Seed cotton cleaning machine processing quality prediction, regulation and control method based on digital twinning

Abstract

The invention discloses a seed cotton cleaning machine processing quality prediction and regulation method based on digital twinning, comprising the following steps: (1) collecting material characteristic parameters of to-be-processed seed cotton, obtaining operation parameters and processing quality data of a seed cotton cleaning machine, and establishing a sample data set; (2) constructing a processing quality prediction model of the seed cotton cleaning machine; (3) model training; (4) the processing quality of the seed cotton cleaning machine is predicted, if the actual processing quality meets the requirement of the predicted quality level, processing is carried out according to the current machine parameters, and otherwise, machine parameter optimization is carried out; (5) making a control strategy, constructing a parameter control model, and performing simulation analysis; and (6) obtaining a control instruction according to a simulation analysis result, and inputting the control instruction to a physical control system to complete regulation and control of machine operation parameters. The processing quality can be predicted according to the material characteristics of the seed cotton to be processed and the machine operation parameters, the machine operation parameters are regulated and controlled, the seed cotton processing quality is optimized, and the processing efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of mechanized cotton cleaning and processing, and in particular to a method for predicting and controlling the processing quality of a seed cotton cleaning machine based on digital twins. Background Art

[0002] The processing quality of the seed cotton cleaning machine is related to the equipment operating parameters such as the rotation speed of the spike roller of the seed cotton cleaning machine, the spacing between the spike roller and the grid bars, and the gap between the grid bars. The operating parameters of the seed cotton cleaning machine can only guarantee the processing quality of the seed cotton within a certain range. In order to improve the processing quality of seed cotton, relevant studies have achieved adjustable operating parameters of the seed cotton cleaning machine, but it is necessary to conduct experiments based on the seed cotton processing quality under different parameters to determine the optimal parameters. On the one hand, cotton processing plants did not apply the seed cotton quality detection link in the actual processing process, resulting in a lag in seed cotton processing quality detection, and seed cotton quality prediction was not involved; on the other hand, at the parameter control level of the seed cotton cleaning machine, the current parameter control method is still open-loop control. There are many complex factors in the cleaning and processing of seed cotton that will lead to failure to reach the set parameters, such as unstable cotton flow, sudden high impurity content in seed cotton, etc., which leads to unstable parameter control of the seed cotton cleaning machine. The above problems need to be solved urgently. Summary of the invention

[0003] In response to the above problems, the main purpose of the present invention is to provide a method for predicting and controlling the processing quality of a seed cotton cleaning machine based on digital twins. By predicting the quality of the seed cotton after cleaning, the operating parameters of the seed cotton cleaning machine are optimized, thereby improving the processing quality of the seed cotton.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] A method for predicting and controlling the processing quality of a seed cotton cleaning machine based on digital twins, characterized in that it comprises the following steps:

[0006] (1) Collecting material characteristic parameters of seed cotton to be processed and obtaining machine operation parameters and processing quality data of a seed cotton cleaning machine during processing, and establishing a sample data set;

[0007] (2) Construct a prediction model for the processing quality of seed cotton cleaning machine;

[0008] (3) Model training;

[0009] (4) Predicting the real-time processing quality of the seed cotton cleaning machine; if the actual processing quality meets the requirements of the predicted quality level, the current machine parameters are used for processing; otherwise, the machine parameters are optimized;

[0010] (5) Formulate control strategies, build parameter control models, and conduct simulation analysis;

[0011] (6) The control instructions obtained based on the simulation results are input into the physical control system to complete the regulation of the machine operating parameters.

[0012] The step (1) of collecting material characteristic parameters of the seed cotton to be processed and obtaining machine operation parameters and processing quality data of the seed cotton cleaning machine during processing, and establishing a sample data set specifically includes:

[0013] 1) collecting the moisture regain of the seed cotton to be processed by a seed cotton moisture regain detection device;

[0014] 2) The machine operating parameters such as the rotation speed of the pricking roller, the distance between the pricking roller and the grid bars, the gap between the grid bars and the grid bars, and the inclination of the cleaning part of the seed cotton cleaning machine are collected through magnetic switches, encoders, inclination sensors, etc.

[0015] 3) Use seed cotton quality testing equipment to test the quality of seed cotton after processing, including the impurity content and fiber length of seed cotton.

[0016] 4) The moisture regain of seed cotton from different batches, the machine operating parameters of the seed cotton during processing, and the quality records of the processed seed cotton are used to establish a sample data set and store it in the database for subsequent query and call.

[0017] Furthermore, the data acquisition method of step 1) and step 2) is to communicate the data collected by the sensor with the data acquisition module KEPServer through PLC via Modbus TCP, build an OPC UA server through C# software, use OPC UA communication to collect data to the OPC UA client, and store the collected data in a MySQL database.

[0018] The step (2) constructs a seed cotton cleaning machine processing quality prediction model: an artificial neural network RBF neural network is used, and the radial basis function selected is a Gaussian function, to establish a seed cotton processing quality prediction model, the model includes an input layer, a hidden layer and an output layer: the input layer has n neurons, corresponding to the input feature parameters of the sample data set; the number of hidden layer neurons is r, the number of network outputs is m, the input layer weight ω, the hidden layer bias β, ω is an R×M matrix, β is a 1×M matrix, the specific values ​​are randomly given, and the output layer is the quality of the seed cotton after processing, including the seed cotton impurity content and the seed cotton fiber length.

[0019] The model training of step (3) is as follows: different batches of processing data in the database are used as training samples, and the rotation speed of the spike roller, the spacing between the spike roller and the grid bars, the gap between the grid bars and the grid bars, and the inclination angle of the cleaning part of the seed cotton cleaning machine are used as inputs of the model.

[0020] The processing quality in step (4) specifically includes the impurity content of the seed cotton and the fiber length of the seed cotton.

[0021] The machine parameter optimization in step (4) is to obtain a multivariate function of quality by fitting the nonlinear least square method, and to obtain the optimal solution of the machine operating parameters by the gradient descent method.

[0022] The constructed parameter model in the step (5) includes a thorn roller speed model, a spacing model between the thorn roller and the grid bar, a gap model between the grid bar and the grid bar, and an inclination angle model of the cleaning part of the seed cotton cleaning machine; the control strategy is that the encoder collects the thorn roller speed and transmits it to the PLC, the PLC performs PID control inside, and controls the frequency converter through the digital-to-analog conversion module to adjust the speed of the thorn roller; the simulation analysis model consists of a mechanical model and a digital model.

[0023] Furthermore, the mechanical model was completed in SolidWorks, and the digital model was built in the Simulink module in MATLAB and combined with Adams for dynamics and motion analysis.

[0024] The method for predicting and controlling the processing quality of a seed cotton cleaning machine based on digital twins further includes visualizing the processing quality prediction data of the seed cotton cleaning machine by the client, and displaying the real-time control effect of the machine operating parameters.

[0025] Compared with the prior art, the method for predicting and controlling the processing quality of a seed cotton cleaning machine based on digital twins of the present invention has one or more of the following beneficial technical effects:

[0026] 1. The present invention can use the RBF neural network to predict the quality of seed cotton after processing by the seed cotton cleaning machine, and combine the prediction model with the digital twin. By predicting the quality of seed cotton, it can help the operator adjust the operating parameters of the seed cotton cleaning machine to more appropriate values, optimize the seed cotton processing quality, and improve the processing efficiency.

[0027] 2. The present invention creates a control model for the operating parameters of the seed cotton cleaning machine, which can simulate the actual operating processing effect of the seed cotton cleaning machine under optimized operating parameters, solve the problem of unstable control effect caused by complex environment during processing, and guide the actual operating parameter control. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The accompanying drawings may be used as examples to illustrate the embodiments of the invention or the technical solutions in the prior art. The accompanying drawings show some embodiments of the invention, but are not limited to these examples. Ordinary technicians can obtain other accompanying drawings based on the structures presented in the accompanying drawings without creative work.

[0029] Figure 1 This is a flow chart of the method for predicting and controlling the processing quality of seed cotton cleaning machine based on digital twin;

[0030] Figure 2It is a data collection schematic diagram of the present invention;

[0031] Figure 3 It is the structure diagram of RBF neural network;

[0032] Figure 4 This is a flowchart of seed cotton processing quality prediction based on RBF neural network;

[0033] Figure 5 It is a flow chart for optimizing machine operating parameters. DETAILED DESCRIPTION

[0034] The present invention is further described in detail below through specific examples. The following examples are only illustrative and not restrictive, and the protection scope of the present invention cannot be limited thereto.

[0035] This embodiment discloses a method for predicting and controlling the processing quality of a seed cotton cleaning machine based on digital twins. Figure 1 As shown, Figure 2 As shown, the method comprises the following steps:

[0036] (1) Collecting material characteristic parameters of seed cotton to be processed and obtaining machine operation parameters and processing quality data of a seed cotton cleaning machine during processing, and establishing a sample data set;

[0037] The method of collecting the material characteristic parameters of the seed cotton to be processed and obtaining the machine operation parameters and processing quality data of the seed cotton cleaning machine during processing, and establishing the sample data set includes the following steps:

[0038] 1) collecting the moisture regain of the seed cotton to be processed by a seed cotton moisture regain detection device;

[0039] 2) Collecting machine operating parameters such as the rotation speed of the pricking roller, the spacing between the pricking roller and the grid bars, the gap between the grid bars, and the inclination of the cleaning part of the seed cotton cleaning machine during processing by means of magnetic switches, encoders, inclination sensors, etc.;

[0040] 3) Use seed cotton quality testing equipment to test the quality of the processed seed cotton, including the impurity content and fiber length of the seed cotton;

[0041] 4) The moisture regain of seed cotton from different batches, the machine operating parameters of the seed cotton during processing, and the quality records of the processed seed cotton are used to establish a sample data set and store it in the database for subsequent query and call. The machine operating parameters for processing seed cotton with different moisture regains are different.

[0042] Assume that the production and processing data of the seed cotton cleaning machine includes n batches. Some processing data are shown in Table 1, where n is the production and processing batch, m is the nk is the kth moisture regain data detected when processing the nth batch of seed cotton, r nkis the speed of the spike roller corresponding to the kth moisture regain when processing the nth batch of seed cotton, d nk D is the distance between the spike roller and the grid bar when processing the nth batch of seed cotton corresponding to the kth moisture regain, nk is the gap between the lattices at the kth moisture regain when processing the nth batch of seed cotton, a nk is the inclination angle of the cleaning part of the seed cotton cleaning machine corresponding to the kth moisture regain when processing the nth batch of seed cotton, n nk is the impurity rate of seed cotton detected after processing under the above machine operating parameters when processing the nth batch of seed cotton corresponding to the kth moisture regain, L nk It is the seed cotton fiber length detected after processing under the above machine operating parameters when processing the nth batch of seed cotton corresponding to the kth moisture regain.

[0043] Table 1

[0045] The data acquisition in the present invention is to communicate the data collected by the sensor with the data acquisition module KEPServer through PLC via Modbus TCP, build an OPC UA server through C# software, use OPC UA communication to collect data to the OPC UA client, and store the collected data in the MySQL database.

[0046] In this way, the seed cotton processing qualities under different moisture regains and different machine operating parameters are matched one by one and used as samples for training the seed cotton cleaning machine processing quality prediction model. Of course, this should also include the processing of abnormal data.

[0047] (2) Construct a prediction model for the processing quality of seed cotton cleaning machine;

[0048] The artificial neural network is used to establish a prediction model for seed cotton processing quality. The artificial neural network used is RBF neural network, and the radial basis function selected is Gaussian function. The artificial neural network RBF neural network is used, and the radial basis function selected is Gaussian function. The prediction model for seed cotton processing quality is established. Figure 3 As shown, the model includes an input layer, a hidden layer, and an output layer: the input layer has n neurons, corresponding to the input feature parameters of the sample data set; the number of neurons in the hidden layer is r, the number of outputs of the network is M, and the input layer weights , hidden layer bias , is an R×M matrix, It is a 1×M matrix, the specific values ​​are randomly given, and the output layer is the quality of the seed cotton after processing, including the impurity content of the seed cotton and the fiber length of the seed cotton.

[0049] (3) Model training;

[0050] Using different batches of processing data in the database as training samples, the speed of the spike roller, the spacing between the spike roller and the grid, the gap between the grids, and the inclination of the cleaning part of the seed cotton cleaner are used as inputs of the model. The training model specifically includes the following:

[0051] like Figure 3 , Figure 4 As shown, let the input data vector of the network be , n is the dimension of the input data, the number of neurons in the hidden layer is r, the number of network outputs is m, the radial basis function selected is the Gaussian function, and the network output can be expressed as:

[0052]

[0053] In the formula: is the mth output value of the output layer, m=1,2,…M; is the weight of the rth hidden layer neuron corresponding to the mth output, r=1,2,…R; is the bias value; is the Gaussian function operator, and the Gaussian function form is: , first define the loss function, which is the mean square error between the network output value and the expected value, that is , the gradient descent method is used to adjust the parameters, and the partial derivative of the loss function with respect to each parameter is:

[0054]

[0055]

[0056] ,

[0057] To minimize the loss function, the correction amount is proportional to the negative gradient direction, then:

[0058]

[0059]

[0060]

[0061] , is the learning rate.

[0062] The specific steps to achieve seed cotton processing quality include:

[0063] 1. Call the MySQL database to use the processing quality record data of different batches of seed cotton as sample data for training the seed cotton cleaning machine processing quality prediction model. The speed of the spike roller, the spacing between the spike roller and the grid, the gap between the grid and the grid, and the inclination angle of the cleaning part of the seed cotton cleaning machine are the inputs of the network. The seed cotton impurity rate and cotton fiber length are used as evaluation indicators of processing quality. The training data needs to be normalized and preprocessed.

[0064] 2. Randomly initialize the parameters of the RBF network by the parameter center vector ,width , weight and bias , the hidden layer neurons are Gaussian functions, and the output layer neurons use linear functions as activation functions.

[0065] 3. Input the training data into the RBF neural network and calculate whether the loss function meets the accuracy. If so, set the current parameters.

[0066] 4. Use the gradient descent method to adjust the network parameters in step 2 and continue training. If the maximum number of training times is reached, set the current network parameters.

[0067] 5. After network training is completed, set network parameters.

[0068] 6. Input the test data and predict the quality of seed cotton processing.

[0069] (4) Predicting the real-time processing quality of the seed cotton cleaning machine; if the actual processing quality meets the requirements of the predicted quality level, the current machine parameters are used for processing; otherwise, the machine parameters are optimized;

[0070] Specifically, the predicted processing quality includes the impurity content of seed cotton and the length of cotton fibers. When the actual processing quality reaches the predicted quality level, the current machine operating parameters are used for processing and the process proceeds to step (5). When the actual processing quality does not reach the predicted quality level, the machine operating parameters need to be optimized, such as Figure 5 Specifically, based on the processing data that meets the quality requirements, the multivariate function of quality is obtained by fitting the nonlinear least squares method using MATLAB, and the optimal solution of the machine operating parameters is obtained by the gradient descent method. After the machine operating parameters are optimized, step (5) is performed.

[0071] (5) Formulate control strategies, build parameter control models, and conduct simulation analysis;

[0072] The control strategy adopted is that the encoder collects the speed of the spike roller and transmits it to the PLC, and the PLC performs PID control inside, and controls the frequency converter through the digital-to-analog conversion module to adjust the speed of the spike roller. A parameter control model is constructed for simulation analysis, mainly including the spike roller speed model, the spacing model between the spike roller and the grid bar, the gap model between the grid bar and the grid bar, and the inclination model of the cleaning part of the seed cotton cleaning machine. The simulation analysis model consists of two parts: a mechanical model and a digital model. The mechanical model is completed in SolidWorks, and the digital model is built in the Simulink module in MATLAB and combined with Adams for dynamics and motion analysis.

[0073] (6) The control instructions obtained based on the simulation results are input into the physical control system to complete the regulation of the machine operating parameters.

[0074] Through simulation analysis, physical equipment is selected, RS485 is used as the communication module, Siemens PLC is used as the lower computer controller combined with a digital-to-analog conversion module, and the speed of the thorn roller is adjusted by adjusting the speed of the asynchronous motor through a vector inverter. The spacing between the thorn roller and the grid, the gap between the grids, and the inclination of the cleaning part of the seed cotton cleaner are adjusted by inputting control instructions to the PLC to control the stepper motor and servo motor.

[0075] Specifically, the visualization of the processing quality prediction data of the seed cotton cleaning machine, as well as the simulation of the machine operating parameters and the real-time control effects are displayed through the communication between the TCP / IP module in the Simulink model and the C# software to realize the display of the simulation data. The display interface is developed using Winform, and the real-time control data is communicated with C#S7.NET through the sensor via PLC to complete the collection of real-time control data.

[0076] Finally, it should be noted that the above is only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred arrangement scheme, a person skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for predicting and controlling the processing quality of a seed cotton cleaning machine based on digital twins, characterized in that: The following steps are involved: (1) Collecting material characteristic parameters of seed cotton to be processed and obtaining machine operation parameters and processing quality data of a seed cotton cleaning machine during processing, and establishing a sample data set; (2) Construct a prediction model for the processing quality of seed cotton cleaning machine; (3) Model training; (4) Predicting the real-time processing quality of the seed cotton cleaning machine; if the actual processing quality meets the requirements of the predicted quality level, the current machine parameters are used for processing; otherwise, the machine parameters are optimized; (5) Formulate control strategies, build parameter control models, and conduct simulation analysis; (6) The control instructions obtained based on the simulation results are input into the physical control system to complete the regulation of the machine operating parameters.

2. The method for predicting and controlling the processing quality of a seed cotton cleaning machine based on digital twin according to claim 1, characterized in that: The step (1) of collecting the material characteristic parameters of the seed cotton to be processed and obtaining the machine operation parameters and processing quality data of the seed cotton cleaning machine during processing, and establishing the sample data set specifically includes: 1) collecting the moisture regain of the seed cotton to be processed by a seed cotton moisture regain detection device; 2) Collecting machine operating parameters such as the rotation speed of the pricking roller, the spacing between the pricking roller and the grid bars, the gap between the grid bars, and the inclination of the cleaning part of the seed cotton cleaning machine during processing by means of magnetic switches, encoders, inclination sensors, etc.; 3) Use seed cotton quality testing equipment to test the quality of the processed seed cotton, including the impurity content and fiber length of the seed cotton; 4) Record the moisture regain of seed cotton from different batches, the machine operating parameters of the seed cotton during processing, and the quality of the processed seed cotton, and then establish a sample data set and store it in the database for subsequent query and call.

3. The method for predicting and controlling the processing quality of a seed cotton cleaning machine based on digital twin according to claim 1, characterized in that: The step (2) constructs a seed cotton cleaning machine processing quality prediction model: an artificial neural network RBF neural network is used, and the radial basis function selected is a Gaussian function, to establish a seed cotton processing quality prediction model, the model includes an input layer, a hidden layer and an output layer: the input layer has n neurons, corresponding to the input feature parameters of the sample data set; the number of neurons in the hidden layer is r, the number of outputs of the network is m, and the input layer weights are , hidden layer bias , is an R×M matrix, It is a 1×M matrix, the specific values ​​are randomly given, and the output layer is the quality of the seed cotton after processing, including the impurity content of the seed cotton and the fiber length of the seed cotton.

4. The method for predicting and controlling the processing quality of a seed cotton cleaning machine based on digital twin according to claim 1, characterized in that: The model training of step (3) is as follows: different batches of processing data in the database are used as training samples, and the rotation speed of the spike roller, the spacing between the spike roller and the grid bars, the gap between the grid bars and the grid bars, and the inclination angle of the cleaning part of the seed cotton cleaning machine are used as inputs of the model.

5. The method for predicting and controlling the processing quality of a seed cotton cleaning machine based on digital twin according to claim 1, characterized in that: The processing quality in step (4) specifically includes the impurity content of the seed cotton and the fiber length of the seed cotton.

6. The method for predicting and controlling the processing quality of a seed cotton cleaning machine based on digital twinning according to claim 1, characterized in that: The machine parameter optimization described in step (4) is to obtain the multivariate function of quality by fitting the nonlinear least squares method, and to obtain the optimal solution of the machine operating parameters by the gradient descent method.

7. The method for predicting and controlling the processing quality of a seed cotton cleaning machine based on digital twin according to claim 1, characterized in that: The constructed parameter model in the step (5) includes a thorn roller speed model, a spacing model between the thorn roller and the grid bar, a gap model between the grid bar and the grid bar, and an inclination angle model of the cleaning part of the seed cotton cleaning machine; the control strategy is that the encoder collects the thorn roller speed and transmits it to the PLC, the PLC performs PID control inside, and controls the frequency converter through the digital-to-analog conversion module to adjust the speed of the thorn roller; the simulation analysis model consists of a mechanical model and a digital model.

8. The method for predicting and controlling the processing quality of a seed cotton cleaning machine based on digital twin according to claim 2, characterized in that: The data acquisition method described in step 1) and step 2) is to communicate the data collected by the sensor with the data acquisition module KEPServer through PLC via ModbusTCP, build an OPC UA server through C# software, use OPC UA communication to collect data to the OPC UA client, and store the collected data in a MySQL database.

9. The method for predicting and controlling the processing quality of a seed cotton cleaning machine based on digital twin according to claim 7, characterized in that: The mechanical model was completed in SolidWorks, and the digital model was built in the Simulink module in MATLAB and combined with Adams for dynamics and motion analysis.

10. The method for predicting and controlling the processing quality of a seed cotton cleaning machine based on digital twins according to claims 1-9, characterized in that: It further includes the client's visualization of the seed cotton cleaning machine's processing quality prediction data and the display of the real-time control effects of the machine's operating parameters.

Images