Engine bolt tightening torque control and adjustment method

Abstract

The invention relates to the technical field of bolt tightening torque measurement, and discloses an engine bolt tightening torque control and adjustment method, which comprises the following steps: step S101, collecting bolt parameters of M bolts; step S102, constructing graph structure data according to the bolt parameters of the M bolts; s103, constructing a training data set for training the tightening torque prediction model through an analog simulation platform; s104, inputting the graph structure data into the trained tightening torque prediction model, wherein the output value represents the target tightening torque of the M bolts; according to the method, the graph structure data is constructed according to the bolt parameters of the multiple bolts, and the nonlinear mapping relation between the graph structure data and the target tightening torque of the multiple bolts is established through the neural network model, so that the assembling quality of the engine bolts is improved.

Description

Technical Field

[0001] The invention relates to the technical field of bolt tightening torque measurement, and more specifically, to a method for controlling and adjusting the tightening torque of engine bolts. Background Art

[0002] In engine manufacturing, the tightening torque of bolts is directly related to the reliability and service life of the engine. The engine contains a large number of bolt connections, such as cylinder head bolts, crankcase bolts, etc. These connections need to withstand high temperature, high pressure and high vibration working conditions. Therefore, it is crucial to ensure that the bolt connections have sufficient clamping force.

[0003] The existing engine bolt assembly determines the bolt tightening torque through a neural network model, that is, the bolt parameters and engine connection parameters are collected, and a nonlinear mapping relationship between the bolt parameters, engine connection parameters and the bolt tightening torque is established through the neural network model. The training samples used to train the neural network model are obtained through a finite element analysis model, which comprehensively considers the effects of temperature changes and vibration changes during engine operation on the bolt tightening torque, thereby improving the quality of engine bolt assembly; however, the existing engine bolt assembly method only considers the quality of a single bolt assembly, and does not consider the effect of simultaneous assembly of multiple bolts on the engine operation stability, resulting in low quality of engine bolt assembly. Summary of the invention

[0004] The present invention provides a method for controlling and adjusting the tightening torque of engine bolts, which solves the technical problem that the prior art in the related art does not take into account the influence of simultaneous assembly of multiple bolts on the running stability of the engine, resulting in low quality of engine bolt assembly.

[0005] The present invention provides a method for controlling and adjusting the tightening torque of an engine bolt, comprising the following steps:

[0006] Step S101, collecting bolt parameters of M bolts;

[0007] Bolt parameters include: diameter, length, torque coefficient, thermal expansion coefficient, elastic modulus, pre-tightening torque and Poisson's ratio;

[0008] Where M is a custom parameter;

[0009] Step S102, constructing graph structure data according to the bolt parameters of the M bolts;

[0010] The graph structure data includes: M nodes, initial vectors of nodes, and edges between nodes;

[0011] M nodes establish data mapping with M bolts respectively;

[0012] The initial vector of the node is represented by the pre-processed bolt parameters of the bolt that establishes data mapping with the node;

[0013] The edges between nodes include: if the hole spacing between bolts is less than or equal to the distance threshold, then edges are built between nodes that establish data mapping with the bolts, where the distance threshold is a custom parameter;

[0014] Step S103, constructing a training data set for training a tightening torque prediction model through a simulation platform;

[0015] The input of the tightening torque prediction model is graph structure data;

[0016] The output of the tightening torque prediction model is the target tightening torque of M bolts;

[0017] The training data set includes U training samples, each of which includes a training data and a sample label, where U is a custom parameter;

[0018] The training data corresponds to graph structure data, and the sample labels correspond to the target tightening torques of M bolts;

[0019] Step S104: input the graph structure data into the trained tightening torque prediction model, and the output value represents the target tightening torque of the M bolts.

[0020] Furthermore, the pre-tightening torque F p The calculation formula is as follows:

[0021]

[0022] Where M represents the tightening torque of the bolt, f t Indicates the thread friction coefficient, f h It represents the friction coefficient between the contact surface of the bolt and the connecting part, and d represents the bolt diameter.

[0023] Further, the bolt parameters of the bolt are preprocessed to obtain an initial vector, including the following steps:

[0024] Step S201, directly fill in the missing values ​​of the bolt parameters with zero values;

[0025] Step S202, normalizing the bolt parameters by the minimum-maximum method to obtain an initial vector;

[0026]

[0027] Where 1≤p≤7, z p Represents the value of the pth dimension of the normalized bolt parameter, x p Represents the value of the pth dimension of the bolt parameter before normalization, xmin and x max Respectively represent the minimum and maximum values ​​of the bolt parameters.

[0028] Furthermore, the edge between nodes also includes: if the similarity value between the initial vectors of any two nodes is greater than or equal to a similarity threshold, an edge is constructed between the two nodes;

[0029] The similarity value Sim between the initial vectors of the i-th node and the j-th node i，j The calculation formula is as follows:

[0030]

[0031] Where 1≤i≤M, 1≤j≤M, Vec i and Vec j Respectively represent the initial vectors of the i-th node and the j-th node, Dis(Vec i , Vec j ) represents the Euclidean distance between the initial vectors of the i-th node and the j-th node.

[0032] Further, the tightening torque prediction model includes: M first hidden layers, 1 second hidden layer, 1 extraction layer and M classifiers;

[0033] The i-th first hidden layer inputs the initial vector of the i-th node and outputs the first update vector, where 1≤i≤M;

[0034] The M first hidden layers share weight parameters and bias parameters;

[0035] The second hidden layer inputs graph structure data and outputs a graph update matrix, where the graph update matrix includes M row vectors, each row vector corresponding to a second update vector of a node;

[0036] The extraction layer is used to input the second update vectors of the M nodes into the M classifiers, and the classification spaces of the M classifiers respectively represent the target tightening torques of the M bolts.

[0037] Furthermore, the calculation formula of the first hidden layer includes:

[0038] UpdteVec i =sigmoid(W2×(InitVec i T ×W1+UpMatrix i )+b);

[0039]

[0040] Where 1≤i≤M, 1≤u≤7, 1≤v≤7, UpdteVeci InitVec represents the first update vector of the output of the first hidden layer of the i-th layer. i Represents the initial vector of the i-th node of the i-th first hidden layer input, UpMatrix i Represents the upper triangular matrix transformed by the initial vector of the i-th node. The number of rows and columns of the upper triangular matrix is ​​the same as the number of dimensions of the initial vector. Represents the element value of the u-th row and v-th column of the upper triangular matrix transformed by the initial vector of the i-th node, represents the value of the u-th dimension of the initial vector of the i-th node, W1 and W2 represent the first weight parameter and the second weight parameter respectively, b represents the bias parameter, T represents the transpose operation, and sigmoid represents the sigmoid activation function.

[0041] Furthermore, the calculation formula of the second hidden layer includes:

[0042]

[0043] Where 1≤i≤M, 1≤j≤M, S represents the graph update matrix output by the second hidden layer, Represents the i-th row vector of the graph update matrix, that is, the second update vector of the i-th node, Col i represents the set of nodes that have edges connected to the i-th node, h i and h j Represent the first update vector of the i-th node and the j-th node respectively, A ij , B ij and W ij Respectively represent the correlation coefficient, the number of edges and the weight parameter between the i-th node and the j-th node, Init represents the number of dimensions of the first update vector of the node, and W i represents the weight parameter corresponding to the i-th node, e represents a natural constant, and Pile M It indicates that the second update vectors of M nodes are stacked, and sigmoid indicates the sigmoid activation function.

[0044] Furthermore, a training data set for training a tightening torque prediction model is constructed through a simulation platform, including the following steps:

[0045] Step S301, constructing a finite element analysis model of a bolt and an engine connection part through a simulation platform, and constructing graph structure data as training data of a training sample according to steps S101 to S102;

[0046] Step S302, randomly generating K tightening torques between a lower limit value of the tightening torque and an upper limit value of the tightening torque;

[0047] Among them, K, lower limit of tightening torque and upper limit of tightening torque are all custom parameters;

[0048] Step S303, respectively calculating the sum of the environmental loads of the M bolts at different ambient temperatures under each tightening torque;

[0049] Different ambient temperatures are custom parameters;

[0050] Step S304, respectively calculating the sum of the dynamic loads of the M bolts at each tightening torque at different engine speeds;

[0051] The different engine speeds are all custom parameters;

[0052] Step S305, respectively summing up the sum of the environmental loads of the M bolts at different ambient temperatures under each tightening torque and the sum of the dynamic loads of the bolts at different engine speeds under the corresponding tightening torque to obtain a comprehensive score;

[0053] Step S306, taking the M tightening torques with the smallest comprehensive scores as sample labels of a training sample;

[0054] Step S307, repeat steps S302 to S306 until U training samples are obtained as the training data set.

[0055] Furthermore, the environmental load Load Env The calculation formula is as follows:

[0056] Load Env =F p -(E×Area×α×T);

[0057] Among them, F p , E, Area and α represent the pre-tightening torque, stiffness, cross-sectional area and thermal expansion coefficient of the bolt, respectively, and T represents the ambient temperature.

[0058] Furthermore, dynamic load dyn The calculation formula is as follows:

[0059]

[0060] Among them, F p , m and MaxSwing represent the pre-tightening torque, weight and maximum amplitude of the bolt respectively, n represents the engine speed, and β represents the proportional coefficient ranging from 0 to 1.

[0061] The beneficial effect of the present invention is that the present invention constructs graph structure data according to the bolt parameters of multiple bolts, and establishes a nonlinear mapping relationship between the graph structure data and the target tightening torque of multiple bolts through a neural network model, thereby improving the quality of engine bolt assembly. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a flow chart of a method for controlling and adjusting the tightening torque of an engine bolt of the present invention;

[0063] Figure 2 It is a flow chart of preprocessing the bolt parameters of the bolt to obtain the initial vector of the present invention;

[0064] Figure 3 It is a flow chart of constructing a training data set for training a tightening torque prediction model through a simulation platform of the present invention. DETAILED DESCRIPTION

[0065] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that the discussion of these embodiments is only to enable those skilled in the art to better understand and implement the subject matter described herein, and the functions and arrangements of the elements discussed may be changed without departing from the scope of protection of the contents of this specification. Each example may omit, replace or add various processes or components as needed. In addition, the features described relative to some examples may also be combined in other examples.

[0066] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in one or more embodiments of the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in one or more embodiments of the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0067] like Figure 1 to Figure 3 As shown, a method for controlling and adjusting the tightening torque of engine bolts comprises the following steps:

[0068] Step S101, collecting bolt parameters of M bolts;

[0069] Bolt parameters include: diameter, length, torque coefficient, thermal expansion coefficient, elastic modulus, pre-tightening torque and Poisson's ratio;

[0070] Where M is a custom parameter, which is set according to the number of bolts actually required by the engine;

[0071] Step S102, constructing graph structure data according to the bolt parameters of the M bolts;

[0072] The graph structure data includes: M nodes, initial vectors of nodes, and edges between nodes;

[0073] M nodes establish data mapping with M bolts respectively;

[0074] The initial vector of the node is represented by the pre-processed bolt parameters of the bolt that establishes data mapping with the node;

[0075] The edges between nodes include: if the hole spacing between bolts is less than or equal to the distance threshold, then edges are built between nodes that establish data mapping with the bolts, where the distance threshold is a custom parameter;

[0076] Step S103, constructing a training data set for training a tightening torque prediction model through a simulation platform;

[0077] The input of the tightening torque prediction model is graph structure data;

[0078] The output of the tightening torque prediction model is the target tightening torque of M bolts;

[0079] The training data set includes U training samples, each training sample includes a training data and a sample label, where U is a custom parameter, preferably, U is set to 2000;

[0080] The training data corresponds to graph structure data, and the sample labels correspond to the target tightening torques of M bolts;

[0081] Step S104: input the graph structure data into the trained tightening torque prediction model, and the output value represents the target tightening torque of the M bolts.

[0082] It should be noted that the diameter, length, torque coefficient, thermal expansion coefficient, elastic modulus and Poisson's ratio of the bolt are all provided by the bolt manufacturer. The torque coefficient indicates the friction coefficient between the bolt and the engine connection, the thermal expansion coefficient indicates the degree of expansion or contraction of the bolt material with temperature changes, the elastic modulus indicates the ability to resist stretching or compression during the elastic deformation stage, and the Poisson's ratio indicates the contraction ratio in the vertical direction when the bolt material is stretched in the direction of force.

[0083] In one embodiment of the present invention, the pre-tightening torque F pThe calculation formula is as follows:

[0084]

[0085] Where M represents the tightening torque of the bolt, f t Indicates the thread friction coefficient, f h It represents the friction coefficient between the contact surface of the bolt and the connecting part, and d represents the bolt diameter.

[0086] It should be noted that the tightening torque of the bolt is obtained through an electronic torque tool, and the thread friction coefficient and the friction coefficient between the bolt and the contact surface of the connecting part are provided by the bolt manufacturer and will not be elaborated here.

[0087] In one embodiment of the present invention, Figure 2 As shown, the bolt parameters of the bolt are preprocessed to obtain the initial vector, including the following steps:

[0088] Step S201, directly fill in the missing values ​​of the bolt parameters with zero values;

[0089] Step S202, normalizing the bolt parameters by the minimum-maximum method to obtain an initial vector;

[0090]

[0091] Where 1≤p≤7, z p Represents the value of the pth dimension of the normalized bolt parameter, x p Represents the value of the pth dimension of the bolt parameter before normalization, x min and x max Respectively represent the minimum and maximum values ​​of the bolt parameters.

[0092] In one embodiment of the present invention, the edge between nodes further includes: if the similarity value between the initial vectors of any two nodes is greater than or equal to a similarity threshold, then an edge is constructed between the two nodes;

[0093] The similarity value Sim between the initial vectors of the i-th node and the j-th node i，j The calculation formula is as follows:

[0094]

[0095] Where 1≤i≤M, 1≤j≤M, Vec i and Vec j Respectively represent the initial vectors of the i-th node and the j-th node, Dis(Vec i , Vec j ) represents the Euclidean distance between the initial vectors of the i-th node and the j-th node.

[0096] In one embodiment of the present invention, the tightening torque prediction model includes: M first hidden layers, 1 second hidden layer, 1 extraction layer and M classifiers;

[0097] The i-th first hidden layer inputs the initial vector of the i-th node and outputs the first update vector, where 1≤i≤M;

[0098] The M first hidden layers share weight parameters and bias parameters;

[0099] The second hidden layer inputs graph structure data and outputs a graph update matrix, where the graph update matrix includes M row vectors, each row vector corresponding to a second update vector of a node;

[0100] The extraction layer is used to input the second update vectors of the M nodes into the M classifiers, and the classification spaces of the M classifiers respectively represent the target tightening torques of the M bolts.

[0101] In one embodiment of the present invention, the calculation formula of the first hidden layer includes:

[0102] UpdteVec i =sigmoid(W2×(InitVec i T ×W1+UpMatrix i )+b);

[0103]

[0104] Where 1≤i≤M, 1≤u≤7, 1≤v≤7, UpdteVec i InitVec represents the first update vector of the output of the first hidden layer of the i-th layer. i Represents the initial vector of the i-th node of the i-th first hidden layer input, UpMatrix i Represents the upper triangular matrix transformed by the initial vector of the i-th node. The number of rows and columns of the upper triangular matrix is ​​the same as the number of dimensions of the initial vector. Represents the element value of the u-th row and v-th column of the upper triangular matrix transformed by the initial vector of the i-th node, represents the value of the u-th dimension of the initial vector of the i-th node, W1 and W2 represent the first weight parameter and the second weight parameter respectively, b represents the bias parameter, T represents the transpose operation, and sigmoid represents the sigmoid activation function.

[0105] It should be noted that the size of the node's initial vector is 1×7, and the size of the transposed initial vector is 7×1. The first weight parameter can be designed as a vector of size 1×7. The transposed initial vector is multiplied by the first weight parameter to obtain a 7×7 matrix, which is added to the upper triangular matrix to obtain a 7×7 matrix. The second weight parameter is designed as a vector of size 1×7. Then the first update vector obtained by multiplying the second weight parameter by the 7×7 matrix is ​​a 1×7 vector.

[0106] In one embodiment of the present invention, the calculation formula of the second hidden layer includes:

[0107]

[0108]

[0109] Where 1≤i≤M, 1≤j≤M, S represents the graph update matrix output by the second hidden layer, Represents the i-th row vector of the graph update matrix, that is, the second update vector of the i-th node, Col i represents the set of nodes that have edges connected to the i-th node, h i and h j Represent the first update vector of the i-th node and the j-th node respectively, A ij , B ij and W ij Respectively represent the association coefficient, the number of edges and the weight parameter between the i-th node and the j-th node, Init represents the number of dimensions of the first update vector of the node, and W i represents the weight parameter corresponding to the i-th node, e represents a natural constant, and Pile M It indicates that the second update vectors of M nodes are stacked, and sigmoid indicates the sigmoid activation function.

[0110] In one embodiment of the present invention, Figure 3 As shown, a training data set for training a tightening torque prediction model is constructed through a simulation platform, including the following steps:

[0111] Step S301, constructing a finite element analysis model of a bolt and an engine connection part through a simulation platform, and constructing graph structure data as training data of a training sample according to steps S101 to S102;

[0112] Step S302, randomly generating K tightening torques between a lower limit value of the tightening torque and an upper limit value of the tightening torque;

[0113] Wherein K, the lower limit value of the tightening torque and the upper limit value of the tightening torque are all custom parameters. Preferably, K is set to 5, and the lower limit value of the tightening torque and the upper limit value of the tightening torque can be set according to the standards formulated by the country or the industry;

[0114] Step S303, respectively calculating the sum of the environmental loads of the M bolts at different ambient temperatures under each tightening torque;

[0115] The different ambient temperatures are all custom parameters;

[0116] For example, the ambient temperature can be set to 6 groups, that is, the upper and lower temperature limits that the engine can bear are divided into equal temperature intervals. For example, the upper and lower temperature limits that the engine can bear are 100°C and 0°C respectively, then the 6 groups of ambient temperatures are 0°C, 20°C, 40°C, 60°C, 80°C and 100°C respectively;

[0117] Step S304, respectively calculating the sum of the dynamic loads of the M bolts at each tightening torque at different engine speeds;

[0118] The different engine speeds are all custom parameters;

[0119] For example, the engine speed can be set to 6 groups, that is, the maximum speed and the minimum speed of the engine are divided into equal speed intervals. For example, the maximum speed and the minimum speed of the engine are 3000RPM and 0RPM respectively, then the 6 groups of engine speeds are 0RPM, 600RPM, 1200RPM, 1800RPM, 2400RPM and 3000RPM respectively;

[0120] Step S305, respectively summing up the sum of the environmental loads of the M bolts at different ambient temperatures under each tightening torque and the sum of the dynamic loads of the bolts at different engine speeds under the corresponding tightening torque to obtain a comprehensive score;

[0121] Step S306, taking the M tightening torques with the smallest comprehensive scores as sample labels of a training sample;

[0122] Step S307, repeat steps S302 to S306 until U training samples are obtained as the training data set.

[0123] In one embodiment of the present invention, the environmental load Load Env The calculation formula is as follows:

[0124] Load Env =F p -(E×Area×α×T);

[0125] where F p, E, Area and α represent the pre-tightening torque, stiffness, cross-sectional area and thermal expansion coefficient of the bolt, respectively, and T represents the ambient temperature.

[0126] In one embodiment of the present invention, the dynamic load Load dyn The calculation formula is as follows:

[0127]

[0128] where F p , m and MaxSwing represent the pre-tightening torque, weight and maximum amplitude of the bolt respectively, n represents the engine speed, and β represents the proportional coefficient ranging from 0 to 1.

[0129] The above describes an embodiment of the present embodiment, but the present embodiment is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present embodiment, ordinary technicians in this field can also make many forms, all of which are within the protection of the present embodiment.

Claims

1. A method for controlling and adjusting the tightening torque of engine bolts, characterized in that: The following steps are involved: Step S101, collecting bolt parameters of M bolts; Bolt parameters include: diameter, length, torque coefficient, thermal expansion coefficient, elastic modulus, pre-tightening torque and Poisson's ratio; Where M is a custom parameter; Step S102, constructing graph structure data according to the bolt parameters of the M bolts; The graph structure data includes: M nodes, initial vectors of nodes, and edges between nodes; M nodes establish data mapping with M bolts respectively; The initial vector of the node is represented by the pre-processed bolt parameters of the bolt that establishes data mapping with the node; The edges between nodes include: if the hole spacing between bolts is less than or equal to the distance threshold, then edges are built between nodes that establish data mapping with the bolts, where the distance threshold is a custom parameter; Step S103, constructing a training data set for training a tightening torque prediction model through a simulation platform; The input of the tightening torque prediction model is graph structure data; The output of the tightening torque prediction model is the target tightening torque of M bolts; The training data set includes U training samples, each of which includes a training data and a sample label, where U is a custom parameter; The training data corresponds to graph structure data, and the sample labels correspond to the target tightening torques of M bolts; Step S104: input the graph structure data into the trained tightening torque prediction model, and the output value represents the target tightening torque of the M bolts.

2. The engine bolt tightening torque control and adjustment method according to claim 1 is characterized in that: Pre-tightening torque F p The calculation formula is as follows: Where M represents the tightening torque of the bolt, f t Indicates the thread friction coefficient, f h It represents the friction coefficient between the contact surface of the bolt and the connecting part, and d represents the bolt diameter.

3. The engine bolt tightening torque control and adjustment method according to claim 1, characterized in that: Preprocessing the bolt parameters of the bolt to obtain the initial vector includes the following steps: Step S201, directly fill in the missing values ​​of the bolt parameters with zero values; Step S202, normalizing the bolt parameters by the minimum-maximum method to obtain an initial vector; where 1≤p≤7, z p Represents the value of the pth dimension of the normalized bolt parameter, x p Represents the value of the pth dimension of the bolt parameter before normalization, x min and x max Respectively represent the minimum and maximum values ​​of the bolt parameters.

4. The engine bolt tightening torque control and adjustment method according to claim 1, characterized in that: The edge between nodes also includes: if the similarity value between the initial vectors of any two nodes is greater than or equal to the similarity threshold, an edge is constructed between the two nodes; The similarity value Sim between the initial vectors of the i-th node and the j-th node i，j The calculation formula is as follows: Where 1≤i≤M, 1≤j≤M, Vec i and Vec j Respectively represent the initial vectors of the i-th node and the j-th node, Dis(Vec i , Vec j ) represents the Euclidean distance between the initial vectors of the i-th node and the j-th node.

5. The engine bolt tightening torque control and adjustment method according to claim 1, characterized in that: The tightening torque prediction model includes: M first hidden layers, 1 second hidden layer, 1 extraction layer and M classifiers; The i-th first hidden layer inputs the initial vector of the i-th node and outputs the first update vector, where 1≤i≤M; The M first hidden layers share weight parameters and bias parameters; The second hidden layer inputs graph structure data and outputs a graph update matrix, where the graph update matrix includes M row vectors, each row vector corresponding to a second update vector of a node; The extraction layer is used to input the second update vectors of the M nodes into the M classifiers, and the classification spaces of the M classifiers respectively represent the target tightening torques of the M bolts.

6. The engine bolt tightening torque control and adjustment method according to claim 5, characterized in that: The calculation formula for the first hidden layer includes: UpdteVec i =sigmoid(W2×(InitVec i T ×W1+UpMatrix i )+b); Where 1≤i≤M, 1≤u≤7, 1≤v≤7, UpdteVec i InitVec represents the first update vector of the output of the first hidden layer of the i-th layer. i Represents the initial vector of the i-th node of the i-th first hidden layer input, UpMatrix i Represents the upper triangular matrix transformed by the initial vector of the i-th node. The number of rows and columns of the upper triangular matrix is ​​the same as the number of dimensions of the initial vector. Represents the element value of the u-th row and v-th column of the upper triangular matrix transformed by the initial vector of the i-th node, represents the value of the u-th dimension of the initial vector of the i-th node, W1 and W2 represent the first weight parameter and the second weight parameter respectively, b represents the bias parameter, T represents the transpose operation, and sigmoid represents the sigmoid activation function.

7. The engine bolt tightening torque control and adjustment method according to claim 5, characterized in that: The calculation formula for the second hidden layer includes: Where 1≤i≤M, 1≤j≤M, S represents the graph update matrix output by the second hidden layer, Represents the i-th row vector of the graph update matrix, that is, the second update vector of the i-th node, Col i represents the set of nodes that have edges connected to the i-th node, h i and h j Represent the first update vector of the i-th node and the j-th node respectively, A ij , B ij and W ij Respectively represent the association coefficient, the number of edges and the weight parameter between the i-th node and the j-th node, Init represents the number of dimensions of the first update vector of the node, and W i represents the weight parameter corresponding to the i-th node, e represents a natural constant, and Pile M It indicates that the second update vectors of M nodes are stacked, and sigmoid indicates the sigmoid activation function.

8. The engine bolt tightening torque control and adjustment method according to claim 1, characterized in that: The training data set for training the tightening torque prediction model is constructed through the simulation platform, including the following steps: Step S301, constructing a finite element analysis model of a bolt and an engine connection part through a simulation platform, and constructing graph structure data as training data of a training sample according to steps S101 to S102; Step S302, randomly generating K tightening torques between a lower limit value of the tightening torque and an upper limit value of the tightening torque; Among them, K, lower limit of tightening torque and upper limit of tightening torque are all custom parameters; Step S303, respectively calculating the sum of the environmental loads of the M bolts at different ambient temperatures under each tightening torque; The different ambient temperatures are all custom parameters; Step S304, respectively calculating the sum of the dynamic loads of the M bolts at each tightening torque at different engine speeds; The different engine speeds are all custom parameters; Step S305, respectively summing up the sum of the environmental loads of the M bolts at different ambient temperatures under each tightening torque and the sum of the dynamic loads of the bolts at different engine speeds under the corresponding tightening torque to obtain a comprehensive score; Step S306, taking the M tightening torques with the smallest comprehensive scores as sample labels of a training sample; Step S307, repeat steps S302 to S306 until U training samples are obtained as the training data set.

9. The engine bolt tightening torque control and adjustment method according to claim 8, characterized in that: Environmental Load Env The calculation formula is as follows: Load Env =F p -(E×Area×α×T); where F p , E, Area and α represent the pre-tightening torque, stiffness, cross-sectional area and thermal expansion coefficient of the bolt, respectively, and T represents the ambient temperature.

10. The engine bolt tightening torque control and adjustment method according to claim 8, characterized in that: Dynamic Load dyn The calculation formula is as follows: where F p , m and MaxSwing represent the pre-tightening torque, weight and maximum amplitude of the bolt respectively, n represents the engine speed, and β represents the proportional coefficient ranging from 0 to 1.

Images