A method and system for predicting flow field characteristics of a cotton picker pneumatic conveying system based on digital twinning

Abstract

This invention provides a method and system for predicting the flow field characteristics of a cotton harvester pneumatic conveying system based on digital twins, including the following steps: acquiring simulation data of the cotton harvester pneumatic conveying system, data preprocessing, core sample screening, constructing a node velocity prediction model, generating a rendering cloud map, and building a visualization module based on digital twins. This invention establishes a model for the input fan speed and multiple sets of different types of output parameters through polynomial chaotic expansion, and uses digital twin technology to visualize the planar flow field at the air outlet and the planar flow field of the impeller ring. This invention obtains the flow field data of each node in the cotton harvester pneumatic conveying system through a numerical simulation model, identifies key nodes using a data clustering algorithm, reduces computational costs, and improves response speed; it uses polynomial chaotic expansion to establish a velocity prediction model for key nodes, and then generates a flow field cloud map based on the node prediction results, achieving rapid prediction of the flow field characteristics of the cotton harvester pneumatic conveying system.

Description

Technical Field

[0001] This invention belongs to the interdisciplinary field of agricultural engineering, big data and artificial intelligence, and in particular relates to a method and system for predicting the flow field characteristics of a cotton harvester pneumatic conveying system based on digital twins. Background Technology

[0002] my country's cotton production has seen rapid mechanization, with machine-harvested cotton accounting for a gradually increasing proportion. Harvesting is a key step in mechanized cotton production, which was previously mainly done manually, resulting in low efficiency and high costs. To improve harvesting efficiency and reduce costs, cotton harvesters have become the main mode of cotton harvesting in my country.

[0003] The cotton harvester separates seed cotton from the cotton plant through the picking head spindle, then separates the seed cotton from the picking spindle through the cotton stripping disc, and finally transports the seed cotton to the cotton collection box through the pneumatic conveying system, thus completing the cotton harvesting process.

[0004] In the pneumatic conveying system of a cotton harvester, a centrifugal fan provides the airflow for transporting seed cotton. However, in existing pneumatic conveying systems, the fan speed and the angle of the air distribution plate are fixed, resulting in a fixed flow field characteristic that cannot be adjusted in real time according to the cotton harvester's operating conditions.

[0005] To achieve adaptive control of fan speed, air distribution plate angle, and other parameters according to operating conditions, it is necessary to obtain the flow field characteristics of the pneumatic conveying system under given parameters such as fan speed.

[0006] Existing flow field measurement methods mostly use Pitot tubes and other measuring devices, which can only measure the velocity at a single point and cannot obtain parameters such as cross-sectional velocity, i.e., the measurement of flow field characteristics.

[0007] Commonly used flow field analysis methods include analytical methods and numerical simulation methods. Among them, analytical methods simplify the scenario to a certain extent, resulting in lower solution accuracy; numerical simulation methods offer high solution accuracy but take a long time to solve, which cannot meet the needs of rapid flow field estimation in pneumatic conveying systems.

[0008] Therefore, how to obtain the flow field characteristics of the pneumatic conveying system based on numerical simulation analysis data, and how to achieve rapid acquisition of flow field characteristics through technologies such as digital twins, has become an urgent issue to be addressed in obtaining the flow field characteristics of the pneumatic conveying system of cotton harvesters. Summary of the Invention

[0009] To address the aforementioned technical problems, this invention provides a method and system for predicting the flow field characteristics of a cotton harvester's pneumatic conveying system based on digital twins. It establishes a model of the input fan speed and multiple sets of different types of output parameters using polynomial chaotic expansion (PCE), and utilizes digital twin technology to visualize the planar flow field at the air outlet and the planar flow field of the impeller ring. This invention obtains flow field data for each node of the cotton harvester's pneumatic conveying system through numerical simulation models, identifies key nodes using data clustering algorithms, reduces computational costs, and improves response speed. It then uses polynomial chaotic expansion to establish velocity prediction models for key nodes, and generates flow field cloud maps based on the node prediction results, enabling rapid prediction of the flow field characteristics of the cotton harvester's pneumatic conveying system.

[0010] Note that the description of these objectives does not preclude the existence of other objectives. One aspect of the invention does not require achieving all of the above objectives. Objectives other than those described above can be extracted from the description, drawings, and claims.

[0011] The present invention achieves the above-mentioned technical objectives through the following technical means.

[0012] A method for predicting the flow field characteristics of a pneumatic conveying system for a cotton harvester based on digital twins includes the following steps:

[0013] Step S1: Acquisition of simulation data for the pneumatic conveying system of the cotton harvester: Establish a high-fidelity numerical simulation model for predicting the flow field characteristics of the pneumatic conveying system of the cotton harvester, and obtain the simulation data of the flow field performance of the pneumatic conveying system of the cotton harvester based on the simulation model.

[0014] Step S2, Data Preprocessing: The flow field simulation data of the cotton harvester pneumatic conveying system obtained in Step S1 is preprocessed to identify and remove outliers and anomalies within the data, forming training datasets for each node.

[0015] Step S3, Core Sample Screening: The training dataset obtained in Step S2 is processed. Considering the problem of too many data samples, cost and noise, the training data is clustered by fuzzy clustering algorithm, and the core data after screening is used as the training dataset.

[0016] Step S4: Node velocity prediction model construction: Based on polynomial chaotic expansion, establish velocity prediction models for each node, use leave-one-out method to construct training samples and validation samples, use the training dataset obtained in step S3 to construct velocity prediction models for each node of the plane flow field at the cotton harvester outlet and the plane flow field of the impeller ring expansion, and use the validation set to validate the velocity prediction models.

[0017] Step S5, Rendering Cloud Map Generation: Preprocess the prediction model obtained in step S4, optimize the number of prediction models, optimize the rendering cloud map generation rate, reduce rendering time, speed up the display speed of cloud map, and call the prediction model to calculate and generate real-time rendering cloud map display based on the rendering software.

[0018] Step S6, Visualization: Build a visualization module based on digital twins. Based on the prediction model obtained in step S5, establish a visualization interface display based on the software architecture.

[0019] In the above scheme, step S1, obtaining simulation data of the flow field performance of the cotton harvester's pneumatic conveying system based on the simulation model, specifically includes the following steps:

[0020] Step S1.1: Establish a high-fidelity numerical simulation model for predicting the flow field characteristics of the pneumatic conveying system of the cotton harvester, based on the overall structure of the fan duct of the cotton harvester.

[0021] Step S1.2: Perform mesh node division for the planar flow field at the air outlet of the cotton harvester and the planar flow field of the impeller ring;

[0022] Step S1.3: Simulate and obtain the performance data of all grid nodes of the air outlet planar flow field and the impeller annular unfolded planar flow field under the corresponding fan operation data.

[0023] In the above scheme, step S1 calculates the flow velocity data of each node using computational fluid dynamics, finite element method or finite volume method to obtain flow field simulation data of the cotton harvester pneumatic conveying system.

[0024] In the above scheme, step S2 uses isolated forests to identify and remove outliers, which solves the problem of abnormal training model data.

[0025] In the above scheme, in step S3, a fuzzy clustering algorithm is used to reduce the number of samples, thereby reducing computational costs and accelerating the response speed of the prediction model. The core data after filtering is used as the training dataset and saved to the database.

[0026] In the above scheme, step S4, which establishes a flow field characteristic prediction model based on polynomial chaotic expansion, specifically includes the following steps:

[0027] Step S4.1: Based on the processed dataset, generate training and validation sets using the leave-one-out method;

[0028] Step S4.2: Establish a flow field characteristic prediction model using polynomial chaotic expansion;

[0029] Step S4.3: Validate the performance of the flow field characteristic prediction model using the validation set and the coefficient of determination. To evaluate the accuracy and reliability of the model.

[0030] Furthermore, step S4.2, which involves establishing a flow field characteristic prediction model based on polynomial chaotic expansion, includes the following steps:

[0031] Construct a basis function y based on the input and output and their relationship;

[0032] Calculate the chaotic PC coefficients of an unknown polynomial;

[0033] The relationship between input x and output y is expressed as follows:

[0034] ;

[0035] This represents a polynomial chaotic expansion model that attempts to approximate or fit the true system output response y.

[0036] It is the symbol for the set of n-dimensional non-negative integers;

[0037] p is the maximum order of the polynomial chaotic expansion;

[0038] Where y is the output;

[0039] x is the input;

[0040] The tensor product of univariate orthogonal polynomials satisfies the following formula:

[0041] ;

[0042] This refers to the i-th variable in the input;

[0043] Let be the polynomial chaotic coefficients to be determined, i.e., the PC coefficients;

[0044] The total number of unknown PC coefficients P satisfies the following formula:

[0045] ;

[0046] P is the maximum order of the input variable;

[0047] The dimension of the input variable;

[0048] Let be a polynomial whose total order |k| does not exceed the given order k.

[0049] To calculate the PC value, we need to find the polynomial chaotic expansion PCE that satisfies the following formula:

[0050]

[0051] Where N is the number of input-output samples;

[0052] The above formula can be rewritten as:

[0053] ;

[0054] in, ;

[0055] This refers to representing a P-dimensional real vector.

[0056] To find the PC coefficient vector;

[0057] ;

[0058] y is the output response vector. Represents an N-dimensional real vector;

[0059] ;

[0060] This is a measurement matrix, where each column contains the evaluation values ​​of the PC basis for N samples;

[0061] Represents an N A 3D real matrix;

[0062] The PC coefficients are calculated by minimizing the residual between the model response and the PCE approximation;

[0063] for The unknown coefficients are calculated using least squares regression, as shown in the following formula:

[0064] ;

[0065] when Then, the problem of solving for the PC coefficients is transformed into a problem of minimizing ℓ1:

[0066] ;

[0067] The ℓ2 norm constraint is used to consider the truncation error of the PCE. ;

[0068] The constraint ℓ1 minimization problem in the above formula can be reformulated as a regularized (unconstrained) optimization problem:

[0069] ;

[0070] Let be the ℓ1 norm of the PC coefficients;

[0071] This represents the degree of fit between the model and the actual model response in the ℓ2 norm sense.

[0072] In the above scheme, step S5, optimizing the rendering cloud map generation rate, includes the following steps:

[0073] Step S5.1: Based on the numerical simulation model, obtain the flow field data of multiple nodes at different wind turbine speeds, and retain the coordinate data of each node. Load the flow field data and coordinate data, establish a proxy model based on polynomial chaotic expansion, take the coordinate data as input and the node wind speed at a certain speed as output, and construct a prediction model at different speeds based on polynomial chaotic expansion.

[0074] Step S5.2: Load all node velocity prediction models from step S5.1. Based on the real-time response rate requirements of the cloud map, divide the horizontal and vertical coordinates into k equal parts within the boundary of the cloud map, determine the intersection of the divisions as the horizontal and vertical coordinate values, and combine to generate (k+1)×(k+1) uniform node coordinate values. Using the coordinates of all uniform nodes as input, load the prediction model and predict all data under each rotational speed. This invention uses the coordinates of all uniform nodes as input to establish a proxy model using polynomial chaotic expansion, and predicts all data under each working condition. Compared with directly calling the simulation node data and calculating and rendering the real-time cloud map display based on rendering software, the processed uniform node data is more uniform, the amount of calculation is smaller, and the response speed is faster.

[0075] Step S5.3: Load all data of uniform nodes, take the speed of each wind turbine as input and the flow velocity data corresponding to each node as output, generate a flow velocity prediction model of (k+1)×(k+1) uniform nodes, call all prediction models through rendering software according to the input wind turbine speed and calculate the node wind velocity value under each coordinate, and arrange the wind velocity data on the cloud map plane according to the corresponding coordinate order, and finally use rendering software to interpolate and generate cloud map according to the data numerical relationship in the plane.

[0076] In the above scheme, the visualization interface in step S6 includes real-time display of the average wind speed in the central region of the flow field, the highest wind speed in the central region, the lowest wind speed in the central region, the outlet flow rate, the average operating power of the fan, as well as real-time display of the planar flow field at the outlet and the expanded planar flow field of the impeller ring.

[0077] Preferably, the average wind speed in the central region of the flow field, the highest wind speed in the central region, the lowest wind speed in the central region, the outlet flow rate, and the average operating power of the fan are displayed in real time using a line graph, and the planar flow field at the outlet and the planar flow field of the impeller ring are displayed in real time using a cloud map.

[0078] A system for implementing the method for predicting the flow field characteristics of a pneumatic conveying system for a cotton harvester based on digital twins includes a simulation data acquisition module, a data preprocessing module, a core sample screening module, a node flow velocity prediction model construction module, a rendering cloud map generation module, and a visualization module.

[0079] The simulation data acquisition module is used to establish a high-fidelity numerical simulation model for predicting the flow field characteristics of the pneumatic conveying system of the cotton harvester, and to obtain simulation data of the flow field performance of the pneumatic conveying system of the cotton harvester based on the simulation model.

[0080] The data preprocessing module is used to preprocess the obtained flow field simulation data of the cotton harvester pneumatic conveying system, identify and remove outliers and anomalies in the data, and form training datasets for each node.

[0081] The core sample filtering module is used to process the obtained training dataset. Considering the problems of too many data samples, cost and noise, the training data is clustered by fuzzy clustering algorithm, and the core data after filtering is used as the training dataset.

[0082] The node velocity prediction model building module is used to build a velocity prediction model for each node based on polynomial chaotic expansion. It uses the leave-one-out method to construct training samples and validation samples. It uses the training dataset obtained in step S3 to build velocity prediction models for each node of the plane flow field at the cotton harvester outlet and the plane flow field of the impeller ring expansion. It uses the validation set to validate the velocity prediction model.

[0083] The cloud map generation module is used to preprocess the obtained prediction models, optimize the number of prediction models, optimize the cloud map generation rate, reduce rendering time, speed up the display of cloud maps, and call the prediction models to calculate and generate real-time rendered cloud maps based on the rendering software.

[0084] The visualization module is used to build a visualization platform based on digital twins. It is based on the prediction model obtained in step S5 and establishes a visualization interface display based on the software architecture.

[0085] Compared with the prior art, the beneficial effects of the present invention are:

[0086] According to one aspect of the present invention, the method first establishes a high-fidelity numerical simulation model of the flow field of the pneumatic conveying system of a cotton harvester, and divides the flow field of the pneumatic conveying system, such as the cotton harvester's air outlet and impeller ring, into grid nodes, acquires operational data and flow field data of each node in the pneumatic conveying system of the cotton harvester, preprocesses outliers and anomalies in the flow field data, then filters the core data using a fuzzy clustering algorithm, and establishes a flow field node velocity prediction model using polynomial chaotic expansion; the method then uses rendering software to call the prediction model to generate prediction data, interpolates the data to generate cloud maps, and finally displays the data through a visualization panel.

[0087] According to one aspect of the present invention, the system of the present invention includes a simulation data acquisition module, a data preprocessing module, a core sample screening module, a node flow velocity prediction model construction module, a rendering cloud map generation module, and a visualization module. The entire device can clearly display the working status of the cotton harvester in real time and monitor the flow field performance data of the cotton harvester's pneumatic conveying system.

[0088] This invention utilizes digital twin technology and polynomial chaotic expansion to predict the flow field characteristics of a cotton harvester's pneumatic conveying system. It enables real-time monitoring of the flow field of the cotton harvester's pneumatic conveying system, solving the problem of the inability to quickly predict and obtain the flow field characteristics of the cotton harvester's pneumatic conveying system, and providing a reference for the control of the pneumatic conveying system.

[0089] Note that the description of these effects does not preclude the existence of other effects. One aspect of the invention does not necessarily have to have all of the above.

[0090] The effects described above are obvious from the description, drawings, claims, etc. Attached Figure Description

[0091] Figure 1 A schematic diagram of the flow field characteristic prediction method for a cotton harvester pneumatic conveying system based on digital twin according to an embodiment of the present invention;

[0092] Figure 2 A high-fidelity numerical simulation result of a cotton harvester fan according to an embodiment of the present invention is shown in the figure.

[0093] Figure 3 A schematic diagram of an outlier identification algorithm according to one embodiment of the present invention;

[0094] Figure 4 A schematic diagram illustrating the effect of a case clustering algorithm according to one embodiment of the present invention;

[0095] Figure 5 A rendering cloud map of the planar flow field at the air outlet, according to one embodiment of the present invention;

[0096] Figure 6 An example of a case study of the impeller annulus unfolded planar flow field rendering cloud map according to one embodiment of the present invention;

[0097] Figure 7 A real-time line graph showing the fan speed, fan power, and minimum flow velocity in the central flow field of a cotton harvester according to one embodiment of the present invention;

[0098] Figure 8 A real-time line graph comparing the maximum flow velocity, minimum flow velocity, average flow velocity, and real-time measured flow velocity in the central flow field of an embodiment of the present invention. Detailed Implementation

[0099] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0100] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "front," "rear," "left," "right," "upper," "lower," "axial," "radial," "vertical," "horizontal," "inner," and "outer," etc., indicating orientation or positional relationships based on the orientation or positional relationships shown in the accompanying drawings, are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0101] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0102] Implementation Case 1

[0103] Figure 1 The image shows a preferred embodiment of the method for predicting the flow field characteristics of a cotton harvester pneumatic conveying system based on digital twins. The method includes the following steps:

[0104] Step S1: Establish a high-fidelity numerical simulation model for predicting the flow field characteristics of the pneumatic conveying system of the cotton harvester. Divide the planar flow field at the air outlet of the pneumatic conveying system and the planar flow field of the impeller ring into 3154 nodes, retain the coordinate position corresponding to each node, and obtain 10 sets of flow field data for each node in the speed range of 1000 rpm to 5000 rpm based on the simulation model.

[0105] Step S2: Preprocess the flow field simulation performance data of the cotton harvester pneumatic conveying system obtained in Step S1, and use isolated forest to identify and remove outliers and anomalies in the data to form training datasets for each node.

[0106] Step S3: Process the training dataset from step S2. Considering the large number of data samples, cost, and noise, cluster the training data using a fuzzy clustering algorithm and use the filtered core data as the training dataset.

[0107] Step S4: Establish velocity prediction models for each node based on polynomial chaotic expansion. Construct training and validation samples using the leave-one-out method. Utilize the training dataset obtained in Step S3 to construct velocity prediction models for each node in the planar flow field at the cotton harvester outlet and the impeller annular expanded planar flow field. Then, validate the velocity prediction models using the validation set and employ the coefficient of determination. To evaluate the accuracy and reliability of the model, the trained node flow rate prediction model is stored in the model storage module for use by the digital twin system.

[0108] Step S5: Preprocess the prediction model obtained in step S4, optimize the number of prediction models, optimize the generation rate of the rendering cloud map, reduce the rendering time, speed up the display speed of the cloud map, and call the prediction model to calculate and generate a real-time rendering cloud map based on the rendering software.

[0109] Step S6: Build a visualization module based on digital twin. Based on the prediction model obtained in step S5, establish a visualization interface display based on the software architecture. The visualization interface displays real-time line graphs of average flow velocity in the central region of the flow field, maximum flow velocity in the central region, minimum flow velocity in the central region, air outlet flow rate, and average operating power of the fan, as well as real-time display of the planar flow field at the air outlet and the planar cloud map of the impeller annulus.

[0110] In step S1, numerical analysis methods, including but not limited to computational fluid dynamics, finite element method, and finite volume method, are used to calculate the flow velocity data at each node to obtain simulation data of the flow field of the cotton harvester's pneumatic conveying system. For example... Figure 2 The image shown is a high-fidelity numerical simulation result of a cotton harvester fan. Based on the simulation model, grid nodes were divided, and multi-node flow velocity data was obtained through simulation, providing data support for the subsequent establishment of a flow field characteristic prediction model based on polynomial chaotic expansion. The specific steps for identifying outliers in isolated forests in step S2 include:

[0111] Step S2.1: Use a random hyperplane to cut any data space to obtain two data subspaces, then use a random hyperplane to cut the data subspaces again, repeating this operation until there is only one data point in each subspace;

[0112] Step S2.2: Construct an isolated forest consisting of a specified number of isolated binary trees;

[0113] Step S2.3: Randomly select several sample points from dataset D as the sample set for generating a single isolated binary tree. ;

[0114] Step S2.4, from the sample set Randomly select a feature and a cut value ;

[0115] Step S2.5, if node All samples included in the features The maximum and minimum values ​​are respectively and Then there is ;

[0116] Step S2.6, Samples with regard to features The value is less than the cutting value Then the sample is assigned to a node. If the node is assigned to the left of the node, then assign it to the right of the node.

[0117] Step S2.7: Repeat the process for nodes. The left and right child nodes are divided to generate an isolated binary tree;

[0118] Step S2.8: When a child node contains multiple identical data entries, or only one data entry, or when the isolated binary tree has reached the set maximum height, stop generating the isolated binary tree.

[0119] Step S2.9: Based on the test sample points Path length in each isolated binary tree ,calculate Outliers, path length

[0120] The smaller the value, the higher the anomaly score for that sample point.

[0121]

[0122] Where m is the sample set The total number of sample points;

[0123] For all path lengths The average value;

[0124] It is the normalized height of the tree. k is the input value. is Euler's constant.

[0125] like Figure 3 The diagram shows a case study of an outlier identification algorithm. By using isolated forests to identify outliers, the algorithm solves the problem of outliers in the training model data.

[0126] The specific steps for fuzzy clustering in step S3 to screen samples include:

[0127] Step S3.1: Construct the following cost function training dataset point Data sets:

[0128]

[0129] Where J is the cost function, representing the sum of weighted distances from data points to their cluster centers throughout the entire clustering process;

[0130] n is the total number of data points, and n equals 3154;

[0131] c represents the number of clusters, dividing the nodes into 300 clusters;

[0132] =[ , , ..., ] T For each data point in the training dataset,

[0133] Representing data The Middle One attribute,

[0134] Representing data The Middle One attribute,

[0135] Representing data The Middle One attribute, It is a positive integer;

[0136] For each type of data, a class prototype is provided.

[0137] For fuzzy coefficients;

[0138] Indicates Euclidean distance;

[0139] Membership degree represents the number of data points. Belonging to the The class has the possibility of satisfying the following conditions:

[0140] ;

[0141] Step S3.2: Construct the membership matrix =[ ] Obtain the clustering cost function The necessary iterative condition for minimization:

[0142] ( )

[0143] ( )

[0144] Step S3.3: Iterate the above formula repeatedly until the difference between the previous generation and the next generation of all elements in the membership matrix is ​​less than a certain set threshold or the maximum number of iterations is reached and the calculation terminates, thus obtaining the optimal data classification result;

[0145] Step S3.4: For each subclass, select the class prototype as a representative sample. After sample screening, there are 1675 flow rate data corresponding to the rotation speed under each node, which reduces the amount of computation for model building.

[0146] like Figure 4 The diagram shows the effect of the clustering algorithm. By using fuzzy clustering algorithm to select representative samples, the problem of excessive data volume in the training model is solved.

[0147] By treating the cluster prototypes obtained from data clustering as core data, the goal of reducing the training data sample size and computational cost can be achieved.

[0148] Represent the core dataset as ;

[0149] Step S4, establishing the flow field prediction model for the nodal velocity of the pneumatic conveying outlet of the cotton harvester and the impeller annular expanded planar flow field, includes the following steps:

[0150] Step S4.1: Construct a basis function y based on the input and output and their relationship. In this embodiment, the basis function y is constructed based on the fan speed and node flow velocity data.

[0151] Step S4.2: Calculate the unknown PC coefficients;

[0152] Furthermore, based on the relationship between the input x and the output y in step S3.1 (i.e., the relationship between the fan speed of the cotton harvester and the air velocity at the fan outlet), the formula is as follows:

[0153] ;

[0154] Where y is the output, which is the wind speed at the nodes of the watershed;

[0155] x is the input, which is the fan speed;

[0156] The tensor product of univariate orthogonal polynomials satisfies the following formula:

[0157] ;

[0158] Let be the polynomial chaotic coefficients (PC coefficients) to be determined.

[0159] The total number of unknown PC coefficients P satisfies the following formula:

[0160] ;

[0161] P is the maximum order of the input variable;

[0162] n is the dimension of the input variable;

[0163] Total order A polynomial of order p not exceeding a given value.

[0164] To calculate the PC value, we need to find the polynomial chaotic expansion PCE that satisfies the following formula:

[0165]

[0166] Where N is the number of input-output samples;

[0167] The above formula can be rewritten as:

[0168] ;

[0169] in, ;

[0170] This refers to a real vector with P dimensions;

[0171] To find the PC coefficient vector;

[0172] ;

[0173] y is the output response vector. Represents an N-dimensional real vector;

[0174] ;

[0175] This is a measurement matrix, where each column contains the evaluation values ​​of the PC basis for N samples;

[0176] Represents an N A 3D real matrix;

[0177] The PC coefficients are calculated by minimizing the residual between the model response and the PCE approximation.

[0178] for The unknown coefficients are calculated using least squares regression, as shown in the following formula:

[0179] ;

[0180] when Then, the problem of solving for the PC coefficients is transformed into a problem of minimizing ℓ1:

[0181] ;

[0182] The ℓ2 norm constraint is used to consider the truncation error of the PCE. ;

[0183] The constraint ℓ1 minimization problem in the above formula can be reformulated as a regularized (unconstrained) optimization problem:

[0184] ;

[0185] Let be the ℓ1 norm of the PC coefficients;

[0186] The degree of fit between the model and the actual model response in the ℓ2 norm sense;

[0187] In step S4, the leave-one-out method generates training and validation samples from the constructed dataset. Specifically, the leave-one-out method is a commonly used cross-validation method for evaluating model performance. For the flow velocity prediction model of each node in the flow field characteristic prediction, this invention uses each sample in the dataset as the validation set and the remaining samples as the training set for training and validation. This process is repeated n times (n is the number of samples in the dataset). In this embodiment, since the sample set for each node is 10, the process is repeated 10 times (10 is the number of samples in the dataset). The coefficient of determination for all validation results of each flow velocity prediction model is calculated. The average value is then used as the final evaluation result for the performance of each model.

[0188] Coefficient of determination Its definition is as follows:

[0189]

[0190] It is the true output value of the i-th sample in the validation set;

[0191] It is the predicted output value of the i-th sample in the validation set;

[0192] m is the sample size of the validation set;

[0193] It is the average of the actual output values;

[0194] The closer the value is to 1, the higher the accuracy of the prediction model. The average accuracy of the leave-one-out cross-validation method for node models. All values ​​are above 0.95, meeting practical requirements;

[0195] The optimization of the real-time response rate of the flow field rendering cloud map of the cotton harvester's pneumatic conveying system in step S5 is as follows:

[0196] Step S5.1: Call the flow velocity data of multiple nodes in the pneumatic conveying system at different fan speeds in step S3, and retain the coordinate data corresponding to each node;

[0197] Step S5.2: Load the flow velocity data and coordinate data from step S5.1. Take the planar coordinate data as input and the flow velocity of the node corresponding to the coordinate data at a certain speed of the wind turbine as output. Construct a prediction model for different speeds based on polynomial chaotic expansion.

[0198] Step S5.3: Load all prediction models from Step S5.2. Based on the real-time response rate requirements of the cloud map, divide both the horizontal and vertical coordinates within the boundary of the cloud map into 20 equal parts (the boundary range is divided into 20 equal parts based on a trade-off between response rate and cloud map accuracy). Determine the horizontal and vertical coordinate data values ​​at the boundaries, and generate 441 uniform node coordinates by combining the pairwise horizontal and vertical coordinate values. Using the coordinates of all uniform nodes as input, load the prediction model to predict all data at each rotational speed.

[0199] Step S5.4: Load all the data from step S5.3, using the wind turbine speed as input and the flow velocity data corresponding to each node as output, to generate a flow velocity prediction model with 441 uniform nodes. Based on the input wind turbine speed, the rendering software calls all prediction models and calculates the node flow velocity value at each coordinate. The flow velocity data is then arranged onto the cloud map plane according to the corresponding coordinate order. Finally, the rendering software interpolates the data values ​​within the plane to generate the cloud map.

[0200] Figure 5The image shows the rendering of the flow field at the air outlet, revealing the details of the velocity in the flow field. The cloud map visually presents the uniformity of the flow field and the distribution of turbulent regions, facilitating the analysis of key system performance characteristics. The specific process corresponds to step S5 in this embodiment. Figure 6 The rendered contour plot of the flow field in the expanded planar region of the impeller ring reveals the details of the velocity distribution. The contour plot visually presents the uniformity of the flow field and the distribution of turbulent regions, facilitating the analysis of key system performance characteristics.

[0201] like Figure 7 This is a real-time line graph showing the fan speed, fan power, and minimum flow velocity in the center of the cotton harvester. Figure 8 A real-time line graph comparing the maximum and minimum flow velocities in the central flow field, the average flow velocity in the central flow field, and the real-time measured flow velocity. Figure 7 and Figure 8 It demonstrates the real-time line graph display effect of key performance data of the flow field, which can dynamically draw and reflect the fluctuation of core performance indicators during system operation, ensuring the stable operation of the cotton harvester's pneumatic conveying system.

[0202] This invention solves the problem of the inability to quickly predict and obtain the flow field characteristics of the pneumatic conveying system of cotton harvesters, and provides a reference for the control of the pneumatic conveying system.

[0203] Example 2

[0204] A system for implementing the flow field characteristic prediction method of the pneumatic conveying system of a cotton harvester based on digital twin as described in Example 1, thus having the beneficial effects of Example 1, will not be repeated here.

[0205] The system of the method for predicting the flow field characteristics of the pneumatic conveying system of a cotton harvester based on digital twins includes a simulation data acquisition module, a data preprocessing module, a core sample screening module, a node flow velocity prediction model construction module, a rendering cloud map generation module, and a visualization module.

[0206] The simulation data acquisition module is used to establish a high-fidelity numerical simulation model for predicting the flow field characteristics of the cotton harvester's pneumatic conveying system, and to divide the flow field at the air outlet into grid nodes. By using numerical simulation software such as computational fluid dynamics, finite element method, and finite volume method, the simulation data of the flow field performance of the cotton harvester's pneumatic conveying system is obtained based on the simulation model.

[0207] The data preprocessing module is used to preprocess the obtained flow field simulation data of the cotton harvester pneumatic conveying system, identify and remove outliers and anomalies in the data, and form training datasets for each node.

[0208] The core sample screening module is used to process the obtained training dataset. Considering the problems of too many data samples, cost and noise, the training data is clustered by fuzzy clustering algorithm, and the core data after screening is used as the training dataset.

[0209] The node velocity prediction model building module is used to establish a velocity prediction model for each node based on polynomial chaotic expansion. It uses the leave-one-out method to construct training and validation samples. The training dataset is used to build velocity prediction models for each node in the planar flow field at the cotton harvester outlet and the planar flow field of the impeller ring. The validation set is used to validate the velocity prediction models. The node velocity prediction model building module, based on the processed core dataset, uses polynomial chaotic regression to establish a velocity prediction model for each node, which is the key to the prediction principle of this digital twin system.

[0210] The rendering cloud map generation module is used to preprocess the obtained prediction model, optimize the number of prediction models, optimize the rendering cloud map generation rate, reduce rendering time, speed up the display speed of the cloud map, and call the prediction model to calculate and generate real-time rendering cloud map display based on the rendering software; the rendering cloud map generation module also reduces the number of prediction models to speed up the generation of cloud map data.

[0211] The visualization module is used to build a visualization platform based on digital twins. It uses the acquired prediction model as a foundation and establishes a visualization interface based on the software architecture. The flow field performance visualization module uses the software architecture to call the flow field characteristic prediction model to generate a visualization interface, enabling real-time monitoring of flow field performance.

[0212] It should be understood that although this specification is described according to various embodiments, not every embodiment contains only one independent technical solution. This way of describing the specification is only for clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

[0213] The detailed descriptions listed above are merely specific illustrations of feasible embodiments of the present invention and are not intended to limit the scope of protection of the present invention. All equivalent embodiments or modifications made without departing from the spirit of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for predicting the flow field characteristics of a pneumatic conveying system for a cotton harvester based on digital twins, characterized in that, Includes the following steps: Step S1: Acquisition of simulation data for the pneumatic conveying system of the cotton harvester: Establish a high-fidelity numerical simulation model for predicting the flow field characteristics of the pneumatic conveying system of the cotton harvester, and obtain the flow field simulation data of the pneumatic conveying system of the cotton harvester based on the simulation model. Step S2, Data Preprocessing: The flow field simulation data of the cotton harvester pneumatic conveying system obtained in Step S1 is preprocessed to identify and remove outliers and anomalies within the data, forming training datasets for each node. Step S3, Core Sample Screening: The training dataset obtained in Step S2 is processed by clustering the training data using a fuzzy clustering algorithm, and the core data after screening is used as the training dataset. Step S4: Node velocity prediction model construction: Based on polynomial chaotic expansion, establish velocity prediction models for each node, use leave-one-out method to construct training samples and validation samples, use the training dataset obtained in step S3 to construct velocity prediction models for each node of the plane flow field at the cotton harvester outlet and the plane flow field of the impeller ring expansion, and use the validation set to validate the velocity prediction models. Step S5, Rendering Cloud Map Generation: Filter the flow rate prediction models for each node obtained in Step S4, optimize the number of flow rate prediction models used, optimize the rendering cloud map generation rate, reduce rendering time, speed up the real-time display speed of the cloud map, and call the flow rate prediction model to calculate and generate a real-time rendered cloud map display based on the rendering software. Step S6, Visualization: Build a visualization module based on digital twins. Based on the flow rate prediction model used to generate the rendering cloud map in step S5, establish a visualization interface display based on the software architecture.

2. The method for predicting the flow field characteristics of a cotton harvester pneumatic conveying system based on digital twins according to claim 1, characterized in that, In step S1, obtaining simulation data of the flow field performance of the cotton harvester's pneumatic conveying system based on the simulation model specifically includes the following steps: Step S1.1: Establish a high-fidelity numerical simulation model for predicting the flow field characteristics of the pneumatic conveying system of the cotton harvester, based on the overall structure of the fan duct of the cotton harvester. Step S1.2: Perform mesh node division for the planar flow field at the air outlet of the cotton harvester and the planar flow field of the impeller ring; Step S1.3: Simulate and obtain the performance data of all grid nodes of the air outlet planar flow field and the impeller annular unfolded planar flow field under the corresponding fan operation data.

3. The method for predicting the flow field characteristics of a cotton harvester pneumatic conveying system based on digital twins according to claim 1, characterized in that, Step S1 calculates the flow velocity data of each node using the finite element method or the finite volume method to obtain the flow field simulation data of the cotton harvester pneumatic conveying system.

4. The method for predicting the flow field characteristics of a cotton harvester pneumatic conveying system based on digital twins according to claim 1, characterized in that, In step S2, isolated forests are used to identify and remove wild values.

5. The method for predicting the flow field characteristics of a cotton harvester pneumatic conveying system based on digital twins according to claim 1, characterized in that, In step S3, a fuzzy clustering algorithm is used to reduce the number of samples, and the core data after filtering is used as the training dataset and saved to the database.

6. The method for predicting the flow field characteristics of a cotton harvester pneumatic conveying system based on digital twins according to claim 1, characterized in that, In step S4, a flow velocity prediction model for each node is established based on polynomial chaotic expansion, which specifically includes the following steps: Step S4.1: Based on the processed dataset, generate training and validation sets using the leave-one-out method; Step S4.2: Establish a flow velocity prediction model using polynomial chaotic expansion; Step S4.3: Validate the performance of the flow velocity prediction model using the validation set and the coefficient of determination. To evaluate the accuracy and reliability of the flow rate prediction model.

7. The method for predicting the flow field characteristics of a cotton harvester pneumatic conveying system based on digital twins according to claim 6, characterized in that, The step S4.2, which uses polynomial chaotic expansion to establish a flow velocity prediction model, includes the following steps: Construct a basis function y based on the input and output and their relationship; Calculate the chaotic PC coefficients of an unknown polynomial; The relationship between input x and output y is expressed as follows: ； Where p is the maximum order of the polynomial chaotic expansion, and y is the output; x is the input; The tensor product of univariate orthogonal polynomials satisfies the following formula: ; Let x be the i-th variable in the input; The PC coefficients are unknown and yet to be determined. The total number of unknown PC coefficients P satisfies the following formula: ； P represents the maximum number of inputs; The dimension of the input variable; Let be a polynomial whose total order |k| does not exceed the given order k. To calculate the PC value, we need to find the polynomial chaotic expansion PCE that satisfies the following formula: Where N is the number of input-output samples; The above equation can be rewritten in matrix form: ； in, ; This refers to a real vector with P dimensions; To find the PC coefficient vector; ； y is the output response vector. Represents an N-dimensional real vector; ； This is a measurement matrix, where each column contains the evaluation values ​​of the PC basis for N samples; Represents an N A 3D real matrix; The PC coefficients are calculated by minimizing the residual between the model response and the PCE approximation; for The unknown coefficients are calculated using least squares regression, as shown in the following formula: ； when Then, the problem of solving for the PC coefficients is transformed into a problem of minimizing the constraint ℓ1: ； The ℓ2 norm constraint is used to consider the truncation error of the PCE. ; The above constraint ℓ1 minimization problem can be rewritten as a regularized optimization problem: ； Let be the ℓ1 norm of the PC coefficients; This represents the degree of fit between the model and the actual model response in the ℓ2 norm sense.

8. The method for predicting the flow field characteristics of a cotton harvester pneumatic conveying system based on digital twins according to claim 1, characterized in that, In step S5, optimizing the rendering cloud map generation rate includes the following steps: Step S5.1: Obtain flow field data of multiple nodes at different fan speeds and retain the coordinate data of each node. Load the flow field data and coordinate data, establish a proxy model based on polynomial chaotic expansion, take the coordinate data as input and the node wind speed at a certain speed as output, and construct a flow velocity prediction model at different speeds based on polynomial chaotic expansion. Step S5.2: Load all node velocity prediction models from step S5.

1. Based on the real-time response rate requirements of the cloud map, divide the horizontal and vertical coordinates into k equal parts within the boundary of the cloud map, where k is a positive integer. Determine the boundary of the division as the horizontal and vertical coordinate values, and combine them to generate (k+1)×(k+1) uniform node coordinate values. Using the coordinates of all uniform nodes as input, load the velocity prediction model and predict all data at each rotation speed. Step S5.3: Load all data of uniform nodes, take the speed of each wind turbine as input and the flow velocity data corresponding to each node as output, generate a flow velocity prediction model of (k+1)×(k+1) uniform nodes, call all flow velocity prediction models through rendering software according to the input wind turbine speed and calculate the node wind velocity value under each coordinate, and arrange the wind velocity data on the cloud map plane according to the corresponding coordinate order, and finally use rendering software to interpolate and generate cloud map according to the data numerical relationship in the plane.

9. The method for predicting the flow field characteristics of a cotton harvester pneumatic conveying system based on digital twins according to claim 1, characterized in that, The visualization interface in step S6 includes real-time display of the average wind speed in the central region of the flow field, the highest wind speed in the central region, the lowest wind speed in the central region, the outlet flow rate, the average operating power of the fan, as well as real-time display of the planar flow field at the outlet and the expanded planar flow field of the impeller ring.

10. A system for implementing the method for predicting the flow field characteristics of a pneumatic conveying system for a cotton harvester based on digital twins as described in any one of claims 1-9, characterized in that, It includes a simulation data acquisition module, a data preprocessing module, a core sample screening module, a node flow velocity prediction model construction module, a rendering cloud map generation module, and a visualization module; The simulation data acquisition module is used to establish a high-fidelity numerical simulation model for predicting the flow field characteristics of the pneumatic conveying system of the cotton harvester, and to obtain the flow field simulation data of the pneumatic conveying system of the cotton harvester based on the simulation model. The data preprocessing module is used to preprocess the obtained flow field simulation data of the cotton harvester pneumatic conveying system, identify and remove outliers and anomalies in the data, and form training datasets for each node. The core sample filtering module is used to process the obtained training dataset. It uses a fuzzy clustering algorithm to cluster the training data and uses the filtered core data as the training dataset. The node velocity prediction model building module is used to build velocity prediction models for each node based on polynomial chaotic expansion. It uses the leave-one-out method to construct training and validation samples, uses the training dataset to build velocity prediction models for each node in the plane flow field at the cotton harvester outlet and the plane flow field of the impeller ring expansion, and uses the validation set to validate the velocity prediction models. The rendering cloud map generation module is used to preprocess the obtained prediction model, optimize the number of flow rate prediction models, optimize the rendering cloud map generation rate, reduce rendering time, speed up the display speed of cloud map, and call the flow rate prediction model to calculate and generate real-time rendering cloud map display based on the rendering software. The visualization module is used to build a visualization platform based on digital twins. It is based on the flow rate prediction model and establishes a visualization interface display based on the software architecture.

Images