre-process the looseness monitoring image, including grayscale, denoising and standardization.

[0063] The method for constructing the opening degree adjustment parameter prediction model comprises:

[0064] The looseness adjustment parameter prediction model adopts the VGG model structure, which includes an input layer, D convolution blocks, 2 fully connected layers and an output layer. The image size of the input layer is uniformly planned to be 224×224×U. D and U are preset values, which are determined by technical personnel according to the amount of data required for training. Generally speaking, D defaults to 5 and U defaults to 3; the convolution blocks include the first convolution block, the second convolution block, the third convolution block, the fourth convolution block and the fifth convolution block: wherein, the convolution blocks all contain two 3×3 convolution layers, the convolution kernel size in the convolution layer is 3×3, the pooling structure of the convolution block includes a pooling layer and a convergence layer, the pooling structure of the first convolution block and the second convolution block are both pooling layers, and the pooling structures of the third convolution block, the fourth convolution block and the fifth convolution block are all convergence layers In the pooling structure of the convolution block, the pooling window size is 2×2 and the step size is 2; the convolution layer of the first convolution block outputs 64 channels, the convolution layer of the second convolution block outputs 128 channels, the convolution layer of the third convolution block outputs 256 channels, the convolution layer of the fourth convolution block outputs 512 channels, and the convolution layer of the fifth convolution block outputs 512 channels; the fully connected layer includes the first fully connected layer and the second fully connected layer; the number of input nodes of the first fully connected layer is 25088, the number of output nodes is 4096, and the activation function is the ReLU function; the number of input nodes of the second fully connected layer is 4096, the number of output nodes is 4096, and the activation function is the ReLU function; the number of input nodes of the output layer is 4096, the number of output nodes is 3, and there is no activation function.

[0065] The training method of the opening degree adjustment parameter prediction model comprises:

[0066] P opening adjustment parameter prediction logs are extracted from the database. Each opening adjustment parameter prediction log includes one opening monitoring image and a set of corresponding opening adjustment parameters and opening abnormality marks. The p opening adjustment parameter prediction logs are divided into a training set and a validation set. The training set is H opening adjustment parameter prediction logs, and the validation set is z opening adjustment parameter prediction logs. The optimizer is the Adam optimizer. The training set is imported into the opening adjustment parameter prediction model to complete the forward calculation from the input layer to the output layer. The cross entropy is used as the loss function. The loss function loos formula is:

[0067]

[0068] In the formula, E i is the actual opening adjustment parameter and opening abnormality mark corresponding to the i-th opening monitoring image, Ea(Fh i ) is the predicted opening adjustment parameter and opening abnormality mark of the i-th opening monitoring image, η is a hyperparameter, is the regularization of the weights of the first fully connected layer and the second fully connected layer; i is the number of the looseness monitoring image;

[0069] According to the optimizer, the parameter values ​​of the parameter prediction model are adjusted according to the degree of relaxation updated. The learning rate is reduced by half after each round of training. The learning rate is determined by the current amount of training data and is 0.001 by default. It can be adjusted. The loss function on the validation set is calculated to observe whether the loss function converges. If it converges, the training ends.

[0070] The method for constructing the mixed uniformity modulation prediction model includes:

[0071] Initialize the mixed uniformity adjustment parameter prediction model structure. The mixed uniformity adjustment parameter prediction model structure adopts an MLP type multi-layer forward network structure. The mixed uniformity adjustment parameters include mixing speed, mixing time and mixing sequence. The mixing sequence includes fiber type, mixing order and mixing ratio. The MLP type multi-layer forward network structure includes an input layer, a hidden layer and an output layer. The input layer includes a first input layer, a second output layer and a third output layer, which correspond to color analysis data, component analysis data and quality data respectively. The hidden layer includes a first hidden layer and a second hidden layer. The number of nodes in the first hidden layer is 128R, with Re as the input layer. lu function is used as the activation function; the number of nodes in the second hidden layer of the second hidden layer is 64R, R is a preset value, and the Relu function is used as the activation function; the output layer includes the first output layer, the second output layer and the third output layer, the number of nodes in the first output layer of the first output layer is 1, and the applicable mixing speed is predicted; the number of nodes in the second output layer of the second output layer is 1, and the applicable mixing time is predicted; the third output layer nodes of the third output layer include node 1, node 2 and node 3, the number of nodes is [3, m+2], node 1 is the predicted fiber type, m is the number of mixed fiber types, node 2 is the predicted mixing order, and node 3 is the predicted mixing ratio;

[0072] After initializing the mixed uniformity adjustment parameter prediction model structure, the mixed uniformity adjustment parameter prediction model is trained using the color analysis data, component analysis data, quality data, and the mixed uniformity adjustment parameters corresponding to the color analysis data, component analysis data, and quality data. The optimizer is Adam, and the loss function is the MSE loss function. The training ends when the loss function converges.

[0073] The training method of the mixing uniformity adjustment parameter prediction model includes:

[0074] The color analysis data, the component analysis data and the quality data are converted into a corresponding set of first feature vectors.

[0075] The color analysis data, component analysis data and quality data are used as inputs of a mixing uniformity adjustment parameter prediction model. The mixing uniformity adjustment parameter prediction model takes a group of mixing uniformity adjustment parameters corresponding to each group of color analysis data, component analysis data and quality data as output, takes the mixing uniformity adjustment parameters actually corresponding to each group of color analysis data, component analysis data and quality data as prediction targets, and takes minimizing the loss function value of the mixing uniformity adjustment parameter prediction model as training target; and stops training when the loss function value of the mixing uniformity adjustment parameter prediction model is less than or equal to a preset target loss value.

[0076] The loss function value of the mixed uniformity adjustment parameter prediction model is the mean square error.

[0077] Mean square error is one of the commonly used loss functions. By transforming the loss function formula The model is trained with minimization as the goal, so that the mixed uniformity adjustment parameter prediction model can better fit the data, thereby improving the performance and accuracy of the model.

[0078] In the loss function, MSE is the loss function value of the mixed uniformity adjustment parameter prediction model, x is the first eigenvector group number; m is the number of eigenvector groups; y is the x is a set of mixing uniformity adjustment parameters output by a mixing uniformity adjustment parameter prediction model corresponding to the first eigenvector of the xth group, It is the mixing uniformity adjustment parameter actually corresponding to the first eigenvector of the xth group.

[0079] Example 2

[0080] like Figure 2 As shown, this embodiment further improves the design on the basis of embodiment 1. The difference is that the intelligent control method for monitoring the production process of needle felt provided in embodiment 1 ignores the influence of temperature and humidity on opening and mixing in actual operation. In the production process of needle felt, high temperature needs to reduce the opening speed, increase the feed amount; increase the mixing speed and shorten the time; low temperature can increase the opening speed, reduce the feed amount; reduce the mixing speed and extend the time; and under high humidity conditions, corresponding temperature compensation measures need to be taken; based on this, the intelligent control method for monitoring the production process of needle felt provided in this embodiment also includes:

[0081] Collect temperature and humidity;

[0082] Input temperature, humidity, opening degree adjustment parameters and mixing uniformity adjustment parameters into the pre-built final adjustment parameter model to output the final adjustment parameters.

[0083] The temperature and humidity are obtained by the temperature sensor and humidity sensor on the air conditioner to obtain specific data. Temperature and humidity have an impact on opening. High temperature will increase the tension of the fiber and reduce its ease of opening. At the same time, the mechanical properties will decrease under high temperature, and the opening effect will be reduced. Low temperature will relax the fiber and make it easy to open, but too low temperature may cause mechanical failure. High humidity will increase the adhesion between fibers and weaken the opening effect. Temperature and humidity have an impact on mixing. High temperature will increase the slippage between fibers, which is conducive to uniform mixing, but may change the fiber structure and cause color difference. At low temperature, the fiber has poor relaxation and it is difficult to form a uniformly distributed structure. High humidity will increase fiber adhesion and hinder uniform mixing and distribution.

[0084] The final adjustment parameters include a final opening degree adjustment parameter, a final mixing uniformity adjustment parameter, a temperature adjustment parameter and a humidity adjustment parameter; the final opening degree adjustment parameter is a value obtained by the opening degree adjustment parameter through the final adjustment parameter model, the final mixing uniformity adjustment parameter is a value obtained by the mixing uniformity adjustment parameter through the final adjustment parameter model, the temperature adjustment parameter is the air conditioner adding or subtracting a certain temperature value, and the humidity adjustment parameter is the air conditioner adding or subtracting a certain humidity value.

[0085] The method for constructing the final adjustment parameter model includes:

[0086] The temperature, humidity, opening degree adjustment parameters and mixing uniformity adjustment parameters are used as analysis data, the final adjustment parameters corresponding to the c groups of analysis data are collected in advance, and the analysis data and the corresponding final adjustment parameters are converted into a corresponding set of second eigenvectors.

[0087] Each group of second eigenvectors is used as the input of the final adjustment parameter model. The final adjustment parameter model takes a group of final adjustment parameters corresponding to each group of analysis data as output, and takes the actual final adjustment parameters corresponding to each group of analysis data as the prediction target. The actual final adjustment parameters are the final adjustment parameters corresponding to the analysis data collected in advance. The training target is to minimize the sum of the prediction errors of all analysis data. The calculation formula of the prediction error is Z K =(a K -μ K ) 2 , where Z K is the prediction error, K = 1, 2, ..., c, K is the group number of the second eigenvector corresponding to the analysis data, α K is the final adjustment parameter corresponding to the Kth group of analysis data, μ K is the actual final adjustment parameter corresponding to the Kth group of analysis data; the final adjustment parameter model is trained until the sum of the prediction errors reaches convergence and the training is stopped.

[0088] The final adjustment parameter model is a deep neural network model.

[0089] Example 3

[0090] like Figure 3 As shown, the intelligent control system for monitoring the needle felt production process includes:

[0091] Collection module: used to collect temperature, humidity, looseness monitoring images and mixing uniformity monitoring data during the production process of needle felt;

[0092] Preprocessing module: used to preprocess the opening degree monitoring image and mixing uniformity monitoring data;

[0093] Opening adjustment module: used to input the preprocessed opening degree monitoring image into the pre-built opening degree adjustment parameter prediction model, and output the opening degree adjustment parameter and opening abnormality mark;

[0094] Mixing adjustment module: used to input the pre-processed mixing uniformity monitoring data into the pre-built mixing uniformity adjustment parameter prediction model, and output the mixing uniformity adjustment parameters;

[0095] Final adjustment module: used to input temperature, humidity, opening degree adjustment parameters and mixing uniformity adjustment parameters into the pre-built final adjustment parameter model, and output the final adjustment parameters;

[0096] Temperature control module: used to receive the temperature adjustment parameter and humidity adjustment parameter in the final adjustment parameter, and control the air conditioner to perform temperature and humidity control according to the temperature adjustment parameter and humidity adjustment parameter;

[0097] Automatic opening module: used for receiving the opening abnormality mark and the final opening degree adjustment parameter in the final adjustment parameter, controlling the opening equipment to open according to the final opening degree adjustment parameter and the opening abnormality mark, obtaining the raw material batching, and controlling the start of the transport line to transport the raw material batching to the mixing equipment;

[0098] Mixing module: used to receive the final mixing uniformity adjustment parameter in the final adjustment parameter, control the mixing equipment to mix the raw materials according to the final mixing uniformity adjustment parameter to obtain uniform materials, and control the start of the transport line to transport the uniform materials to the combing machine.

[0099] Example 4

[0100] See also Figure 4 As shown, according to another aspect of the present application, an electronic device 500 is also provided. The electronic device 500 may include one or more processors and one or more memories. The memories store computer readable codes, and when the computer readable codes are executed by one or more processors, the intelligent control method for monitoring the needle felt production process as described above may be executed.

[0101] The system according to the embodiment of the present application can also be used Figure 4 The electronic device architecture shown in FIG. Figure 4 As shown, the electronic device 500 may include a bus 501, one or more CPUs 502, a ROM 503, a RAM 504, a communication port 505 connected to a network, an input / output 506, a hard disk 507, etc. The storage device in the electronic device 500, such as the ROM 503 or the hard disk 507, may store the intelligent control method for monitoring the needle felt production process provided by the present application. Furthermore, the electronic device 500 may also include a user interface 508. Of course, Figure 4 The architecture shown is only exemplary and can be omitted according to actual needs when implementing different devices. Figure 4 One or more components of an electronic device are shown.

[0102] Example 5

[0103] See also Figure 5 As shown, it is a computer-readable storage medium according to an embodiment of the present application. Computer-readable instructions are stored on the computer-readable storage medium. When the computer-readable instructions are executed by the processor, the intelligent control method for monitoring the needle felt production process according to the embodiment of the present application described with reference to the above figures can be executed. The storage medium includes, but is not limited to, for example, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory (cache), etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0104] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0105] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent control method for monitoring the needle felt production process, characterized in that: Methods include: Collect the openness monitoring images and mixing uniformity monitoring data during the needle felt production process; the mixing uniformity monitoring data includes color analysis data, component analysis data and quality data; Preprocessing the opening degree monitoring images and mixing uniformity monitoring data; Input the preprocessed opening degree monitoring image into the pre-built opening degree adjustment parameter prediction model, and output the opening degree adjustment parameter and opening abnormality mark; The preprocessed mixing uniformity monitoring data is input into a pre-built mixing uniformity adjustment parameter prediction model, and the mixing uniformity adjustment parameter is output; The method for constructing the mixing uniformity adjustment parameter prediction model includes: Initialize the mixed uniformity adjustment parameter prediction model structure. The mixed uniformity adjustment parameter prediction model structure adopts an MLP type multi-layer forward network structure. The mixed uniformity adjustment parameters include mixing speed, mixing time and mixing sequence. The mixing sequence includes fiber type, mixing order and mixing ratio. The MLP type multi-layer forward network structure includes an input layer, a hidden layer and an output layer. The input layer includes a first input layer, a second output layer and a third output layer, which correspond to color analysis data, component analysis data and quality data respectively. The hidden layer includes a first hidden layer and a second hidden layer. The number of nodes in the first hidden layer is 128R, with Re as the input layer. lu function is used as the activation function; the number of nodes in the second hidden layer of the second hidden layer is 64R, R is a preset value, and the Relu function is used as the activation function; the output layer includes the first output layer, the second output layer and the third output layer, the number of nodes in the first output layer of the first output layer is 1, and the applicable mixing speed is predicted; the number of nodes in the second output layer of the second output layer is 1, and the applicable mixing time is predicted; the third output layer nodes of the third output layer include node 1, node 2 and node 3, the number of nodes is [3, m+2], node 1 is the predicted fiber type, m is the number of mixed fiber types, node 2 is the predicted mixing order, and node 3 is the predicted mixing ratio; After initializing the mixed uniformity adjustment parameter prediction model structure, the mixed uniformity adjustment parameter prediction model is trained using the color analysis data, component analysis data, and quality data as well as the mixed uniformity adjustment parameters corresponding to the color analysis data, component analysis data, and quality data. The optimizer is Adam, and the loss function is the MSE loss function. When the loss function converges, the training ends and the training is completed. Receive the opening degree adjustment parameters and the opening abnormality mark, control the opening equipment to open according to the opening degree adjustment parameters and the opening abnormality mark, obtain the raw material batching, and control the transportation line to start and transport the raw material batching to the mixing equipment; Receive the mixing uniformity adjustment parameter, control the mixing equipment to mix the raw materials according to the mixing uniformity adjustment parameter to obtain uniform materials, and control the transportation line to start conveying the uniform materials to the combing machine.

2. The intelligent control method for monitoring the needle felt production process according to claim 1, characterized in that: The method for constructing the opening degree adjustment parameter prediction model comprises: The looseness adjustment parameter prediction model adopts the VGG model structure, which includes an input layer, D convolution blocks, 2 fully connected layers and an output layer. The image size of the input layer is uniformly planned to be 224×224×U, D and U are preset values, and the convolution blocks include the first convolution block, the second convolution block, the third convolution block, the fourth convolution block and the fifth convolution block; wherein, the convolution blocks all contain two 3×3 convolution layers, and the convolution kernel size in the convolution layer is 3×3. The pooling structure of the convolution block includes a pooling layer and a convergence layer. The pooling structures of the first convolution block and the second convolution block are both pooling layers, and the pooling structures of the third convolution block, the fourth convolution block and the fifth convolution block are all convergence layers. In the pooling structure of the convolution block, the pooling window size is 2×2, with a stride of 2; the convolution layer of the first convolution block has 64 output channels, the convolution layer of the second convolution block has 128 output channels, the convolution layer of the third convolution block has 256 output channels, the convolution layer of the fourth convolution block has 512 output channels, and the convolution layer of the fifth convolution block has 512 output channels; the fully connected layer includes the first fully connected layer and the second fully connected layer; the number of input nodes of the first fully connected layer is 25088, the number of output nodes is 4096, and the activation function is the ReLU function, the number of input nodes of the second fully connected layer is 4096, the number of output nodes is 4096, the activation function is the ReLU function, the number of input nodes of the output layer is 4096, the number of output nodes is 3, and there is no activation function.

3. The intelligent control method for monitoring the needle felt production process according to claim 2, characterized in that: The training method of the opening degree adjustment parameter prediction model comprises: P opening adjustment parameter prediction logs are extracted from the database. Each opening adjustment parameter prediction log includes one opening monitoring image and a set of corresponding opening adjustment parameters and opening abnormality marks. The p opening adjustment parameter prediction logs are divided into a training set and a validation set. The training set is H opening adjustment parameter prediction logs, and the validation set is z opening adjustment parameter prediction logs. The optimizer is the Adam optimizer. The training set is imported into the opening adjustment parameter prediction model to complete the forward calculation from the input layer to the output layer. The cross entropy is used as the loss function. The loss function loos formula is: In the formula, E i is the actual opening adjustment parameter and opening abnormality mark corresponding to the i-th opening monitoring image, Ea(Fh i ) is the predicted opening adjustment parameter and opening abnormality mark of the i-th opening monitoring image, η is a hyperparameter, is the regularization of the weights of the first fully connected layer and the second fully connected layer; i is the number of the looseness monitoring image; According to the optimizer, the parameter values ​​of the parameter prediction model are adjusted according to the degree of relaxation updated. The learning rate is reduced by half after each round of training. The loss function on the validation set is calculated to observe whether the loss function converges. If it converges, the training is terminated.

4. The intelligent control method for monitoring the needle felt production process according to claim 3, characterized in that: The training method of the mixing uniformity adjustment parameter prediction model includes: Converting the color analysis data, the component analysis data and the quality data into a corresponding set of first feature vectors; The color analysis data, component analysis data and quality data are used as inputs of a mixing uniformity adjustment parameter prediction model. The mixing uniformity adjustment parameter prediction model takes a group of mixing uniformity adjustment parameters corresponding to each group of color analysis data, component analysis data and quality data as output, takes the mixing uniformity adjustment parameters actually corresponding to each group of color analysis data, component analysis data and quality data as prediction targets, and takes minimizing the loss function value of the mixing uniformity adjustment parameter prediction model as training target; and stops training when the loss function value of the mixing uniformity adjustment parameter prediction model is less than or equal to a preset target loss value.

5. The intelligent control method for monitoring the needle felt production process according to claim 1, characterized in that: Also includes: Collect temperature and humidity; Input temperature, humidity, opening degree adjustment parameters and mixing uniformity adjustment parameters into the pre-built final adjustment parameter model to output the final adjustment parameters.

6. The intelligent control method for monitoring the needle felt production process according to claim 5, characterized in that: The method for constructing the final adjustment parameter model includes: Taking temperature, humidity, opening degree adjustment parameters and mixing uniformity adjustment parameters as analysis data, collecting final adjustment parameters corresponding to c groups of analysis data in advance, and converting the analysis data and the corresponding final adjustment parameters into a corresponding set of second eigenvectors; Each group of second eigenvectors is used as the input of the final adjustment parameter model. The final adjustment parameter model takes a group of final adjustment parameters corresponding to each group of analysis data as output, and takes the actual final adjustment parameters corresponding to each group of analysis data as the prediction target. The actual final adjustment parameters are the final adjustment parameters corresponding to the analysis data collected in advance. The training target is to minimize the sum of the prediction errors of all analysis data. The calculation formula of the prediction error is Z K =(a K -μ K ) 2 , where Z K is the prediction error, K = 1, 2, ..., c, K is the group number of the second eigenvector corresponding to the analysis data, α K is the final adjustment parameter corresponding to the Kth group of analysis data, μ K is the actual final adjustment parameter corresponding to the Kth group of analysis data; the final adjustment parameter model is trained until the sum of the prediction errors reaches convergence and the training is stopped; the final adjustment parameter model is a deep neural network model.

7. An intelligent control system for monitoring the needle felt production process, characterized in that: The system is used to implement the intelligent control method for monitoring the needle felt production process according to any one of claims 1 to 6, and the system comprises: Collection module: used to collect temperature, humidity, looseness monitoring images and mixing uniformity monitoring data during the production process of needle felt; Preprocessing module: used to preprocess the opening degree monitoring image and mixing uniformity monitoring data; Opening adjustment module: used to input the preprocessed opening degree monitoring image into the pre-built opening degree adjustment parameter prediction model, and output the opening degree adjustment parameter and opening abnormality mark; Mixing adjustment module: used to input the pre-processed mixing uniformity monitoring data into the pre-built mixing uniformity adjustment parameter prediction model, and output the mixing uniformity adjustment parameters; Final adjustment module: used to input temperature, humidity, opening degree adjustment parameters and mixing uniformity adjustment parameters into the pre-built final adjustment parameter model, and output the final adjustment parameters; Final control module: The air conditioner controls the temperature and humidity according to the final adjustment parameters; the opening equipment and mixing equipment are adjusted according to the final adjustment parameters.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the intelligent control method for monitoring the needle felt production process according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed, the intelligent control method for monitoring the needle felt production process according to any one of claims 1 to 6 is implemented.

Images