Model training method, data prediction method, computing device and readable storage medium
By constructing and combining multiple neural networks, a data prediction agent model is formed, which solves the problem that accurate data cannot be obtained during long-term use of engineering machinery, and achieves higher performance data prediction effects.
Patent Information
- Application Number
- CN202411665047.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-20
- Publication Date
- 2025-03-11
AI Technical Summary
In the prior art, long-term use of engineering machinery cannot obtain accurate stress and displacement data, resulting in problems such as reduced measurement accuracy and missing data.
A model training method is adopted to build multiple neural networks, train and combine network models based on the operation data and target data of the engineering machinery to form a data prediction agent model to improve the accuracy of data prediction.
A higher performance data prediction agent model is realized, which can obtain more accurate data such as stress and displacement of engineering machinery, solving the data accuracy and integrity problems.
Smart Images
Figure CN119599147A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of data processing technology, and specifically to a model training method, a data prediction method, a computing device, and a readable storage medium. Background Art
[0002] With the rapid development of engineering machinery manufacturing technology, engineering machinery is widely used in various scenarios. After engineering machinery such as tower cranes are put into use, fatigue damage to the structure of the components is likely to occur due to the fact that some structures of the components are weak points in terms of static strength and fatigue strength, as well as the difference between the actual use conditions and the preset use conditions. Usually, a variety of sensors are installed at the key components of engineering machinery to detect data such as displacement and stress during the operation of the engineering machinery. The remaining life of the engineering machinery is predicted based on the data detected by the sensors to obtain the quality analysis results of the engineering machinery and avoid safety accidents.
[0003] However, the deformable parts of the sensor itself, such as the stress plate, will also suffer fatigue damage, causing the signal collected by the sensor to drift, which will affect the measurement accuracy of stress and displacement data, making it impossible to obtain accurate displacement and stress data. In addition, as the working time of the sensor to collect signals continues to accumulate, it is easy to cause data loss due to sensor failure, further making it impossible to accurately obtain data on engineering machinery. Summary of the invention
[0004] The purpose of the embodiments of the present application is to provide a model training method, a data prediction method, a computing device and a readable storage medium to solve the problem in the prior art that accurate data cannot be obtained for engineering machinery that has been used for a long time.
[0005] In order to achieve the above-mentioned purpose, the first aspect of the present application provides a model training method, which includes:
[0006] Constructing a first neural network according to the number of nodes corresponding to the types of operation data of the construction machinery and the nodes corresponding to the target data to be predicted, wherein the target data includes stress and / or displacement;
[0007] Training a first neural network, and determining all first nodes and all second nodes in the trained first neural network, taking the second nodes as initial nodes, wherein the precision of the first nodes is greater than or equal to a preset precision, and the precision of the second nodes is less than the preset precision;
[0008] When the number of the second nodes is not zero, construct the nth neural network corresponding to all the initial nodes;
[0009] Training an nth neural network, and determining all third nodes and all fourth nodes in the trained nth neural network, wherein the precision of the third nodes is greater than or equal to a preset precision, and the precision of the fourth nodes is less than the preset precision;
[0010] When the number of the fourth nodes is not zero, the initial nodes are updated using the fourth nodes, and the step of constructing the nth neural network corresponding to all the initial nodes is performed until the number of the fourth nodes is zero;
[0011] Based on the first neural network and all n-th neural networks, a data prediction agent model is constructed.
[0012] In an embodiment of the present application, a first neural network is constructed according to the number of nodes corresponding to the types of operation data of the engineering machinery and the number of nodes corresponding to the target data to be predicted, including:
[0013] Based on the number of types of operation data of the construction machinery, obtaining at least one input node of the neural network;
[0014] Based on the key points corresponding to the target data to be predicted of the engineering machinery, at least one output node of the neural network is obtained;
[0015] Construct the first neural network based on all input nodes and all output nodes.
[0016] In an embodiment of the present application, at least one output node of a neural network is obtained based on key points corresponding to target data to be predicted of engineering machinery, including:
[0017] Determine at least one key point of the engineering machinery according to the stress cloud map of the engineering machinery, wherein the key point is at least one of a structural position where the stress is greater than a preset stress, a structural position where the load is greater than a preset load threshold, and a structural position where the fatigue life is weak;
[0018] An output node of the neural network is obtained corresponding to each key point.
[0019] In an embodiment of the present application, the construction machinery is a tower crane, the operation data includes at least one of the lifting weight, lifting amplitude, rotation angle and wind speed of the tower crane, and the target data includes stress and / or displacement of key points.
[0020] In an embodiment of the present application, at least one input node of a neural network is obtained based on the number of types of operation data of the engineering machinery, including:
[0021] determining at least one data type corresponding to the operating data;
[0022] An input node of the neural network is obtained corresponding to each data type.
[0023] In an embodiment of the present application, the model training method further includes:
[0024] Obtain historical operation data of construction machinery;
[0025] Simulate the construction machinery based on historical operation data to obtain simulation target data of the construction machinery;
[0026] According to the historical operation data and simulation target data, a relational database between the operation data and target data of the construction machinery is constructed;
[0027] Train the first neural network, and determine all first nodes and all second nodes in the trained first neural network, including:
[0028] training a first neural network based on the relational database, and determining all first nodes and all second nodes in the trained first neural network;
[0029] Train the nth neural network, and determine all third nodes and all fourth nodes in the trained nth neural network, including:
[0030] An nth neural network is trained based on the relational database, and all third nodes and all fourth nodes in the trained nth neural network are determined.
[0031] In an embodiment of the present application, a data prediction agent model is constructed based on the first neural network and all n-th neural networks, including:
[0032] Build a summary network;
[0033] The first neural network, all n-th neural networks and the summary network are combined to obtain a combined network model, wherein the first node and the third node whose accuracy is greater than or equal to the preset accuracy transmit data to the summary network;
[0034] Train the combined network model to obtain the data prediction agent model.
[0035] A second aspect of the present application provides a data prediction method, the data prediction method comprising:
[0036] Obtain real-time operation data of construction machinery;
[0037] The real-time operation data of the construction machinery is input into the data prediction agent model to obtain the predicted target data of the construction machinery, wherein the data prediction agent model is obtained according to the above-mentioned model training method.
[0038] A third aspect of the present application provides a computing device, including:
[0039] a memory configured to store instructions;
[0040] A processor is configured to call instructions from a memory and implement the above-mentioned model training method or the above-mentioned data prediction method when executing the instructions.
[0041] A fourth aspect of the present application provides a machine-readable storage medium having instructions stored thereon, the instructions being used to enable a machine to execute the above-mentioned model training method or the above-mentioned data prediction method.
[0042] The present application provides a model training method, including: constructing a first neural network according to the number of nodes corresponding to the type of operation data of the engineering machinery and the nodes corresponding to the target data to be predicted; training the first neural network, and determining all the first nodes and all the second nodes in the first neural network after training, and taking the second node as the initial node; when the number of the second nodes is not zero, constructing the nth neural network corresponding to all the initial nodes; training the nth neural network, and determining all the third nodes and all the fourth nodes in the nth neural network after training; when the number of the fourth nodes is not zero, using the fourth node to update the initial node, and executing the step of constructing the nth neural network corresponding to all the initial nodes until the number of the fourth nodes is zero; constructing a data prediction proxy model based on the first neural network and all the nth neural networks. Construct multiple neural networks according to the actual accuracy of each node, and summarize the multiple neural networks to obtain the data prediction proxy model. Since the neural network reduces the scale of network parameters, it not only improves the model training efficiency, but also reduces the requirements of the proxy model on the training data, so that a higher performance proxy model can be obtained, and then more accurate data such as stress and displacement of the engineering machinery can be obtained through the proxy model.
[0043] Other features and advantages of the embodiments of the present application will be described in detail in the subsequent specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present application, but do not constitute a limitation on the embodiments of the present application. In the accompanying drawings:
[0045] Figure 1 A schematic diagram of a process flow of a model training method according to an embodiment of the present application is schematically shown;
[0046] Figure 2 An example diagram of a neural network according to an embodiment of the present application is schematically shown;
[0047] Figure 3 An example diagram of a data prediction agent model according to an embodiment of the present application is schematically shown;
[0048] Figure 4 A flow chart of a data prediction method according to an embodiment of the present application is schematically shown. DETAILED DESCRIPTION
[0049] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the specific implementation methods described herein are only used to illustrate and explain the embodiments of the present application, and are not used to limit the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.
[0050] It should be noted that if directional indication is involved in the embodiments of the present application, the directional indication is only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.
[0051] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present application, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by this application.
[0052] Example 1
[0053] Figure 1 The following schematically shows a flow chart of a model training method according to an embodiment of the present application. Figure 1 As shown, the embodiment of the present application provides a model training method, which includes the following steps:
[0054] S110, constructing a first neural network according to the number of nodes corresponding to the types of operation data of the engineering machinery and the nodes corresponding to the target data to be predicted, wherein the target data includes stress and / or displacement.
[0055] The type of construction machinery is based on the settings and is not limited here. For ease of understanding, the construction machinery in the embodiment of the present application is a tower crane, that is, a tower crane. Usually, a tower crane needs to obtain target data at multiple positions, and there are also differences in the mapping relationship between operating data such as external loads and different positions. Considering the dynamic load effect, there is a wave equation between load and stress. It is difficult to ensure that the accuracy of all nodes of the algorithm model meets the requirements by directly using a feedforward network to fit the nonlinear mapping relationship between load and stress, and it is difficult to accurately obtain the data of the construction machinery. In this embodiment, the first neural network is constructed according to the nodes corresponding to the number of types of operating data of the construction machinery and the nodes corresponding to the target data to be predicted.
[0056] In an embodiment of the present application, a first neural network is constructed according to the number of nodes corresponding to the types of operation data of the engineering machinery and the number of nodes corresponding to the target data to be predicted, including:
[0057] Based on the number of types of operation data of the construction machinery, obtaining at least one input node of the neural network;
[0058] Based on the key points corresponding to the target data to be predicted of the engineering machinery, at least one output node of the neural network is obtained;
[0059] Construct the first neural network based on all input nodes and all output nodes.
[0060] Based on the number of types of operation data of engineering machinery, at least one input node of the neural network is obtained, that is, the number of output nodes of the neural network is equal to the number of types of operation data. Usually, engineering machinery such as tower cranes include multiple components or structures. In actual fatigue damage analysis and other scenarios, only the target data of key components or key structures of the tower crane need to be obtained. Based on the key points corresponding to the target data to be predicted of the engineering machinery, at least one output node of the neural network is obtained, that is, the number of output nodes of the neural network is equal to the number of key points. Construct the first neural network based on all input nodes and all output nodes.
[0061] In an embodiment of the present application, at least one output node of a neural network is obtained based on key points corresponding to target data to be predicted of engineering machinery, including:
[0062] Determine at least one key point of the engineering machinery according to the stress cloud map of the engineering machinery, wherein the key point is at least one of a structural position where the stress is greater than a preset stress, a structural position where the load is greater than a preset load threshold, and a structural position where the fatigue life is weak;
[0063] An output node of the neural network is obtained corresponding to each key point.
[0064] The key components or key structures of the construction machinery are strength-checked. According to the stress cloud map of the construction machinery, the structural positions where the stress is greater than the preset stress, the structural positions where the load is greater than the preset load threshold, and the structural positions where the fatigue life is weak are determined, and then at least one key point of the construction machinery is determined. An output node of the neural network is obtained for each key point, and the output of the proxy model is the target data of each key point of the tower crane.
[0065] In an embodiment of the present application, the construction machinery is a tower crane, the operation data includes at least one of the lifting weight, lifting amplitude, rotation angle and wind speed of the tower crane, and the target data includes stress and / or displacement of key points.
[0066] Usually, the operation data includes multiple types of data. When the construction machinery is a tower crane, the operation data includes at least one of the lifting weight, lifting amplitude, rotation angle and wind speed of the tower crane. The target data is one of stress and displacement. It should be understood that the difference in target data will lead to different numbers of nodes in the neural network, which will not be elaborated here.
[0067] In an embodiment of the present application, at least one input node of a neural network is obtained based on the number of types of operation data of the engineering machinery, including:
[0068] determining at least one data type corresponding to the operating data;
[0069] An input node of the neural network is obtained corresponding to each data type.
[0070] Determine at least one data type corresponding to the operation data. For ease of understanding, in the embodiment of the present application, the number of types of operation data is 4, that is, the operation data also includes the lifting weight, lifting amplitude, rotation angle and wind speed of the tower crane, and the number of input nodes of the proxy model is 4. An input node of the neural network is obtained corresponding to each data type, and the inputs of the proxy model are the lifting weight, lifting amplitude, rotation angle and wind speed of the tower crane.
[0071] See also Figure 2 , Figure 2 An example diagram of a neural network according to an embodiment of the present application is schematically shown.
[0072] Taking the first neural network as an example, the number of input nodes of the neural network is I, the number of output nodes is J, the number of layers of the neural network is 2, and the data of the intermediate nodes in the first layer is The number of intermediate nodes in the second layer is The structure of the first neural network is shown in the figure. The input of the neural network is xi = [G, H, D, W], where xi is the input data of the i-th input node, G is the lifting weight, H is the lifting amplitude, D is the rotation angle, W is the wind speed, and the output of the neural network is Y i =[y1,y,...,y J ], where Y i is the output data of the i-th output node, y J For the J The predicted target data of key points.
[0073] S120, training a first neural network, and determining all first nodes and all second nodes in the trained first neural network, taking the second nodes as initial nodes, wherein the accuracy of the first nodes is greater than or equal to a preset accuracy, and the accuracy of the second nodes is less than the preset accuracy.
[0074] The first neural network is trained multiple times to obtain a first neural network after training. The first neural network after training is tested to determine the first node that meets the accuracy requirement and the second node that does not meet the accuracy requirement, that is, the accuracy of the first node is greater than or equal to the preset accuracy, and the accuracy of the second node is less than the preset accuracy. The second node is used as an initial node to construct a new neural network.
[0075] S130, when the number of the second nodes is not zero, constructing an nth neural network corresponding to all the initial nodes.
[0076] When the number of the second nodes is not zero, it is determined that there is a neural network in the first neural network that does not meet the accuracy requirement, and the nth neural network corresponding to all the initial nodes is constructed. The value of n changes continuously according to the number of neural networks constructed. When the first neural network is constructed, the initial value of n is 2.
[0077] S140, training the nth neural network, and determining all third nodes and all fourth nodes in the trained nth neural network, wherein the accuracy of the third nodes is greater than or equal to a preset accuracy, and the accuracy of the fourth nodes is less than a preset accuracy.
[0078] The nth neural network is trained multiple times to obtain the trained nth neural network. The trained nth neural network is tested to determine the third node that meets the accuracy requirement and the fourth node that does not meet the accuracy requirement, that is, the accuracy of the third node is greater than or equal to the preset accuracy, and the accuracy of the fourth node is less than the preset accuracy.
[0079] S150, when the number of the fourth nodes is not zero, using the fourth nodes to update the initial nodes, and executing the step of constructing the nth neural network corresponding to all the initial nodes until the number of the fourth nodes is zero.
[0080] When the number of the fourth node is not zero, after n neural networks are constructed, there are still nodes that do not meet the accuracy requirements. The fourth node is used to update the initial node, and the count of n is increased by one. The steps of constructing the nth neural network corresponding to all initial nodes are executed, and the new neural network is repeatedly constructed and trained until the number of the fourth node is zero, that is, the number of nodes that do not meet the accuracy requirements is zero.
[0081] S160, building a data prediction agent model based on the first neural network and all n-th neural networks.
[0082] According to the actual accuracy of each node, multiple neural networks are constructed, and multiple neural networks are aggregated to obtain a data prediction proxy model. Since the neural network reduces the scale of network parameters, it not only improves the model training efficiency, but also reduces the requirements of the proxy model for training data, and can obtain a higher performance proxy model, and then more accurate data such as stress and displacement of engineering machinery can be obtained through the proxy model.
[0083] In an embodiment of the present application, the model training method further includes:
[0084] Obtain historical operation data of construction machinery;
[0085] Simulate the construction machinery based on historical operation data to obtain simulation target data of the construction machinery;
[0086] According to the historical operation data and simulation target data, a relational database between the operation data and target data of the construction machinery is constructed;
[0087] Train the first neural network, and determine all first nodes and all second nodes in the trained first neural network, including:
[0088] training a first neural network based on the relational database, and determining all first nodes and all second nodes in the trained first neural network;
[0089] Train the nth neural network, and determine all third nodes and all fourth nodes in the trained nth neural network, including:
[0090] An nth neural network is trained based on the relational database, and all third nodes and all fourth nodes in the trained nth neural network are determined.
[0091] Acquire historical operation data of the construction machinery. Simulate the construction machinery based on the historical operation data to obtain simulation target data of the construction machinery. Construct a relational database between the operation data and target data of the construction machinery based on the historical operation data and the simulation target data.
[0092] Finite element analysis is performed on the components or structures of the tower crane to obtain the weak parts of the static strength and fatigue strength of the components or structures of the tower crane, and then the location and number of key points of the tower crane are determined. The operation data of the tower crane and the stress data of the corresponding key points are obtained through simulation tests or existing Internet of Things data platforms to form a relational database. By establishing a dynamic simulation model and obtaining external loads through a large number of simulations, a relational database is formed.
[0093] A training sample is obtained based on a relational database, and then a first neural network is trained based on the relational database, and all first nodes and all second nodes in the trained first neural network are determined, and an nth neural network is trained based on the relational database, and all third nodes and all fourth nodes in the trained nth neural network are determined.
[0094] In an embodiment of the present application, a data prediction agent model is constructed based on the first neural network and all n-th neural networks, including:
[0095] Build a summary network;
[0096] The first neural network, all n-th neural networks and the summary network are combined to obtain a combined network model, wherein the first node and the third node whose accuracy is greater than or equal to the preset accuracy transmit data to the summary network;
[0097] Train the combined network model to obtain the data prediction agent model.
[0098] When the number of the fourth node is zero, determine that all nodes meet the accuracy requirements and build a summary network. Use the output of each neural network as the input of the summary network, combine the first neural network, all nth neural networks and the summary network to obtain a combined network model. The first node and the third node whose accuracy is greater than or equal to the preset accuracy transmit the output data to the summary network, and the data output by the second node and the fourth node whose accuracy is less than the preset accuracy can be discarded. Train the combined network model to obtain a data prediction proxy model. The performance of the data prediction proxy model can be further improved by the combined network model, thereby obtaining more accurate target data.
[0099] See also Figure 3 , Figure 3 An example diagram of a data prediction agent model according to an embodiment of the present application is schematically shown.
[0100] Based on the first neural network and all n-th neural networks, the first neural network and all n-th neural networks are combined to construct a data prediction proxy model. As shown in the figure, the value of n in this embodiment is 2 to 4, that is, the first neural network, the second neural network, the third neural network and the fourth neural network are constructed. The first neural network, the second neural network, the third neural network, the fourth neural network and the summary network are trained to obtain sub-model 1, sub-model 2, sub-model 3, sub-model 4 and the summary model respectively, and the sub-model 1, sub-model 2, sub-model 3, sub-model 4 and the summary model are combined to obtain the combined network model. The combined network model is trained to obtain the data prediction proxy model.
[0101] The present application provides a model training method, including: constructing a first neural network according to the number of nodes corresponding to the type of operation data of the engineering machinery and the nodes corresponding to the target data to be predicted; training the first neural network, and determining all the first nodes and all the second nodes in the first neural network after training, and taking the second node as the initial node; when the number of the second nodes is not zero, constructing the nth neural network corresponding to all the initial nodes; training the nth neural network, and determining all the third nodes and all the fourth nodes in the nth neural network after training; when the number of the fourth nodes is not zero, using the fourth node to update the initial node, and executing the step of constructing the nth neural network corresponding to all the initial nodes until the number of the fourth nodes is zero; constructing a data prediction proxy model based on the first neural network and all the nth neural networks. Construct multiple neural networks according to the actual accuracy of each node, and summarize the multiple neural networks to obtain the data prediction proxy model. Since the neural network reduces the scale of network parameters, it not only improves the model training efficiency, but also reduces the requirements of the proxy model on the training data, so that a higher performance proxy model can be obtained, and then more accurate data such as stress and displacement of the engineering machinery can be obtained through the proxy model.
[0102] Example 2
[0103] Figure 4 The following schematically shows a flow chart of a data prediction method according to an embodiment of the present application. Figure 4 As shown, the embodiment of the present application provides a data prediction method, which includes the following steps:
[0104] S210, acquiring real-time operation data of the construction machinery.
[0105] The construction machinery is a tower crane. When it is necessary to obtain the predicted data of the tower crane, the real-time operation data of the tower crane is obtained. It should be understood that the real-time operation data is set according to actual needs, and can be load data such as the lifting weight, lifting amplitude, rotation angle and wind speed of the tower crane. The predicted data is also set according to actual needs, and can be stress and displacement, etc., which are not limited here.
[0106] S220, inputting the real-time operation data of the construction machinery into a data prediction agent model to obtain predicted target data of the construction machinery, wherein the data prediction agent model is obtained according to the above-mentioned model training method.
[0107] The real-time operation data is input into the data prediction proxy model to obtain the predicted target data of the tower crane output by the data prediction proxy model. Since the neural network reduces the scale of network parameters, it not only improves the model training efficiency, but also reduces the requirements of the proxy model for training data, and can obtain a higher performance proxy model. Through the proxy model, more accurate data such as stress and displacement of engineering machinery can be obtained.
[0108] The present application also provides a computing device, including:
[0109] a memory configured to store instructions;
[0110] A processor is configured to call instructions from a memory and implement the above-mentioned model training method or the above-mentioned data prediction method when executing the instructions.
[0111] The processor includes a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set, and the problem of failing to obtain accurate data for long-term used engineering machinery in the prior art can be solved by adjusting kernel parameters.
[0112] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.
[0113] An embodiment of the present application also provides a machine-readable storage medium, on which instructions are stored, and the instructions are used to enable a machine to execute the above-mentioned model training method or the above-mentioned data prediction method.
[0114] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.
[0115] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0116] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0117] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0118] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0119] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.
[0120] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.
[0121] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.
[0122] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.
Claims
1. A model training method, characterized in that: The model training method comprises: Constructing a first neural network according to the number of nodes corresponding to the types of operation data of the engineering machinery and the nodes corresponding to the target data to be predicted, wherein the target data includes stress and / or displacement; Training the first neural network, and determining all first nodes and all second nodes in the trained first neural network, and taking the second nodes as initial nodes, wherein the precision of the first nodes is greater than or equal to a preset precision, and the precision of the second nodes is less than the preset precision; When the number of the second nodes is not zero, constructing an nth neural network corresponding to all the initial nodes; Training the nth neural network, and determining all third nodes and all fourth nodes in the trained nth neural network, wherein the precision of the third nodes is greater than or equal to a preset precision, and the precision of the fourth nodes is less than the preset precision; When the number of the fourth nodes is not zero, the initial nodes are updated using the fourth nodes, and the step of constructing the nth neural network corresponding to all the initial nodes is performed until the number of the fourth nodes is zero; Based on the first neural network and all the n-th neural networks, a data prediction agent model is constructed.
2. The model training method according to claim 1, characterized in that: The first neural network is constructed according to the nodes corresponding to the number of types of the operation data of the engineering machinery and the nodes corresponding to the target data to be predicted, including: Based on the number of types of operation data of the construction machinery, obtaining at least one input node of the neural network; Based on the key points corresponding to the target data to be predicted of the engineering machinery, at least one output node of the neural network is obtained; A first neural network is constructed according to all the input nodes and all the output nodes.
3. The model training method according to claim 2, characterized in that: The step of obtaining at least one output node of a neural network based on key points corresponding to target data to be predicted of the engineering machinery includes: Determine at least one key point of the engineering machinery according to the stress cloud map of the engineering machinery, wherein the key point is at least one of a structural position where the stress is greater than a preset stress, a structural position where the load is greater than a preset load threshold, and a structural position where the fatigue life is weak; An output node of the neural network is obtained corresponding to each key point.
4. The model training method according to claim 3, characterized in that: The engineering machinery is a tower crane, the operating data includes at least one of the lifting weight, lifting amplitude, rotation angle and wind speed of the tower crane, and the target data includes the stress and / or displacement of the key points.
5. The model training method according to claim 2, characterized in that: The method of obtaining at least one input node of the neural network based on the number of types of operation data of the engineering machinery includes: determining at least one data type corresponding to the operating data; An input node of the neural network is obtained corresponding to each data type.
6. The model training method according to claim 1, characterized in that: The model training method also includes: Acquiring historical operation data of the engineering machinery; Simulating the engineering machinery based on the historical operation data to obtain simulation target data of the engineering machinery; Constructing a relational database between the operation data and the target data of the construction machinery according to the historical operation data and the simulation target data; The step of training the first neural network and determining all first nodes and all second nodes in the trained first neural network includes: Training the first neural network based on the relational database, and determining all first nodes and all second nodes in the trained first neural network; The training of the nth neural network and determining all third nodes and all fourth nodes in the trained nth neural network include: The nth neural network is trained based on the relational database, and all third nodes and all fourth nodes in the trained nth neural network are determined.
7. The model training method according to claim 1, characterized in that: The step of constructing a data prediction agent model based on the first neural network and all the n-th neural networks includes: Build a summary network; Combining the first neural network, all the n-th neural networks and the summary network to obtain a combined network model, wherein the first node and the third node having an accuracy greater than or equal to a preset accuracy both transmit data to the summary network; The combined network model is trained to obtain a data prediction agent model.
8. A data prediction method, characterized in that: The data prediction method comprises: Obtain real-time operation data of construction machinery; The real-time operation data of the construction machinery is input into a data prediction agent model to obtain the predicted target data of the construction machinery, wherein the data prediction agent model is obtained according to the model training method according to any one of claims 1 to 7.
9. A computing device, characterized in that include: a memory configured to store instructions; A processor is configured to call the instructions from the memory and to implement the model training method according to any one of claims 1 to 7, or the data prediction method in claim 8 when executing the instructions.
10. A machine-readable storage medium, characterized in that: The machine-readable storage medium stores instructions for causing a machine to execute the model training method according to any one of claims 1 to 7, or the data prediction method according to claim 8.