Thick plate impact energy prediction reliability evaluation method and device and electronic equipment

Abstract

The embodiment of the invention provides a thick plate impact energy prediction reliability evaluation method and device and electronic equipment, and relates to the field of data processing, and the method comprises the steps: determining a first direct trust value feature and a first recommended trust value feature value between a to-be-predicted sample and each historical sample in each historical data, based on each first direct trust value feature and each corresponding first recommendation trust value feature value, determining each first global trust degree value of the to-be-predicted sample and each historical data, based on each first global trust degree value, obtaining a first global trust degree matrix, and aiming at each one-dimensional trusted vector in the first global trust degree matrix, obtaining a first trust degree matrix; obtaining a plurality of first sub-vectors based on a sliding window, calculating the mean values of the plurality of first sub-vectors, and inputting each mean value into the trained impact energy prediction reliability evaluation optimization model to obtain the impact energy variance error of the to-be-predicted sample, thereby achieving the evaluation of the predicted impact energy. And the production process standard control of the subsequent process can be effectively guided on site.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and more specifically, to a method, apparatus, and electronic device for predicting the reliability of impact energy of thick plates. Background Technology

[0002] At present, the research on the prediction of impact energy of thick plates has become a hot topic in the industry. Accurate prediction of impact energy can not only help the production site to carry out quality early warning, but also reduce the complex and delayed sampling impact test process, greatly reduce the need for experimental testing, and reduce the occurrence of quality disputes.

[0003] Traditional research on impact performance prediction typically employs probabilistic statistics and convolutional neural network (RNN) algorithms for estimation. More in-depth studies utilize radial basis function (RBF) neural networks to avoid the problems of adjusting neural network weight parameters and inaccurate determination of hidden layer centers that exist in traditional neural network algorithms used for function approximation or regression. The selection of factors influencing the basis function of neural networks and the adjustment of weight parameters are mainly addressed through path optimization algorithms. For example, the paper "A Method for Predicting the Mechanical Properties of Steel Plates" mentions using attention mechanisms to optimize convolutional neural networks (RNNs) for predicting the mechanical properties of steel plates. However, it does not provide effective solutions for extracting features of performance influencing factors, adapting the algorithm to different scenarios, or effectively addressing the inherent limitations of the algorithm in handling long-term dependencies and avoiding gradient vanishing problems. These issues in the aforementioned literature may lead to unreliable prediction results. Summary of the Invention

[0004] The purpose of this invention is to provide a method, apparatus and electronic equipment for predicting the reliability of impact energy in thick plates, which can determine the accuracy of the predicted impact energy, thereby helping to effectively guide the production process standard control of subsequent processes on site.

[0005] To achieve the above objectives, the technical solutions adopted in the embodiments of this application are as follows: In a first aspect, embodiments of this application provide a method for predicting the impact energy of a thick plate, the method comprising: Determine the first direct trust value feature and the first recommendation trust value feature value between the sample to be predicted and each historical sample in each historical data; Based on each of the first direct trust value features and the corresponding first recommended trust value features, the first global trust value of the sample to be predicted and each of the historical data is determined. Based on each first global trust value, the first global trust matrix is ​​obtained; For each one-dimensional trustworthy vector in the first global trust matrix, multiple first sub-vectors are obtained based on a sliding window, wherein the first global trust matrix includes multiple one-dimensional trustworthy vectors; Calculate the mean of multiple first sub-vectors; The mean values ​​are input into the trained impact energy prediction reliability evaluation optimization model to obtain the impact energy variance error of the sample to be predicted.

[0006] In an optional implementation, the method further includes: Determine the second direct trust value feature and the second recommended trust value feature value among the historical samples in the historical data; Based on the second direct trust value feature and the second recommended trust value feature, a second global trust value is determined among the historical samples; Construct a second global trust matrix based on each of the second global trust values; A sliding window mechanism is used to extract multiple second sub-vectors for each dimension of the trustworthy vector in the second global trust matrix; Calculate the mean of multiple second sub-vectors; The average algorithm variance between the true impact energy label value and the predicted impact energy label value of the historical samples is calculated and used as the impact energy variance error label. The impact energy prediction reliability assessment optimization model is trained based on the mean and the corresponding impact energy variance error label to obtain the trained impact energy prediction reliability assessment optimization model.

[0007] In an optional implementation, the step of determining the second direct trust value feature among historical samples in the historical data includes: For the historical sample i, determine the risk coefficient of the historical sample i; Based on the risk coefficient, a second direct trust value feature is determined among the historical samples in the historical data.

[0008] In an optional implementation, the step of determining the risk coefficient of the historical sample i includes: Determine the centroid values ​​of multiple attributes contained in the historical sample i; Calculate the Eulerian distance between each attribute in the historical sample i and the centroid value; The ratio of the Eulerian distance to the distance threshold is calculated and used as the risk coefficient of the attribute.

[0009] In an optional implementation, the step of determining the second direct trust value feature among the historical samples in the historical data based on the risk coefficient includes: Determine the first set of attributes for historical sample i; Determine the set of second attributes of historical sample j; Based on the first attribute set and the second attribute set, the similarity values ​​of each attribute curve of the historical sample i and the historical sample j are determined; Based on the similarity value and each of the risk coefficients, a second direct trust value feature is determined among the historical samples in the historical data.

[0010] In an optional implementation, the second direct trust value feature is calculated using the following formula: ; in, Let be the second direct trust value feature between historical sample i and historical sample j, m be the total number of attributes between historical sample i and historical sample j, S(im,jm) be the similarity value of each attribute curve between historical sample i and historical sample j, and (i,m) represent the risk coefficient of each attribute in historical sample i.

[0011] In an optional implementation, the second recommendation trust value feature is calculated using the following formula: ; Where R(im,jm) is the second recommendation trust value between historical sample i and historical sample j. This represents the ratio of the number of recommended nodes M with similarity values ​​greater than a preset threshold to the total number of attributes. Recommended nodes M are obtained from the historical samples i and j, where the node M simultaneously satisfies that the average similarity value of its attributes with nodes i and j is greater than a set threshold.

[0012] In an optional implementation, the second global trust value is calculated using the following formula: ; in Let be the global trust value between historical sample i and historical sample j, 0 1. R(im,jm) is the second recommendation trust value between historical sample i and historical sample j. The second direct trust value feature is the historical sample i and historical sample j.

[0013] Secondly, embodiments of this application provide a reliability assessment device for predicting the impact energy of thick plates, the device comprising: The determination module is used to determine the first direct trust value feature and the first recommended trust value feature value between the sample to be predicted and each historical sample in each historical data; based on each first direct trust value feature and the corresponding first recommended trust value feature value, determine each first global trust value between the sample to be predicted and each of the historical data; and based on each first global trust value, obtain the first global trust matrix. The acquisition module is used to acquire multiple first sub-vectors based on a sliding window for each one-dimensional trustworthy vector in the first global trust matrix, wherein the first global trust matrix includes multiple one-dimensional trustworthy vectors. A calculation module is used to calculate the mean of multiple first sub-vectors; The prediction module is used to input the mean values ​​into the trained impact energy prediction reliability evaluation optimization model to obtain the impact energy variance error of the sample to be predicted.

[0014] Thirdly, embodiments of this application provide an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the thick plate impact energy prediction reliability assessment method.

[0015] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the thick plate impact energy prediction reliability assessment method.

[0016] This application has the following beneficial effects: This application determines the first direct trust value feature and the first recommended trust value feature value between the sample to be predicted and each historical sample in each historical data. Based on each first direct trust value feature and the corresponding first recommended trust value feature value, it determines each first global trust value between the sample to be predicted and each historical data. Based on each first global trust value, it obtains a first global trust matrix. For each one-dimensional trustworthy vector in the first global trust matrix, it obtains multiple first sub-vectors based on a sliding window. The first global trust matrix includes multiple one-dimensional trustworthy vectors. It calculates the mean of multiple first sub-vectors and inputs each mean into a trained impact energy prediction reliability assessment optimization model to obtain the impact energy variance error of the sample to be predicted. This enables the assessment of the predicted impact energy and helps to effectively guide the production process standard control of subsequent processes on-site. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A block diagram of an electronic device provided in an embodiment of the present invention; Figure 2This is one of the flowcharts illustrating a reliability assessment method for predicting the impact energy of a thick plate, provided by an embodiment of the present invention. Figure 3 This is a second schematic flowchart of a reliability assessment method for predicting the impact energy of a thick plate, provided in an embodiment of the present invention. Figure 4 The third flowchart illustrates a reliability assessment method for predicting the impact energy of a thick plate, as provided in an embodiment of the present invention. Figure 5 The fourth flowchart illustrates a reliability assessment method for predicting the impact energy of a thick plate, as provided in an embodiment of the present invention. Figure 6 The fifth flowchart illustrates a reliability assessment method for predicting the impact energy of a thick plate, as provided in an embodiment of the present invention. Figure 7 This is a structural block diagram of a thick plate impact energy prediction reliability assessment device provided in an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0020] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0021] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0022] In the description of this invention, it should be noted that if terms such as "upper," "lower," "inner," or "outer" are used to indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship in which the product of this invention is usually placed, they are only for the convenience of describing this 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 this invention.

[0023] Furthermore, the terms "first" and "second" are used only to distinguish descriptions and should not be interpreted as indicating or implying relative importance.

[0024] In the description of this application, it should also be noted that, unless otherwise expressly specified and limited, the terms "set up," "install," "connect," and "link" 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 application based on the specific circumstances.

[0025] Through extensive research, the inventors discovered that traditional studies on the prediction of impact performance typically employ probability statistics and convolutional neural network (RNN) algorithms for estimation. More in-depth research utilizes radial basis function neural networks to avoid the problems of adjusting neural network weight parameters and inaccurate determination of hidden layer centers that exist in traditional neural network algorithms used for function approximation or regression.

[0026] No effective solutions have been provided for extracting features of performance-influencing factors and adapting the algorithm to different scenarios. Furthermore, there are no effective solutions for the algorithm's inability to handle long-term dependencies and avoid the gradient vanishing problem. The problems in the aforementioned literature may lead to unreliable prediction results.

[0027] In view of the above-mentioned problems, this embodiment provides a method, apparatus, and electronic device for predicting the reliability of impact energy of thick plates. It can determine the first direct trust value feature and the first recommended trust value feature value between the sample to be predicted and each historical sample in the historical data. Based on each first direct trust value feature and the corresponding first recommended trust value feature value, it determines each first global trust value between the sample to be predicted and each historical data. Based on each first global trust value, it obtains a first global trust matrix. For each one-dimensional trustworthy vector in the first global trust matrix, it obtains multiple first sub-vectors based on a sliding window. The first global trust matrix includes multiple one-dimensional trustworthy vectors. It calculates the mean of the multiple first sub-vectors and inputs each mean into a trained impact energy prediction reliability assessment optimization model to obtain the impact energy variance error of the sample to be predicted. This enables the assessment of the predicted impact energy, thereby helping to effectively guide the production process standard control of subsequent processes on-site. The solution provided in this embodiment is described in detail below.

[0028] This embodiment provides an electronic device capable of evaluating the reliability of thick plate impact energy prediction. In one possible implementation, the electronic device can be a user terminal, such as, but not limited to, a server, smartphone, personal computer (PC), tablet computer, personal digital assistant (PDA), mobile internet device (MID), etc.

[0029] Please refer to Figure 1 , Figure 1 This is a schematic diagram of the structure of the electronic device 100 provided in the embodiments of this application. The electronic device 100 may further include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown. Figure 1 The components shown can be implemented using hardware, software, or a combination thereof.

[0030] The electronic device 100 includes a thick plate impact energy prediction reliability assessment device 110, a memory 120, and a processor 130.

[0031] The components of the memory 120 and processor 130 are electrically connected directly or indirectly to achieve data transmission or interaction. For example, these components can be electrically connected to each other through one or more communication buses or signal lines. The thick plate impact energy prediction reliability assessment device 110 includes at least one software function module that can be stored in the memory 120 in the form of software or firmware or embedded in the operating system (OS) of the electronic device 100. The processor 130 is used to execute the executable modules stored in the memory 120, such as the software function modules and computer programs included in the thick plate impact energy prediction reliability assessment device 110.

[0032] The memory 120 may be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), etc. The memory 120 is used to store programs, and the processor 130 executes the programs after receiving execution instructions.

[0033] Please refer to Figure 2 , Figure 2 For application Figure 1 The flowchart below shows a method for predicting the reliability of a thick plate impact energy of an electronic device 100. The method includes a detailed explanation of each step.

[0034] S201: Determine the first direct trust value feature and the first recommended trust value feature between the sample to be predicted and each historical sample in each historical data.

[0035] S202: Based on the first direct trust value features and the corresponding first recommended trust value features, determine the first global trust value of the sample to be predicted and each historical data.

[0036] S203: Based on each first global trust value, obtain the first global trust matrix.

[0037] S204: For each one-dimensional trustworthy vector in the first global trust matrix, obtain multiple first sub-vectors based on a sliding window.

[0038] The first global trust matrix includes multiple one-dimensional trust vectors.

[0039] S205: Calculate the mean of multiple first subvectors.

[0040] S206: Input the mean values ​​into the trained impact energy prediction reliability evaluation optimization model to obtain the impact energy variance error of the sample to be predicted.

[0041] The first step involves collecting data from the sample to be predicted. This data is collected from the production system, specifically from thick plate production data, primarily including relevant influencing factor parameter values. The second step is data processing. This involves merging the sample to be predicted with historical data, cleaning the merged data to remove invalid and abnormal data, normalizing the influencing factor data in the cleaned data, and then calculating the first direct confidence value feature and the first recommended confidence value feature value between the sample to be predicted and each historical sample.

[0042] Based on the first direct trust value features and the corresponding first recommended trust value features, the first global trust value of the sample to be predicted and each historical data is determined. Based on each first global trust value, the first global trust matrix is ​​obtained. For each one-dimensional trust vector in the first global trust matrix, multiple first sub-vectors are obtained based on a sliding window. The first global trust matrix includes multiple one-dimensional trust vectors. The mean of the multiple first sub-vectors is calculated and used as the input of the trained impact energy prediction reliability assessment optimization model, so that the sample to be predicted is consistent with the sample of the impact energy prediction model.

[0043] This invention utilizes a trained impact energy prediction reliability assessment optimization model to estimate the uncertainty index of the impact energy prediction of the sample to be predicted. This model avoids misleading production due to abnormal prediction results by measuring the reliability of the prediction model, shortens the material flow cycle, and reduces costs. By adopting the solution of this invention, a reliability assessment index for prediction results can be provided, reducing the occurrence of products with quality disputes caused by performance issues, and providing guidance for on-site sampling methods.

[0044] To address the reliability issues of existing thick plate impact prediction models, such as unstable prediction accuracy, especially when introduced into new scenarios, the prediction model often fails to meet customer requirements for product quality, and operators cannot promptly and accurately assess the model's suitability and reliability. The impact energy prediction reliability assessment and optimization model can provide a timely, third-party perspective on the reliability of traditional prediction model results, offering a solution to the uncertainty of traditional performance prediction model results.

[0045] When predicting impact energy, it is necessary to optimize the impact energy prediction reliability assessment model based on a well-trained model.

[0046] like Figure 3 As shown, the method for training the impact energy prediction reliability assessment optimization model includes the following steps: S301: Determine the second direct trust value feature and the second recommended trust value feature value among the historical samples in the historical data.

[0047] S302: Determine the second global trust value among historical samples based on the second direct trust value feature and the second recommended trust value feature.

[0048] S303: Construct a second global trust matrix based on each second global trust value.

[0049] S304: Use a sliding window mechanism to extract multiple second sub-vectors of each dimension of the trustworthy vector in the second global trust matrix.

[0050] S305: Calculate the mean of multiple second subvectors.

[0051] S306: Calculate the average algorithm variance between the true and predicted impact energy labels of historical samples, and use it as the impact energy variance error label.

[0052] S307: Train the impact energy prediction reliability assessment optimization model to be trained based on each mean and the corresponding impact energy variance error label to obtain the trained impact energy prediction reliability assessment optimization model.

[0053] This application embodiment is a method for preparatory evaluation of the reliability of prediction results based on an existing impact energy prediction model. First, the trained impact energy prediction reliability evaluation optimization model needs to determine the second direct trust value feature, the second recommended trust value feature value, the second global trust matrix, and the impact energy variance error label among various historical samples in the historical data. The second global trust matrix and the impact energy variance error label are then calculated to train the impact energy prediction reliability evaluation optimization model. Next, the first global trust matrix between the sample to be predicted and each historical sample in the historical data is determined. Using the credibility feature of the first global trust matrix between the sample to be predicted and the historical samples as input factors, the trained impact energy prediction reliability evaluation optimization model is used to estimate the impact energy prediction reliability value of the sample to be predicted.

[0054] The historical data can be collected by acquiring a large amount of historical slab sample characteristic data from the on-site production and manufacturing system, cleaning and normalizing the collected data, and using the normalized data as historical data.

[0055] For each historical sample, the second global trust value is obtained in the above manner. The calculation method for the second global trust value is as follows: , where 0 1, The second global trust value for the correlation of production process parameters between historical sample i and historical sample j.

[0056] A second global trust matrix is ​​constructed based on each second global trust value, and its form is as follows: .

[0057] Using the second global trust matrix composed of the second global trust values ​​of each historical sample calculated above, the credibility features of each historical sample are extracted.

[0058] Specifically, this may include the following steps: For the second global trust matrix, determine multiple second sub-vectors for each dimension of the trust vector. For example, the one-dimensional trust vector includes: f(0,0) f(0,1)…f(0,m-1), f(1,0) f(1,1)…f(1,m-1)…f(k-1) f(k-1,1)…f(k-1,m-1).

[0059] For each one-dimensional trustworthy vector, a sliding window mechanism is used to extract multiple second sub-vectors.

[0060] Specifically, a sliding window mechanism is used to obtain X sets of data for each historical sample. The size of the window is determined based on the number of samples in the actual experiment. X sets of second vectors are then obtained; in this example, X is set to 6. The sliding window mechanism is used to select 6 sets of second sub-vectors with the same window length y = 8. The window length can be set to (200, 300, 400, 500, 600, 700), forming... Vector matrices. Calculate the arithmetic mean of six vector matrices.

[0061] The historical samples also include the true and predicted impact energy labels for each historical sample. The average algorithmic variance between the true and predicted impact energy labels for the historical samples is calculated and used as the impact energy variance error label.

[0062] The mean of each second sub-vector is used as the input to the impact energy prediction reliability assessment optimization model, and the impact energy variance error label is used as the input to the impact energy prediction reliability assessment optimization model. Based on each mean and the corresponding impact energy variance error label, the impact energy prediction reliability assessment optimization model to be trained is trained to obtain the trained impact energy prediction reliability assessment optimization model.

[0063] In one example, this application attempts corresponding algorithms for various scenarios and ultimately selects the nonlinear regression algorithm BILSTM (Bidirectional Long Short-Term Memory) as the optimization model framework for the reliability assessment of impact energy prediction. This example can be extended to any regression network algorithm.

[0064] This application can perform error analysis on traditional prediction results by integrating an optimized BILSTM algorithm model with a recommendation trust mechanism, determine the applicability of the prediction model in various scenarios, and optimize the model's feedback on production process control capabilities through impact energy prediction reliability assessment, thereby significantly reducing the risk of product quality disputes and effectively guiding the extraction of production sample intervals to reduce experimental costs.

[0065] There are multiple ways to determine the second direct trust value feature among historical samples in historical data. In one implementation, such as... Figure 4 As shown, it includes the following steps: S401: For the historical sample i, determine the risk coefficient of the historical sample i.

[0066] S402: Determine the second direct trust value feature among historical samples in historical data based on the risk coefficient.

[0067] There are multiple ways to determine the risk coefficient of historical sample i. In one method, such as... Figure 5 As shown, it includes the following steps: S501: Determine the centroid values ​​of multiple attributes contained in historical sample i.

[0068] S502: Calculate the Eulerian distance between each attribute and the centroid value in historical sample i.

[0069] S503: Calculate the ratio of the Eulerian distance to the distance threshold, which serves as the risk coefficient for the attribute.

[0070] It should be noted that the historical samples may include multiple attributes, and the influencing factors derived from actual production analysis mainly include: C, Si, Mn, P, S, Ni, Vi, Ti, CEV, final rolling temperature, red-hot temperature, finishing rolling temperature, furnace time, tensile strength, yield strength, elongation, and other attributes.

[0071] Each attribute is normalized, and based on the normalized historical samples, the centroid value of each attribute is determined. For example, if the normalized historical sample i is represented as {0.5, 0.2, 0.5, 0.8, 0.7, 0.1, 0.2, 0.3, 0.8, 0.4, 0.1, 0.2, 0.3, 0.7, 0.6, 0.5}, then the mean of each attribute in the normalized historical sample is calculated and used as the centroid value of the historical sample i containing multiple attributes. That is, the centroid value of the normalized historical sample i is as follows: (0.5+0.2+0.5+0.8+0.7+0.1+0.2+0.3+0.8+0.4+0.1+0.2+0.3+0.7+0.6+0.5) / 16=0.43125.

[0072] The Eulerian distance between historical sample i and the centroid value is calculated using the following formula: ; in, Let be the Eulerian distance between each attribute and the centroid value in historical sample i. x i1 , x i2 ... x im (i) represents historical sample i. x i1 , x i2 ... x im Let be the normalized attributes in historical sample i, and m be the total number of attributes in historical sample i.

[0073] The Eulerian distance and distance threshold of each attribute in historical sample i are compared to realize the transformation of Eulerian distance into risk coefficient of each attribute. Attributes with a ratio of Eulerian distance to distance threshold of less than 0.4 are marked as important attributes. The smaller the ratio, the more important the attribute is in historical sample i.

[0074] There are several ways to determine the second direct trust value feature among historical samples in historical data based on risk coefficients. In one implementation method, such as... Figure 6 As shown, it includes the following steps: S601: Determine the first set of attributes for historical sample i.

[0075] S602: Determine the set of second attributes of historical sample j.

[0076] S603: Based on the first attribute set and the second attribute set, determine the similarity values ​​of the attribute curves of historical sample i and historical sample j.

[0077] The core objective of determining the similarity values ​​of the attribute curves of historical sample i and historical sample j is to compare the similarity of the time series curves (i.e., "attribute curves") of two historical samples (i and j) on multiple attributes. The attribute curve of each sample is represented in time series form (for example, the curve of the red temperature attribute of sample i changing with time is represented as [(t1,v1), (t2, v2), ...]). The smaller the value calculated by the Euclidean distance formula, the more similar the attribute curves are.

[0078] S604: Based on similarity values ​​and risk coefficients, determine the second direct trust value feature among historical samples in historical data.

[0079] Specifically, the second direct trust value feature is calculated using the following formula: ; in, Let be the second direct trust value feature between historical sample i and historical sample j, m be the total number of attributes between historical sample i and historical sample j, S(im,jm) be the similarity value of each attribute curve between historical sample i and historical sample j, and (i,m) represent the risk coefficient of each attribute in historical sample i.

[0080] The second recommendation trust value feature is calculated using the following formula: ; Where R(im,jm) is the second recommendation trust value between historical sample i and historical sample j. This represents the ratio of the number of similarity values ​​between the recommended node M and historical sample i that are greater than a preset threshold to the total number of attributes.

[0081] The recommended node M is obtained from historical samples i and j. The condition for determining node M as a recommended node is that the node must simultaneously satisfy the condition that the mean similarity of attributes with nodes i and j is greater than a set threshold. The similarity of each attribute of nodes i and j is obtained by Euler's formula.

[0082] The second global trust value is calculated using the following formula: ; in Let be the global trust value between historical sample i and historical sample j, 0 1. R(im,jm) is the second recommendation trust value between historical sample i and historical sample j. The second direct trust value feature is the historical sample i and historical sample j.

[0083] Based on the first and second attribute sets, the similarity values ​​of each attribute curve for historical sample i and historical sample j are determined. The similarity values ​​are calculated using the Manhattan distance method as follows: .

[0084] It is the first attribute in the first attribute set. It is the first attribute in the second attribute set. For the m-th parameter in the first attribute set, Let m be the m-th attribute in the second attribute set.

[0085] Please refer to Figure 7 This application embodiment also provides an application for Figure 1The thick plate impact energy prediction reliability assessment device 110 of the electronic device 100 includes: The determination module 111 is used to determine the first direct trust value feature and the first recommended trust value feature value between the sample to be predicted and each historical sample in each historical data; based on each first direct trust value feature and the corresponding first recommended trust value feature value, determine each first global trust value between the sample to be predicted and each historical data; and based on each first global trust value, obtain the first global trust matrix. The acquisition module 112 is used to acquire multiple first sub-vectors based on a sliding window for each one-dimensional trustworthy vector in the first global trust matrix, wherein the first global trust matrix includes multiple one-dimensional trustworthy vectors. The calculation module 113 is used to calculate the mean of multiple first sub-vectors; The prediction module 114 is used to input the mean values ​​into the trained impact energy prediction reliability evaluation optimization model to obtain the impact energy variance error of the sample to be predicted.

[0086] This application also provides an electronic device 100, which includes a processor 130 and a memory 120. The memory 120 stores computer-executable instructions, which, when executed by the processor 130, implement the thick plate impact energy prediction reliability assessment method.

[0087] This application embodiment also provides a computer-readable storage medium storing a computer program, which, when executed by the processor 130, implements the thick plate impact energy prediction reliability assessment method.

[0088] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0089] Furthermore, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part. If the function is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0090] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0091] The above descriptions are merely various embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A reliability assessment method for predicting the impact energy of thick plates, characterized in that, The method includes: Determine the first direct trust value feature and the first recommendation trust value feature value between the sample to be predicted and each historical sample in each historical data; Based on each of the first direct trust value features and the corresponding first recommended trust value features, the first global trust value of the sample to be predicted and each of the historical data is determined. Based on each first global trust value, the first global trust matrix is ​​obtained; For each one-dimensional trustworthy vector in the first global trust matrix, multiple first sub-vectors are obtained based on a sliding window, wherein the first global trust matrix includes multiple one-dimensional trustworthy vectors; Calculate the mean of multiple first sub-vectors; The mean values ​​are input into the trained impact energy prediction reliability evaluation optimization model to obtain the impact energy variance error of the sample to be predicted.

2. The method according to claim 1, characterized in that, The method further includes: Determine the second direct trust value feature and the second recommended trust value feature value among the historical samples in the historical data; Based on the second direct trust value feature and the second recommended trust value feature, a second global trust value is determined among the historical samples; Construct a second global trust matrix based on each of the second global trust values; A sliding window mechanism is used to extract multiple second sub-vectors for each dimension of the trustworthy vector in the second global trust matrix; Calculate the mean of multiple second sub-vectors; The average algorithm variance between the true impact energy label value and the predicted impact energy label value of the historical samples is calculated and used as the impact energy variance error label. The impact energy prediction reliability assessment optimization model is trained based on the mean and the corresponding impact energy variance error label to obtain the trained impact energy prediction reliability assessment optimization model.

3. The method according to claim 2, characterized in that, The step of determining the second direct trust value feature among historical samples in the historical data includes: For the historical sample i, determine the risk coefficient of the historical sample i; Based on the risk coefficient, a second direct trust value feature is determined among the historical samples in the historical data.

4. The method according to claim 3, characterized in that, The step of determining the risk coefficient of the historical sample i includes: Determine the centroid values ​​of multiple attributes contained in the historical sample i; Calculate the Eulerian distance between each attribute in the historical sample i and the centroid value; The ratio of the Eulerian distance to the distance threshold is calculated and used as the risk coefficient of the attribute.

5. The method according to claim 3, characterized in that, The step of determining the second direct trust value feature among the historical samples in the historical data based on the risk coefficient includes: Determine the first set of attributes for historical sample i; Determine the set of second attributes of historical sample j; Based on the first attribute set and the second attribute set, the similarity values ​​of each attribute curve of the historical sample i and the historical sample j are determined; Based on the similarity value and each of the risk coefficients, a second direct trust value feature is determined among the historical samples in the historical data.

6. The method according to claim 5, characterized in that, The second direct trust value feature is calculated using the following formula: ； in, Let be the second direct trust value feature between historical sample i and historical sample j, m be the total number of attributes between historical sample i and historical sample j, S(im,jm) be the similarity value of each attribute curve between historical sample i and historical sample j, and (i,m) represent the risk coefficient of each attribute in historical sample i.

7. The method according to claim 6, characterized in that, The second recommendation trust value feature is calculated using the following formula: ； Where R(im,jm) is the second recommendation trust value between historical sample i and historical sample j. This represents the ratio of the number of recommended nodes M with similarity values ​​greater than a preset threshold to the total number of attributes. Recommended nodes M are obtained from the historical samples i and j, where the average similarity of the node M with the attributes of nodes i and j is greater than a set threshold.

8. The method according to claim 2, characterized in that, The second global trust value is calculated using the following formula: ； in The global trust value between historical sample i and historical sample j, 0 1. R(im,jm) is the second recommendation trust value between historical sample i and historical sample j. The second direct trust value feature is the historical sample i and historical sample j.

9. A reliability assessment device for predicting the impact energy of thick plates, characterized in that, The device includes: The determination module is used to determine the first direct trust value feature and the first recommended trust value feature value between the sample to be predicted and each historical sample in each historical data; based on each first direct trust value feature and the corresponding first recommended trust value feature value, determine each first global trust value between the sample to be predicted and each of the historical data; and based on each first global trust value, obtain the first global trust matrix. The acquisition module is used to acquire multiple first sub-vectors based on a sliding window for each one-dimensional trustworthy vector in the first global trust matrix, wherein the first global trust matrix includes multiple one-dimensional trustworthy vectors. A calculation module is used to calculate the mean of multiple first sub-vectors; The prediction module is used to input the mean values ​​into the trained impact energy prediction reliability evaluation optimization model to obtain the impact energy variance error of the sample to be predicted.

10. An electronic device, characterized in that, The method includes a memory and a processor, the memory storing a computer program, characterized in that the processor executes the computer program to implement the steps of the method according to any one of claims 1-8.

Images