CN-122023879-A - Method for estimating strength of connection firmness confirmation action of oil filling riser and related products

Abstract

The invention discloses a strength estimation method for confirming connection firmness of a crane tube and a related product, the method comprises the following steps of S1, obtaining a target area image time sequence, S2, inputting the target area image time sequence into a human body posture estimation model to be processed to obtain a human body skeleton data time sequence, S3, inputting the human body skeleton data time sequence into a pre-trained GT-ForceNet model to obtain pulling strength and pulling type of workers in each frame of target area image, wherein the GT-ForceNet model comprises a double-layer graph convolution network, a splicing layer, a flexible layer, an LSTM network, a full-connection layer group and a softmax layer which are sequentially connected. The invention can precisely quantify the applied force in the pulling process and can remarkably improve the safety and reliability of operation.

Inventors

• YE CHENG
• LIU XUEJUN
• LU XUEMEI
• YUE BIN
• LU YAWEI
• ZHANG BOJUN
• LI MENG
• WEI SONGJIE
• JI KANG
• ZHANG BINBIN
• CHEN CHAO
• FEI XUEZHI
• WANG ZHIRONG

Assignees

• 南京市锅炉压力容器检验研究院
• 南京理工大学
• 机械工业上海蓝亚石化设备检测所有限公司
• 南京工业大学
• 南京金创有色金属科技发展有限公司

Dates

Publication Date

20260512

Application Date

20251231

Claims (10)

1. 1. The method for estimating the strength of the connection firmness confirming action of the oil filling riser is characterized by comprising the following steps of: s1, acquiring a target area image time sequence, wherein the target area image time sequence comprises N frames of target area images; S2, inputting the target area image time sequence into a human body posture estimation model for processing to obtain a human skeleton data time sequence; and S3, inputting the human skeleton data time sequence into a pre-trained GT-ForceNet model to obtain the pulling force and the pulling type of workers in each frame of target area image, wherein the GT-ForceNet model comprises a double-layer graph convolution network, a splicing layer, a flame layer, an LSTM network, a full-connection layer group and a softmax layer which are sequentially connected, and the full-connection layer group comprises a first full-connection layer and a second full-connection layer which are parallel.
2. 2. The method for estimating the strength of the connection firmness confirming action of a crane tube according to claim 1, wherein S3 comprises the steps of: S31, the human skeleton data time sequence { is obtained Inputting into the double-layer graph rolling network to obtain a high-layer spatial characteristic time sequence { And (3) a process for preparing the same, wherein, Time series { representing the human skeletal data N-th frame of human skeletal data in }; time series { representing the high-level spatial features An nth frame high-level spatial feature in }; ; S32, characterizing the high-level space Velocity matrix of elbow-shoulder Acceleration matrix of elbow-shoulder And elbow-shoulder angular velocity matrix Inputting the splicing layer to perform splicing operation to obtain splicing characteristics ; S33, splicing the characteristic time sequence Sequentially processing the flat layer, the LSTM network and the first full-connection layer to obtain the pulling force of workers in each frame of target area image; Splice the characteristic time sequence Sequentially processing the flat layer, the LSTM network, the second full-connection layer and the softmax layer to obtain the pulling type of workers in each frame of target area image; Wherein, the Representing the temporal sequence of stitching features An nth frame splice feature of (a).
3. 3. The method for estimating the strength of the arm connection firmness confirming action according to claim 2, wherein S31 is implemented based on the following formula: When n=1: ; ); When (when) When (1): ; )); Wherein, the Time series { representing low-level spatial features An nth frame low-level spatial feature in }; time series { representing low-level spatial features N-1 th frame low-level spatial features in }; representing a graph convolutional network function; representing a splicing operation; Representing a contiguous matrix of predefined human skeletal maps.
4. 4. The method for estimating the strength of the arm connection firmness confirming action according to claim 2, wherein S31 is implemented based on the following formula: When (when) When (1): ; )); when n=n: ; ); Wherein, the Time series { representing low-level spatial features An nth frame low-level spatial feature in }; time series { representing low-level spatial features N+1th frame low-level spatial features in }; representing a graph convolutional network function; representing a splicing operation; Representing a contiguous matrix of predefined human skeletal maps.
5. 5. The method for estimating the strength of a connection reliability confirmation action of a crane tube according to claim 1, further comprising the steps of: And S4, optimizing and adjusting the pulling force of the workers in each frame of target area image to obtain the optimized pulling force of the workers in each frame of target area image.
6. 6. The method for estimating the strength of a connection firmness confirming action of a crane tube according to claim 5, wherein S4 is specifically implemented based on the following formula: ×( ); In the formula, Confidence representing the pull strength of the worker in the i-th frame target area image when In the time-course of which the first and second contact surfaces, The value is 0; representing the pulling force of workers in the n-th frame target area image obtained in the step S33; and (4) representing the optimized pulling force of the worker in the n-th frame target area image obtained in the step S4.
7. 7. A system for estimating the strength of a crane tube connection firmness confirming action, which is characterized by comprising an acquisition module, a human body posture estimation model and a pre-trained GT-ForceNet model, wherein the method for estimating the strength of the crane tube connection firmness confirming action is used for realizing the crane tube connection firmness confirming action according to any one of claims 1-6; The acquisition module is used for acquiring a target area image time sequence, wherein the target area image time sequence comprises N frames of target area images; the human body posture estimation model is used for extracting bone data of the target area image time sequence to obtain a human body bone data time sequence; The pre-trained GT-ForceNet model is used for processing the human skeleton data time sequence to obtain the pulling force and the pulling type of workers in each frame of target area image, wherein the GT-ForceNet model comprises a double-layer graph convolution network, a splicing layer, a flame layer, an LSTM network and a full connection layer which are connected in sequence.
8. 8. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method for estimating the strength of the arm connection firmness confirming action according to any of claims 1 to 6.
9. 9. A computer program product comprising a computer program which when executed by a processor implements a method of estimating the strength of a joint strength confirmation action of a crane pipe as claimed in any one of claims 1 to 6.
10. 10. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the method for estimating the strength of a joint robustness confirming action of a crane tube according to any one of claims 1 to 6.

Description

Method for estimating strength of connection firmness confirmation action of oil filling riser and related products Technical Field The invention relates to the technical field of filling of oil and gas tank trucks, in particular to a strength estimation method for confirming connection firmness of a loading arm and related products. Background At present, in the filling operation of an oil and gas tank truck, a worker usually intuitively checks the connection firmness of the oil and gas tank truck and a connecting port of the tank truck by manually pulling the oil filling riser; however, the method completely depends on personal experience and subjective judgment of workers, lacks objective and quantitative strength evaluation standards, is likely to have misjudgment risks, and cannot accurately feed back the real fastening degree of the connection in real time, so that great potential safety hazards such as oil gas leakage, fire disaster and the like are buried; therefore, how to provide a method for estimating the strength of the connection firmness of the oil filling riser, which can precisely quantify the strength applied during the pulling process, so as to remarkably improve the safety and reliability of the operation, is a problem to be solved by those skilled in the art. Disclosure of Invention In view of the foregoing, the present invention is directed to a method for estimating strength of a connection reliability confirmation of a crane tube and related products, which overcome or at least partially solve the foregoing problems. In order to achieve the above purpose, the present invention adopts the following technical scheme: In a first aspect, a method for estimating strength of a connection firmness confirmation action of a crane tube is provided, including the following steps: s1, acquiring a target area image time sequence, wherein the target area image time sequence comprises N frames of target area images; S2, inputting the target area image time sequence into a human body posture estimation model for processing to obtain a human skeleton data time sequence; and S3, inputting the human skeleton data time sequence into a pre-trained GT-ForceNet model to obtain the pulling force and the pulling type of workers in each frame of target area image, wherein the GT-ForceNet model comprises a double-layer graph convolution network, a splicing layer, a flame layer, an LSTM network, a full-connection layer group and a softmax layer which are sequentially connected, and the full-connection layer group comprises a first full-connection layer and a second full-connection layer which are parallel. Preferably, S3 specifically comprises the following steps: S31, the human skeleton data time sequence { is obtained Inputting into the double-layer graph rolling network to obtain a high-layer spatial characteristic time sequence {And (3) a process for preparing the same, wherein,Time series { representing the human skeletal dataN-th frame of human skeletal data in }; time series { representing the high-level spatial features An nth frame high-level spatial feature in };; S32, characterizing the high-level space Velocity matrix of elbow-shoulderAcceleration matrix of elbow-shoulderAnd elbow-shoulder angular velocity matrixInputting the splicing layer to perform splicing operation to obtain splicing characteristics; S33, splicing the characteristic time sequenceSequentially processing the flat layer, the LSTM network and the first full-connection layer to obtain the pulling force of workers in each frame of target area image; Splice the characteristic time sequence Sequentially processing the flat layer, the LSTM network, the second full-connection layer and the softmax layer to obtain the pulling type of workers in each frame of target area image; Wherein, the Representing the temporal sequence of stitching featuresAn nth frame splice feature of (a). Preferably, S31 is implemented based on the following formula: When n=1: ; ); When (when) When (1): ; )); Wherein, the Time series { representing low-level spatial featuresAn nth frame low-level spatial feature in }; time series { representing low-level spatial features N-1 th frame low-level spatial features in }; representing a graph convolutional network function; representing a splicing operation; Representing a contiguous matrix of predefined human skeletal maps. Preferably, S31 is implemented based on the following formula: When (when) When (1): ; )); when n=n: ; ); Wherein, the Time series { representing low-level spatial featuresAn nth frame low-level spatial feature in }; time series { representing low-level spatial features N+1th frame low-level spatial features in }; representing a graph convolutional network function; representing a splicing operation; Representing a contiguous matrix of predefined human skeletal maps. Preferably, the strength estimation method further includes the steps of: And S4, optimizing and adjusting the pulling force of