Portal crane stress inversion method
Through artificial intelligence and virtual and real data-driven methods, combined with deep learning technology, the problem of insufficient stress monitoring data of gantry cranes is solved, high accuracy inversion of stress data is achieved, and safety control capabilities are improved.
Patent Information
- Application Number
- CN202411309736.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-19
- Publication Date
- 2025-02-25
AI Technical Summary
The stress monitoring of gantry cranes faces the problem of insufficient data. It is difficult for traditional methods to comprehensively monitor the stress distribution of the entire equipment, and the stress data varies greatly under different working conditions, making it difficult to accurately reflect the actual stress state.
Using artificial intelligence and virtual and real data-driven methods, through the rational layout of stress sensors and the multi-dimensional fusion of virtual and real stress data, a dynamic reconstruction and first-level negative feedback deep learning unit is constructed to realize deep training of stress inversion of gantry cranes, and the method is verified and corrected through the second-level negative feedback unit.
It improves the confidence and safety control capabilities of stress inversion of gantry cranes, and can accurately invert the stress data of the entire equipment in terms of speed and accuracy, ensuring the accuracy of stress monitoring under different operating conditions.
Smart Images
Figure CN119514046A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of gantry cranes, and in particular relates to a stress inversion method for gantry cranes. Background Art
[0002] Gantry crane is one of the key equipments in hydropower station, which is mainly responsible for the lifting and lowering of dam gates. The mastery of the stress condition of its mechanical structure is directly related to the safe production and operation of hydropower station. However, due to the large structure of gantry crane and the high requirements for the operating environment, it is impossible to arrange enough stress sensors on it to obtain stress data. First, it is difficult to obtain stress data due to the cost. Second, the problems such as unstable structure, power supply and communication will interfere with or even endanger the normal operation of gantry crane.
[0003] Gantry cranes play a vital role in industrial production and logistics transportation. However, in actual application, the stress monitoring of gantry cranes faces a series of challenges. Traditional stress monitoring methods usually have shortcomings: traditional stress monitoring methods can only cover some key stress points, and it is difficult to fully monitor the stress distribution of the entire gantry crane; in actual projects, the stress data of gantry cranes under different working conditions may vary greatly, and it is difficult to accurately reflect the actual stress state by simply relying on measured data. Summary of the invention
[0004] The technical problem to be solved by the present invention is to provide a gantry crane stress inversion method, which solves the problem of insufficient gantry crane stress data through artificial intelligence and virtual-real data driving, and improves the gantry crane stress inversion confidence and safety management and control capabilities.
[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: A gantry crane stress inversion method, the steps are: Carry out reasonable layout of stress sensors and multi-dimensional fusion of virtual and real stress data; Taking virtual and real stress data as input, a dynamic reconstruction and first-level negative feedback deep learning unit is constructed to realize deep training of gantry crane stress inversion; Construct a two-level negative feedback unit for the measured stress, inversion stress, and deep learning unit of the gantry crane to realize the verification and correction of the inversion method based on the measured stress. Preferably, the reasonable layout of stress sensors and multi-dimensional fusion of virtual and real stress data include the following steps: Based on the finite element simulation tool and mechanical drawings, the key stress nodes of the gantry crane are analyzed and stress sensors are evenly arranged at and between the key stress nodes; The measured stress data of the gantry crane is obtained through the stress sensor, and a timestamp is added to each data. At the same time, based on the same working conditions and positions, the virtual stress data of the gantry crane generated by the finite element simulation tool is obtained, and a timestamp is added to each data. Under normal working conditions, for the key stress nodes of the gantry crane, the collected measured stress data and virtual stress data of the gantry crane are fused according to the order of timestamps. When the timestamps are consistent, the measured stress data are in the front, and finally fused into a set of virtual and real stress fusion data sets A of the key stress nodes of the gantry crane sorted by time; at the same time, a set of virtual and real stress fusion data sets A* of the non-key stress nodes of the gantry crane sorted by time are obtained according to the above method.
[0006] Preferably, under abnormal working conditions, the gantry crane is difficult to actually operate or there are operating risks, that is, the measured stress data cannot be obtained; at this time, the virtual stress data of the gantry crane generated by the finite element simulation tool is obtained, and a timestamp is added to each data; For the key stress nodes of the gantry crane, the obtained virtual stress data are sorted in the order of timestamps and finally fused into a set of virtual stress fusion data sets B of the key stress nodes of the gantry crane sorted by time; at the same time, a set of virtual stress fusion data sets B* of the non-key stress nodes of the gantry crane sorted by time are obtained according to the above method.
[0007] Preferably, the method of taking virtual and real stress data as input, constructing a dynamic reconstruction and first-level negative feedback deep learning unit, and realizing deep training of gantry crane stress inversion comprises the following steps: A deep learning unit for gantry crane stress inversion based on a deep learning algorithm is constructed, and datasets A and A*, datasets B and B* are respectively input into the deep learning unit. The nonlinear relationship between the key stress node data and the non-key stress node data of the gantry crane is established through deep training; at the same time, the output and input of the deep learning unit constitute a first-level negative feedback structure; If the nonlinear relationship between the key stress node data and the non-key stress node data of the gantry crane is not accurately established within the set time, the structural parameters of the gantry crane stress inversion deep learning unit are automatically adjusted dynamically based on the first-level negative feedback structure, and the "construction of a gantry crane stress inversion deep learning unit based on a deep learning algorithm, and input of data sets A and A*, data sets B and data sets B* into the deep learning unit respectively, and the establishment of a nonlinear relationship between the key stress node data and the non-key stress node data of the gantry crane through deep training; at the same time, the output and input of the deep learning unit constitute a first-level negative feedback structure" is re-executed.
[0008] Preferably, the construction of a secondary negative feedback unit for the measured stress, inversion stress, and deep learning unit of a gantry crane to implement the verification and correction of the inversion method based on the measured stress includes the following steps: The gantry crane stress inversion deep learning unit is deployed at the gantry crane operation site, and the measured stress of the gantry crane, the inversion stress, and the deep learning unit form a two-level negative feedback structure; the measured stress data of the key stress nodes of the gantry crane are used as the input of the unit, and the difference between its output and the measured stress data of the non-key stress nodes of the gantry crane is compared; When the difference is within the set threshold range, it indicates that the gantry crane stress inversion meets the requirements; otherwise, the deep learning unit structural parameters are automatically adjusted based on the secondary negative feedback structure, and the difference is compared again.
[0009] A gantry crane stress inversion system, comprising: Gantry crane stress data acquisition and fusion module: used to achieve reasonable layout of stress sensors and multi-dimensional fusion of virtual and real stress data; Gantry crane stress inversion deep learning module: used to construct dynamic reconstruction and first-level negative feedback deep learning units with virtual and real stress data as input, and realize deep training of gantry crane stress inversion; Gantry crane stress inversion verification module: used to construct a two-level negative feedback unit for the measured stress, inversion stress, and deep learning unit of the gantry crane, to realize the verification and correction of the inversion method based on the measured stress.
[0010] Preferably, the program instructions executed by the gantry crane stress data acquisition and fusion module are: based on the finite element simulation tool and the mechanical drawing analysis, the key stress nodes of the gantry crane are obtained, and stress sensors are evenly arranged at the key stress nodes and between the key stress nodes; The measured stress data of the gantry crane is obtained through the stress sensor, and a timestamp is added to each data. At the same time, based on the same working conditions and positions, the virtual stress data of the gantry crane generated by the finite element simulation tool is obtained, and a timestamp is added to each data. Under normal working conditions, for the key stress nodes of the gantry crane, the collected measured stress data and virtual stress data of the gantry crane are fused according to the order of timestamps. When the timestamps are consistent, the measured stress data is in the front, and finally fused into a set of virtual and real stress fusion data sets A of the key stress nodes of the gantry crane sorted by time; at the same time, a set of virtual and real stress fusion data sets A* of the non-key stress nodes of the gantry crane sorted by time is obtained according to the above method; Under abnormal working conditions, the gantry crane is difficult to actually operate or there are operating risks, that is, it is impossible to obtain the measured stress data; at this time, the virtual stress data of the gantry crane generated by the finite element simulation tool is obtained, and a timestamp is added to each data; For the key stress nodes of the gantry crane, the obtained virtual stress data are sorted in the order of timestamps and finally fused into a set of virtual stress fusion data sets B of the key stress nodes of the gantry crane sorted by time; at the same time, a set of virtual stress fusion data sets B* of the non-key stress nodes of the gantry crane sorted by time are obtained according to the above method.
[0011] Preferably, the program instructions executed by the gantry crane stress inversion deep learning module are: A deep learning unit for gantry crane stress inversion based on a deep learning algorithm is constructed, and datasets A and A*, datasets B and B* are respectively input into the deep learning unit. The nonlinear relationship between the key stress node data and the non-key stress node data of the gantry crane is established through deep training; at the same time, the output and input of the deep learning unit constitute a first-level negative feedback structure; If the nonlinear relationship between the key stress node data and the non-key stress node data of the gantry crane is not accurately established within the set time, the structural parameters of the gantry crane stress inversion deep learning unit are automatically adjusted dynamically based on the first-level negative feedback structure, and the "construction of a gantry crane stress inversion deep learning unit based on a deep learning algorithm, and input of data sets A and A*, data sets B and data sets B* into the deep learning unit respectively, and the establishment of a nonlinear relationship between the key stress node data and the non-key stress node data of the gantry crane through deep training; at the same time, the output and input of the deep learning unit constitute a first-level negative feedback structure" is re-executed.
[0012] Preferably, the program instructions executed by the gantry crane stress inversion verification module are: The gantry crane stress inversion deep learning unit is deployed at the gantry crane operation site, and the measured stress of the gantry crane, the inversion stress, and the deep learning unit form a two-level negative feedback structure; the measured stress data of the key stress nodes of the gantry crane are used as the input of the unit, and the difference between its output and the measured stress data of the non-key stress nodes of the gantry crane is compared; When the difference is within the set threshold range, it indicates that the gantry crane stress inversion meets the requirements; otherwise, the deep learning unit structural parameters are automatically adjusted based on the secondary negative feedback structure, and the difference is compared again.
[0013] A computer device comprising: one or more processors; The processor is used to store one or more programs; When the one or more programs are executed by the one or more processors, a gantry crane stress inversion method as described is implemented.
[0014] A computer-readable storage medium stores a computer program, which, when executed, implements a gantry crane stress inversion method as described above.
[0015] The present invention can achieve the following beneficial effects: 1. This method covers the design of gantry crane stress data acquisition and fusion module, gantry crane stress inversion deep learning module, and gantry crane stress inversion verification module. It can change the current problem of incomplete stress monitoring of gantry cranes to a certain extent and improve the confidence and safety management capabilities of gantry crane stress inversion.
[0016] 2. Based on the deep learning algorithm, a deep learning unit for gantry crane stress inversion is constructed, which can build a nonlinear relationship between the key stress node data and the non-key stress node data of the gantry crane in terms of speed and accuracy, so that the stress data of the entire gantry crane can be accurately inverted based on a small amount of measured data; 3. The constructed multi-level negative feedback structure can automatically adjust the structural parameters of the deep learning unit to ensure the correctness of the deep training direction, and verify the feasibility of the method based on actual operation data. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] The present invention will be further described below in conjunction with the accompanying drawings and embodiments: Figure 1 This is a structural block diagram of the present invention. DETAILED DESCRIPTION
[0018] The preferred solution is Figure 1 As shown in the figure, a gantry crane stress inversion method adopts the design of gantry crane stress data acquisition and fusion module, gantry crane stress inversion deep learning module and gantry crane stress inversion verification module. It can change the current problem of incomplete gantry crane stress monitoring to a certain extent and improve the gantry crane stress inversion confidence and safety management and control capabilities.
[0019] The structural block diagram of the present invention is as follows Figure 1 As shown, the specific implementation is as follows: (1) Figure 1 1 in the figure represents the gantry crane stress data acquisition and fusion module, which realizes the reasonable layout of stress sensors and the multi-dimensional fusion of virtual and real stress data. The specific implementation is as follows: ① Based on the finite element simulation tool and mechanical drawings, the key stress nodes of the gantry crane are analyzed and stress sensors are evenly arranged at and between the key stress nodes; ② Under normal working conditions, the measured stress data of the gantry crane is obtained through the stress sensor, and a timestamp is added to each data; at the same time, based on the same working conditions and positions, the virtual stress data of the gantry crane generated by the finite element simulation tool is obtained, and a timestamp is added to each data; ③ In case ②, for the key stress nodes of the gantry crane, the collected measured stress data and virtual stress data of the gantry crane are fused according to the order of timestamps. When the timestamps are consistent, the measured stress data is in the front, and finally fused into a set of virtual and real stress fusion data sets A of the key stress nodes of the gantry crane sorted by time; at the same time, a set of virtual and real stress fusion data sets A* of the non-key stress nodes of the gantry crane sorted by time is obtained according to the above method; ④ Under special working conditions, the gantry crane is difficult to operate or there are certain risks in operation, that is, it is impossible to obtain the measured stress data; at this time, the virtual stress data of the gantry crane generated by the finite element simulation tool is obtained, and a timestamp is added to each data; ⑤ In case ④, for the key stress nodes of the gantry crane, the obtained virtual stress data are sorted according to the order of timestamps, and finally fused into a set of virtual stress fusion data sets B of the key stress nodes of the gantry crane sorted by time; at the same time, a set of virtual stress fusion data sets B* of the non-key stress nodes of the gantry crane sorted by time is obtained according to the above method; (2) Figure 1 2 in the figure represents the deep learning module for gantry crane stress inversion. This module takes virtual and real stress data as input, builds a dynamic reconstruction and first-level negative feedback deep learning unit, and implements deep training for gantry crane stress inversion. The specific implementation is as follows: ① A deep learning unit for gantry crane stress inversion based on deep learning algorithm is constructed, and datasets A and A*, datasets B and B* are input into the deep learning unit respectively. The nonlinear relationship between the key stress node data and the non-key stress node data of the gantry crane is established through deep training; at the same time, the output and input of the deep learning unit constitute a first-level negative feedback structure; ② If the nonlinear relationship between the key stress node data and the non-key stress node data of the gantry crane is not accurately established within a certain period of time, the structural parameters of the gantry crane stress inversion deep learning unit in ① are automatically adjusted dynamically based on the first-level negative feedback structure, and ① is re-executed; the above-mentioned certain period of time and accuracy are user-configurable; (3) Figure 1 3 in the figure represents the gantry crane stress inversion verification module. This module constructs a two-level negative feedback unit for the gantry crane measured stress, inversion stress, and deep learning unit to realize the verification and correction of the inversion method based on the measured stress. The specific implementation is as follows: ① The gantry crane stress inversion deep learning unit obtained in step (2) is deployed at the gantry crane operation site, and the measured stress of the gantry crane, the inversion stress, and the deep learning unit form a two-level negative feedback structure; the measured stress data of the key stress nodes of the gantry crane is used as the input of the unit, and the difference between its output and the measured stress data of the non-key stress nodes of the gantry crane is compared; ② When the difference in ① is within a reasonable range, it indicates that the stress inversion of the gantry crane meets the requirements; otherwise, the deep learning unit structure parameters in ① are automatically adjusted based on the secondary negative feedback structure in ①, and ① is returned.
[0020] A computer device comprising: one or more processors; The processor is used to store one or more programs; When the one or more programs are executed by the one or more processors, a gantry crane stress inversion method as described is implemented.
[0021] A computer-readable storage medium stores a computer program, which, when executed, implements a gantry crane stress inversion method as described above.
[0022] The finite element simulation tools and deep learning algorithms in this method are selected according to user needs and actual conditions.
[0023] In summary, the present invention discloses a gantry crane stress inversion method, including the design of a gantry crane stress data acquisition and fusion module, a gantry crane stress inversion deep learning module, and a gantry crane stress inversion verification module. It can change the current problem of incomplete gantry crane stress monitoring to a certain extent and improve the gantry crane stress inversion confidence and safety management and control capabilities.
[0024] The above embodiments are only preferred technical solutions of the present invention and should not be regarded as limiting the present invention. The protection scope of the present invention shall be the technical solutions recorded in the claims, including equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, equivalent replacement improvements within this scope are also within the protection scope of the present invention.
Claims
1. A gantry crane stress inversion method, characterized in that The following steps are involved: Carry out reasonable layout of stress sensors and multi-dimensional fusion of virtual and real stress data; Taking virtual and real stress data as input, a dynamic reconstruction and first-level negative feedback deep learning unit is constructed to realize deep training of gantry crane stress inversion; Construct a two-level negative feedback unit for the measured stress, inversion stress, and deep learning unit of the gantry crane to realize the verification and correction of the inversion method based on the measured stress.
2. A gantry crane stress inversion method according to claim 1, characterized in that: The reasonable layout of stress sensors and multi-dimensional fusion of virtual and real stress data include the following steps: Based on the finite element simulation tool and mechanical drawings, the key stress nodes of the gantry crane are analyzed and stress sensors are evenly arranged at and between the key stress nodes; The measured stress data of the gantry crane is obtained through the stress sensor, and a timestamp is added to each data. At the same time, based on the same working conditions and positions, the virtual stress data of the gantry crane generated by the finite element simulation tool is obtained, and a timestamp is added to each data. Under normal working conditions, for the key stress nodes of the gantry crane, the collected measured stress data and virtual stress data of the gantry crane are fused according to the order of timestamps. When the timestamps are consistent, the measured stress data are in the front, and finally fused into a set of virtual and real stress fusion data sets A of the key stress nodes of the gantry crane sorted by time; at the same time, a set of virtual and real stress fusion data sets A* of the non-key stress nodes of the gantry crane sorted by time are obtained according to the above method.
3. A gantry crane stress inversion method according to claim 2, characterized in that: Under abnormal working conditions, the gantry crane is difficult to actually operate or there are operational risks, that is, it is impossible to obtain the measured stress data; At this time, the virtual stress data of the gantry crane generated by the finite element simulation tool is obtained, and a timestamp is added to each data; For the key stress nodes of the gantry crane, the obtained virtual stress data are sorted in the order of timestamps and finally fused into a set of virtual stress fusion data sets B of the key stress nodes of the gantry crane sorted by time; at the same time, a set of virtual stress fusion data sets B* of the non-key stress nodes of the gantry crane sorted by time are obtained according to the above method.
4. A gantry crane stress inversion method according to claim 3, characterized in that: The method of taking virtual and real stress data as input, constructing a dynamic reconstruction and first-level negative feedback deep learning unit, and realizing deep training of gantry crane stress inversion includes the following steps: A deep learning unit for gantry crane stress inversion based on a deep learning algorithm is constructed, and datasets A and A*, datasets B and B* are respectively input into the deep learning unit. The nonlinear relationship between the key stress node data and the non-key stress node data of the gantry crane is established through deep training; at the same time, the output and input of the deep learning unit constitute a first-level negative feedback structure; If the nonlinear relationship between the key stress node data and the non-key stress node data of the gantry crane is not accurately established within the set time, the structural parameters of the gantry crane stress inversion deep learning unit are automatically adjusted dynamically based on the primary negative feedback structure, and the "construction of a gantry crane stress inversion deep learning unit based on a deep learning algorithm, and input of data sets A and A*, data sets B and data sets B* into the deep learning unit respectively, and the nonlinear relationship between the key stress node data and the non-key stress node data of the gantry crane is established through deep training; at the same time, the output and input of the deep learning unit constitute a primary negative feedback structure" is re-executed.
5. A gantry crane stress inversion method according to claim 4, characterized in that: The construction of the secondary negative feedback unit for the measured stress, inversion stress, and deep learning unit of the gantry crane to realize the verification and correction of the inversion method based on the measured stress includes the following steps: The gantry crane stress inversion deep learning unit is deployed at the gantry crane operation site, and the measured stress of the gantry crane, the inversion stress, and the deep learning unit form a two-level negative feedback structure; the measured stress data of the key stress nodes of the gantry crane are used as the input of the unit, and the difference between its output and the measured stress data of the non-key stress nodes of the gantry crane is compared; When the difference is within the set threshold range, it indicates that the gantry crane stress inversion meets the requirements; otherwise, the deep learning unit structural parameters are automatically adjusted based on the secondary negative feedback structure, and the difference is compared again.
6. A gantry crane stress inversion system, characterized in that: include: Gantry crane stress data acquisition and fusion module: used to achieve reasonable layout of stress sensors and multi-dimensional fusion of virtual and real stress data; Gantry crane stress inversion deep learning module: used to construct dynamic reconstruction and first-level negative feedback deep learning units with virtual and real stress data as input, and realize deep training of gantry crane stress inversion; Gantry crane stress inversion verification module: used to construct a two-level negative feedback unit for the measured stress, inversion stress, and deep learning unit of the gantry crane, to realize the verification and correction of the inversion method based on the measured stress.
7. A gantry crane stress inversion system according to claim 6, characterized in that: The program instructions executed by the gantry crane stress data acquisition and fusion module are: based on the finite element simulation tool and mechanical drawings, the key stress nodes of the gantry crane are analyzed and stress sensors are evenly arranged at and between the key stress nodes; The measured stress data of the gantry crane is obtained through the stress sensor, and a timestamp is added to each data. At the same time, based on the same working conditions and positions, the virtual stress data of the gantry crane generated by the finite element simulation tool is obtained, and a timestamp is added to each data. Under normal working conditions, for the key stress nodes of the gantry crane, the collected measured stress data and virtual stress data of the gantry crane are fused according to the order of timestamps. When the timestamps are consistent, the measured stress data is in the front, and finally fused into a set of virtual and real stress fusion data sets A of the key stress nodes of the gantry crane sorted by time; at the same time, a set of virtual and real stress fusion data sets A* of the non-key stress nodes of the gantry crane sorted by time is obtained according to the above method; Under abnormal working conditions, the gantry crane is difficult to actually operate or there are operational risks, that is, it is impossible to obtain the measured stress data; At this time, the virtual stress data of the gantry crane generated by the finite element simulation tool is obtained, and a timestamp is added to each data; For the key stress nodes of the gantry crane, the obtained virtual stress data are sorted in the order of timestamps and finally fused into a set of virtual stress fusion data sets B of the key stress nodes of the gantry crane sorted by time; at the same time, a set of virtual stress fusion data sets B* of the non-key stress nodes of the gantry crane sorted by time are obtained according to the above method.
8. A gantry crane stress inversion system according to claim 7, characterized in that: The program instructions executed by the gantry crane stress inversion deep learning module are: A deep learning unit for gantry crane stress inversion based on a deep learning algorithm is constructed, and datasets A and A*, datasets B and B* are respectively input into the deep learning unit. The nonlinear relationship between the key stress node data and the non-key stress node data of the gantry crane is established through deep training; at the same time, the output and input of the deep learning unit constitute a first-level negative feedback structure; If the nonlinear relationship between the key stress node data and the non-key stress node data of the gantry crane is not accurately established within the set time, the structural parameters of the gantry crane stress inversion deep learning unit are automatically adjusted dynamically based on the primary negative feedback structure, and the "construction of a gantry crane stress inversion deep learning unit based on a deep learning algorithm, and input of data sets A and A*, data sets B and data sets B* into the deep learning unit respectively, and the nonlinear relationship between the key stress node data and the non-key stress node data of the gantry crane is established through deep training; at the same time, the output and input of the deep learning unit constitute a primary negative feedback structure" is re-executed.
9. A gantry crane stress inversion system according to claim 8, characterized in that: The program instructions executed by the gantry crane stress inversion verification module are: The gantry crane stress inversion deep learning unit is deployed at the gantry crane operation site, and the measured stress of the gantry crane, the inversion stress, and the deep learning unit form a two-level negative feedback structure; the measured stress data of the key stress nodes of the gantry crane are used as the input of the unit, and the difference between its output and the measured stress data of the non-key stress nodes of the gantry crane is compared; When the difference is within the set threshold range, it indicates that the gantry crane stress inversion meets the requirements; otherwise, the deep learning unit structural parameters are automatically adjusted based on the secondary negative feedback structure, and the difference is compared again.
10. A computer device, characterized in that: include: one or more processors; The processor is used to store one or more programs; When the one or more programs are executed by the one or more processors, a gantry crane stress inversion method as described in any one of claims 1 to 5 is implemented.
11. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed, a gantry crane stress inversion method as described in any one of claims 1 to 4 is implemented.