Crane monitoring method based on artificial intelligence

Abstract

The invention discloses a crane monitoring method based on artificial intelligence, and belongs to the technical field of crane monitoring. The crane abnormal operation analysis model is adopted to learn the crane historical operation images and the operator behavior labels corresponding to the crane historical operation images, and the learned crane abnormal operation analysis model is obtained; and then the state of the operator is monitored by adopting the learned abnormal operation analysis model of the crane, so that the illegal behavior of the operator can be effectively detected, and the safety in the running process of the crane is greatly improved.

Description

Technical Field

[0001] The invention belongs to the technical field of crane monitoring, and in particular relates to a crane monitoring method based on artificial intelligence. Background Art

[0002] With the rapid development of infrastructure construction in my country, cranes are increasingly used in construction, transportation, energy and other fields. However, cranes are prone to malfunctions during long-term, high-load operation, leading to safety accidents. Traditional crane monitoring systems mainly rely on manual inspections and sensor monitoring, but existing crane monitoring technologies mainly focus on monitoring the operating status of the equipment itself, while less attention is paid to the behavior management of operators and the safety management of the working environment. Improper operation of operators and environmental factors may lead to crane accidents. Therefore, it is necessary to study a method for effective management of operators. Summary of the invention

[0003] The present invention provides a crane monitoring method based on artificial intelligence, which is used to solve the problem that the prior art only monitors the operating status of the equipment but less monitors the operators, resulting in crane accidents.

[0004] A crane monitoring method based on artificial intelligence, comprising:

[0005] Acquire the crane historical operation images and the operator behavior labels corresponding to each crane historical operation image; wherein the operator behavior labels include the categories of normal operator operation or abnormal operator behavior;

[0006] An artificial intelligence algorithm is used to build a crane abnormal operation analysis model, and the crane abnormal operation analysis model is used to learn the crane historical operation images and the operator behavior labels corresponding to each crane historical operation image to obtain the crane abnormal operation analysis model after learning;

[0007] Acquire a real-time operation image of the crane, and use the learned abnormal operation analysis model of the crane to analyze the real-time operation image of the crane to obtain a crane operation monitoring result; wherein the crane operation monitoring result includes a category of normal operation of the operator or abnormal behavior of the operator;

[0008] When the crane operation monitoring result is an abnormal operator behavior category, a crane safety operation warning is issued to complete the crane monitoring based on artificial intelligence.

[0009] Furthermore, the crane historical operation images and the operator behavior labels corresponding to each crane historical operation image are obtained, including:

[0010] Acquire historical operation images of the crane, and display the historical operation images of the crane to staff;

[0011] The operator's abnormal behavior category fed back by the staff through human-computer interaction is obtained; wherein the operator's abnormal behavior category includes the operator's normal operation, the operator's fatigued operation, the operator answering the phone or the operator smoking.

[0012] Furthermore, an artificial intelligence algorithm is used to construct a crane abnormal operation analysis model, including: using a CNN algorithm to construct a crane abnormal operation analysis model.

[0013] Furthermore, the crane abnormal operation analysis model is used to learn the crane historical operation images and the operator behavior labels corresponding to each crane historical operation image, and the learned crane abnormal operation analysis model is obtained, including:

[0014] The hyperparameters of the crane abnormal operation analysis model are randomly initialized, and the hyperparameter vector after initialization is a vector to obtain a parameter vector, and multiple parameter vectors are repeatedly obtained;

[0015] The crane abnormal operation analysis model is used to obtain the error value corresponding to each parameter vector from the crane historical operation image and the operator behavior label corresponding to each crane historical operation image;

[0016] Determine the optimal parameter vector according to the error value corresponding to each parameter vector;

[0017] For the optimal parameter vector, perform greedy neighborhood exploration on the optimal parameter vector to obtain the optimal parameter vector after neighborhood exploration;

[0018] For other parameter vectors except the optimal parameter vector, the cosine exploration strategy is used to make all parameter vectors jointly explore around the optimal parameter vector to obtain the parameter vector after joint exploration;

[0019] For the parameter vector after joint exploration, an interpolation exploration strategy that weakens with the number of learning times is used to perform multi-level exploration on the parameter vector to obtain the parameter vector after multi-level exploration;

[0020] For the parameter vector after multi-level exploration, the mutation greedy exploration strategy is used to perform mutation exploration on the parameter vector to obtain the parameter vector after mutation exploration;

[0021] Determine whether the learning end conditions are met. If so, re-determine the optimal parameter vector based on the parameter vector after mutation exploration, and decode the optimal parameter vector into model parameters of the crane abnormal operation analysis model. Apply the decoded model parameters to the crane abnormal operation analysis model to obtain the crane abnormal operation analysis model after learning. Otherwise, return to the step of obtaining the error value corresponding to each parameter vector.

[0022] Furthermore, for the optimal parameter vector, a neighborhood greedy exploration is performed on the optimal parameter vector to obtain the optimal parameter vector after the neighborhood exploration, including:

[0023]

[0024] In the formula, represents the updated value of the optimal parameter vector in the tth learning process, represents the optimal parameter vector in the tth learning process, step1 represents the first step length factor, η1 represents the first convergence factor, iter represents the maximum number of learning times, η2 represents the second convergence factor, η1 and η2 are both random numbers uniformly distributed between (0,1), and X max represents the upper parameter limit vector, X min represents the parameter lower limit vector;

[0025] It is determined whether the error value corresponding to the updated value of the optimal parameter vector is reduced. If so, the updated value of the optimal parameter vector is used as the optimal parameter vector after neighborhood exploration. Otherwise, the original optimal parameter vector is used as the optimal parameter vector after neighborhood exploration.

[0026] Furthermore, for other parameter vectors except the optimal parameter vector, the cosine exploration strategy is used to make all parameter vectors jointly explore around the optimal parameter vector, and the parameter vector after joint exploration is obtained, including:

[0027]

[0028] in, represents the mth parameter vector in the tth learning process, Represents the parameter vector after joint exploration m=1,2,…,M, M represents the total number of other parameter vectors, C represents the step size control factor, π represents the circumference of a circle, and R1 represents a random number between (-1,1); the step size control factor is: C0 represents the maximum step size, e represents a natural constant, and iter represents the maximum number of learning times.

[0029] Furthermore, for the parameter vector after the joint exploration, an interpolation exploration strategy that weakens with the number of learning times is used to perform multi-level exploration on the parameter vector, and the parameter vector after the multi-level exploration is obtained, including:

[0030] Obtain the error value of the parameter vector after each joint exploration, and arrange them in ascending order of error value to obtain the arranged parameter vector;

[0031] According to the parameter vector after arrangement, multi-level exploration is performed on the parameter vector, and the parameter vector after multi-level exploration is obtained as follows:

[0032]

[0033] In the formula, represents the nth parameter vector in the tth learning process, n=1,2,3,…,N, N represents the total number of parameter vectors, Represents the parameter vector after multi-level exploration represents the nth parameter vector in the t-1th learning process, represents the nth parameter vector in the t-2th learning process, represents the nth parameter vector in the t-3th learning process, represents the n-1th parameter vector in the tth learning process, and when n=1, represents the optimal parameter vector, v represents the updated control variable, and

[0034] Furthermore, for the parameter vector after multi-level exploration, a mutation greedy exploration strategy is used to perform mutation exploration on the parameter vector to obtain the parameter vector after mutation exploration, including:

[0035] For the parameter vector after multi-level exploration, determine the mutation value corresponding to each parameter vector as follows:

[0036]

[0037] in, represents the parameter vector after the k-th multi-level exploration in the t-th learning process, represents the variation value corresponding to the parameter vector after the k-th multi-level exploration, α represents the proportional factor, φ represents the intermediate parameter; μ represents a random number, and μ~N(0,1) obeys the normal distribution, N represents the normal distribution; υ represents a random number, and υ~N(0,1); β represents a random number between (0,2), represents the center of all parameter vectors, γ represents the standard gamma function, and π represents pi.

[0038] Determine whether the error value corresponding to the variation value corresponding to the parameter vector is reduced. If so, the variation value corresponding to the parameter vector is used as the parameter vector after variation exploration. Otherwise, the original parameter vector after multi-level exploration is directly used as the parameter vector after variation exploration.

[0039] Further, judging whether the learning end condition is met includes:

[0040] Determine whether the current learning times reaches the maximum learning times or whether the error value of the optimal parameter encoding is less than a preset threshold. If so, determine that the learning end condition is met, otherwise determine that the learning end condition is not met.

[0041] Furthermore, when the crane operation monitoring result is an abnormal operator behavior category, a crane safety operation warning is issued, including:

[0042] When the crane operation monitoring result is an abnormal operator behavior category, a text warning message is generated and the text warning message is transmitted to a device designated by the staff or an audible and visual alarm is issued through an audible and visual alarm device installed at the crane site.

[0043] The present invention provides an artificial intelligence-based crane monitoring method, which constructs a crane abnormal operation analysis model through an artificial intelligence algorithm, and uses the crane abnormal operation analysis model to learn the crane historical operation image and the operator behavior label corresponding to each crane historical operation image to obtain the learned crane abnormal operation analysis model, and then uses the learned crane abnormal operation analysis model to monitor the operator's status, which can effectively detect the operator's illegal behavior and greatly improve the safety of the crane operation process. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0045] Figure 1 A flowchart of a crane monitoring method based on artificial intelligence is provided in an embodiment of the present invention.

[0046] The above drawings have shown clear embodiments of the present invention, which will be described in more detail below. These drawings and text descriptions are not intended to limit the scope of the present invention in any way, but to illustrate the concept of the present invention for those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0047] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of devices and methods consistent with some aspects of the present invention as detailed in the appended claims.

[0048] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0049] like Figure 1 As shown, an embodiment of the present invention provides a crane monitoring method based on artificial intelligence, comprising:

[0050] S1. Acquire historical operation images of the crane and operator behavior labels corresponding to each historical operation image of the crane; wherein the operator behavior labels include normal operator operation or abnormal operator behavior categories;

[0051] The historical operation images of the crane may include the head images of the operator in a normal state, the head images under fatigue operation, the head images when making a phone call, and the head images when smoking. The staff will then label these images with corresponding category labels, so as to obtain sample data for training artificial intelligence. It is worth noting that the head image here is only an example of an embodiment of the present invention, and upper body images can also be collected to analyze more abnormal behavior categories.

[0052] S2. Using an artificial intelligence algorithm to construct a crane abnormal operation analysis model, and using the crane abnormal operation analysis model to learn the crane historical operation images and the operator behavior labels corresponding to each crane historical operation image, to obtain the crane abnormal operation analysis model after learning;

[0053] Artificial intelligence algorithms such as CNN algorithms can be used to build a crane abnormal operation analysis model. After the crane historical operation images and the operator behavior labels corresponding to each crane historical operation image are used to train the crane abnormal operation analysis model, the crane abnormal operation analysis model will have the ability to identify abnormal behavior.

[0054] S3, obtaining a real-time operation image of the crane, and using the learned crane abnormal operation analysis model to analyze the real-time operation image of the crane to obtain a crane operation monitoring result; wherein the crane operation monitoring result includes a category of normal operation of the operator or abnormal behavior of the operator;

[0055] For example, a camera can be set in front of the crane driving position to collect the upper body image including the driver's head area. Then, the upper body image can be analyzed through the learned crane abnormal operation analysis model to obtain the crane operation monitoring result, thereby realizing the monitoring of the driver's status.

[0056] S4. When the crane operation monitoring result is an abnormal operator behavior category, a crane safety operation warning is issued to complete the crane monitoring based on artificial intelligence.

[0057] Crane safe operation warning can be text message warning, sound warning, light warning, sound and light warning and other warning methods, so that the staff can manage the drivers.

[0058] The present invention provides an artificial intelligence-based crane monitoring method, which constructs a crane abnormal operation analysis model through an artificial intelligence algorithm, and uses the crane abnormal operation analysis model to learn the crane historical operation image and the operator behavior label corresponding to each crane historical operation image to obtain the learned crane abnormal operation analysis model, and then uses the learned crane abnormal operation analysis model to monitor the operator's status, which can effectively detect the operator's illegal behavior and greatly improve the safety of the crane operation process.

[0059] In an embodiment of the present invention, obtaining a crane historical operation image and an operator behavior label corresponding to each crane historical operation image includes:

[0060] Acquire historical operation images of the crane, and display the historical operation images of the crane to staff;

[0061] The operator's abnormal behavior category fed back by the staff through human-computer interaction is obtained; wherein the operator's abnormal behavior category includes the operator's normal operation, the operator's fatigued operation, the operator answering the phone or the operator smoking.

[0062] It is worth noting that the above-mentioned operator abnormal behavior categories are only preferred implementations of the embodiments of the present invention, and other abnormal behaviors may also be set and then detected.

[0063] In the embodiment of the present invention, an artificial intelligence algorithm is used to construct a crane abnormal operation analysis model, including: using a CNN (Convolutional Neural Network) algorithm to construct a crane abnormal operation analysis model. Other artificial intelligence algorithms can also be used to identify and analyze the driver's image, which can also realize abnormal operation analysis.

[0064] In the prior art, the hyperparameter optimization of the artificial intelligence algorithm is generally performed by using an intelligent optimization algorithm such as a particle swarm algorithm. Although this prior art has a faster convergence speed, it is very easy to fall into a local optimal value, resulting in poor analysis accuracy of the final crane abnormal operation analysis model. Therefore, an embodiment of the present invention provides a new learning optimization algorithm to improve the abnormal behavior analysis capability.

[0065] In an embodiment of the present invention, a crane abnormal operation analysis model is used to learn the crane historical operation images and the operator behavior labels corresponding to each crane historical operation image to obtain the crane abnormal operation analysis model after learning, including:

[0066] The hyperparameters of the crane abnormal operation analysis model are randomly initialized, and the hyperparameter vector after initialization is a vector to obtain a parameter vector, and multiple parameter vectors are repeatedly obtained;

[0067] For example, the hyperparameters of the crane abnormal operation analysis model have upper and lower limits, so they can be randomly initialized within the upper and lower limits of the hyperparameters.

[0068] The crane abnormal operation analysis model is used to obtain the error value corresponding to each parameter vector from the crane historical operation image and the operator behavior label corresponding to each crane historical operation image; for example, the root mean square error value can be obtained.

[0069] Determine the optimal parameter vector according to the error value corresponding to each parameter vector;

[0070] For the optimal parameter vector, perform greedy neighborhood exploration on the optimal parameter vector to obtain the optimal parameter vector after neighborhood exploration;

[0071] For other parameter vectors except the optimal parameter vector, the cosine exploration strategy is used to make all parameter vectors jointly explore around the optimal parameter vector to obtain the parameter vector after joint exploration;

[0072] For the parameter vector after joint exploration, an interpolation exploration strategy that weakens with the number of learning times is used to perform multi-level exploration on the parameter vector to obtain the parameter vector after multi-level exploration;

[0073] For the parameter vector after multi-level exploration, the mutation greedy exploration strategy is used to perform mutation exploration on the parameter vector to obtain the parameter vector after mutation exploration;

[0074] Determine whether the learning end conditions are met. If so, re-determine the optimal parameter vector based on the parameter vector after mutation exploration, and decode the optimal parameter vector into model parameters of the crane abnormal operation analysis model. Apply the decoded model parameters to the crane abnormal operation analysis model to obtain the crane abnormal operation analysis model after learning. Otherwise, return to the step of obtaining the error value corresponding to each parameter vector.

[0075] In the embodiment of the present invention, for the optimal parameter vector, performing neighborhood greedy exploration on the optimal parameter vector to obtain the optimal parameter vector after the neighborhood exploration includes:

[0076]

[0077] In the formula, represents the updated value of the optimal parameter vector in the tth learning process, represents the optimal parameter vector in the tth learning process, step1 represents the first step length factor, η1 represents the first convergence factor, iter represents the maximum number of learning times, η2 represents the second convergence factor, η1 and η2 are both random numbers uniformly distributed between (0,1), and X max represents the upper parameter limit vector, X min represents the parameter lower limit vector;

[0078] It is determined whether the error value corresponding to the updated value of the optimal parameter vector is reduced. If so, the updated value of the optimal parameter vector is used as the optimal parameter vector after neighborhood exploration. Otherwise, the original optimal parameter vector is used as the optimal parameter vector after neighborhood exploration.

[0079] The greedy exploration of the neighborhood of the optimal parameter vector can make the optimal parameter vector always explore in the optimal area, which can effectively ensure the search efficiency of the algorithm.

[0080] In the embodiment of the present invention, for other parameter vectors except the optimal parameter vector, a cosine exploration strategy is adopted to make all parameter vectors jointly explore around the optimal parameter vector, and obtain the parameter vector after the joint exploration, including:

[0081]

[0082] in, represents the mth parameter vector in the tth learning process, Represents the parameter vector after joint exploration m=1,2,…,M, M represents the total number of other parameter vectors, C represents the step size control factor, π represents the circumference of a circle, and R1 represents a random number between (-1,1); the step size control factor is: C0 represents the maximum step size, e represents a natural constant, and iter represents the maximum number of learning times.

[0083] This joint exploration allows all parameter vectors to perform spiral wave search around the optimal position, which can effectively improve the exploration efficiency and accuracy of the algorithm.

[0084] In the embodiment of the present invention, for the parameter vector after the joint exploration, an interpolation exploration strategy that weakens with the number of learning times is used to perform multi-level exploration on the parameter vector, and the parameter vector after the multi-level exploration is obtained, including:

[0085] Obtain the error value of the parameter vector after each joint exploration, and arrange them in ascending order of error value to obtain the arranged parameter vector;

[0086] According to the parameter vector after arrangement, multi-level exploration is performed on the parameter vector, and the parameter vector after multi-level exploration is obtained as follows:

[0087]

[0088] In the formula, represents the nth parameter vector in the tth learning process, n=1,2,3,…,N, N represents the total number of parameter vectors, Represents the parameter vector after multi-level exploration represents the nth parameter vector in the t-1th learning process, represents the nth parameter vector in the t-2th learning process, represents the nth parameter vector in the t-3th learning process, represents the n-1th parameter vector in the tth learning process, and when n=1, represents the optimal parameter vector, v represents the updated control variable, and

[0089] This multi-level exploration integrates multiple historical position information, allowing the parameter vector to search based on the historical position, avoiding repeated searches, exploring more areas, and moving toward better surrounding positions, thereby increasing the possibility of exploring the global optimal solution.

[0090] In the embodiment of the present invention, for the parameter vector after the multi-level exploration, a mutation greedy exploration strategy is used to perform mutation exploration on the parameter vector to obtain the parameter vector after the mutation exploration, including:

[0091] For the parameter vector after multi-level exploration, determine the mutation value corresponding to each parameter vector as follows:

[0092]

[0093] in, represents the parameter vector after the k-th multi-level exploration in the t-th learning process, represents the variation value corresponding to the parameter vector after the k-th multi-level exploration, α represents the proportional factor, φ represents the intermediate parameter; μ represents a random number, and μ~N(0,1) obeys the normal distribution, N represents the normal distribution; υ represents a random number, and υ~N(0,1); β represents a random number between (0,2), represents the center of all parameter vectors, γ represents the standard gamma function, and π represents pi;

[0094] Determine whether the error value corresponding to the variation value corresponding to the parameter vector is reduced. If so, the variation value corresponding to the parameter vector is used as the parameter vector after variation exploration. Otherwise, the original parameter vector after multi-level exploration is directly used as the parameter vector after variation exploration.

[0095] This mutation exploration performs mutation based on the center of all parameter vectors (i.e., the vector composed of the mean of each dimension), which is conducive to maintaining population diversity and ensuring global search capabilities.

[0096] Optionally, a return simulation algorithm can be used to control the variation exploration to avoid the defects of the greedy strategy, and the out-of-bounds processing of the parameter vector after each search can also be performed.

[0097] In the embodiment of the present invention, determining whether the learning end condition is met includes:

[0098] Determine whether the current learning times reaches the maximum learning times or whether the error value of the optimal parameter encoding is less than a preset threshold. If so, determine that the learning end condition is met, otherwise determine that the learning end condition is not met.

[0099] In an embodiment of the present invention, when the crane operation monitoring result is an abnormal behavior category of the operator, a crane safety operation warning is performed, including:

[0100] When the crane operation monitoring result is an abnormal operator behavior category, a text warning message is generated and the text warning message is transmitted to a device designated by the staff or an audible and visual alarm is issued through an audible and visual alarm device installed at the crane site.

[0101] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take 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.) containing computer-usable program code.

[0102] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks 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 produce 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.

[0103] 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.

[0104] 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.

[0105] A person of ordinary skill in the art can understand that all or part of the steps in realizing the above-mentioned facts and methods can be completed by instructing the relevant hardware through a program, and the program involved or the program can be stored in a computer-readable storage medium. When the program is executed, it includes the following steps: At this time, the corresponding method steps are derived, and the storage medium can be ROM / RAM, a disk, an optical disk, etc.

[0106] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A crane monitoring method based on artificial intelligence, characterized in that: include: Acquire the crane historical operation images and the operator behavior labels corresponding to each crane historical operation image; wherein the operator behavior labels include the categories of normal operator operation or abnormal operator behavior; An artificial intelligence algorithm is used to build a crane abnormal operation analysis model, and the crane abnormal operation analysis model is used to learn the crane historical operation images and the operator behavior labels corresponding to each crane historical operation image to obtain the crane abnormal operation analysis model after learning; Acquire a real-time operation image of the crane, and use the learned abnormal operation analysis model of the crane to analyze the real-time operation image of the crane to obtain a crane operation monitoring result; wherein the crane operation monitoring result includes a category of normal operation of the operator or abnormal behavior of the operator; When the crane operation monitoring result is an abnormal operator behavior category, a crane safety operation warning is issued to complete the crane monitoring based on artificial intelligence.

2. The crane monitoring method based on artificial intelligence according to claim 1 is characterized in that: Obtain the crane's historical operation images and the operator behavior labels corresponding to each crane's historical operation image, including: Acquire historical operation images of the crane, and display the historical operation images of the crane to staff; The operator's abnormal behavior category fed back by the staff through human-computer interaction is obtained; wherein the operator's abnormal behavior category includes the operator's normal operation, the operator's fatigued operation, the operator answering the phone or the operator smoking.

3. The crane monitoring method based on artificial intelligence according to claim 1 is characterized in that: An artificial intelligence algorithm is used to construct a crane abnormal operation analysis model, including: using a CNN algorithm to construct a crane abnormal operation analysis model.

4. The crane monitoring method based on artificial intelligence according to claim 1 is characterized in that: The crane abnormal operation analysis model is used to learn the crane historical operation images and the operator behavior labels corresponding to each crane historical operation image, and the learned crane abnormal operation analysis model is obtained, including: The hyperparameters of the crane abnormal operation analysis model are randomly initialized, and the hyperparameter vector after initialization is a vector to obtain a parameter vector, and multiple parameter vectors are repeatedly obtained; The crane abnormal operation analysis model is used to obtain the error value corresponding to each parameter vector from the crane historical operation image and the operator behavior label corresponding to each crane historical operation image; Determine the optimal parameter vector according to the error value corresponding to each parameter vector; For the optimal parameter vector, perform greedy neighborhood exploration on the optimal parameter vector to obtain the optimal parameter vector after neighborhood exploration; For other parameter vectors except the optimal parameter vector, the cosine exploration strategy is used to make all parameter vectors jointly explore around the optimal parameter vector to obtain the parameter vector after joint exploration; For the parameter vector after joint exploration, an interpolation exploration strategy that weakens with the number of learning times is used to perform multi-level exploration on the parameter vector to obtain the parameter vector after multi-level exploration; For the parameter vector after multi-level exploration, the mutation greedy exploration strategy is used to perform mutation exploration on the parameter vector to obtain the parameter vector after mutation exploration; Determine whether the learning end conditions are met. If so, re-determine the optimal parameter vector based on the parameter vector after mutation exploration, and decode the optimal parameter vector into model parameters of the crane abnormal operation analysis model. Apply the decoded model parameters to the crane abnormal operation analysis model to obtain the crane abnormal operation analysis model after learning. Otherwise, return to the step of obtaining the error value corresponding to each parameter vector.

5. The crane monitoring method based on artificial intelligence according to claim 4 is characterized in that: For the optimal parameter vector, perform greedy neighborhood exploration on the optimal parameter vector to obtain the optimal parameter vector after neighborhood exploration, including: In the formula, represents the updated value of the optimal parameter vector in the tth learning process, represents the optimal parameter vector in the tth learning process, step1 represents the first step length factor, η1 represents the first convergence factor, iter represents the maximum number of learning times, η2 represents the second convergence factor, η1 and η2 are both random numbers uniformly distributed between (0,1), and X max represents the upper parameter limit vector, X min represents the parameter lower limit vector; It is determined whether the error value corresponding to the updated value of the optimal parameter vector is reduced. If so, the updated value of the optimal parameter vector is used as the optimal parameter vector after neighborhood exploration. Otherwise, the original optimal parameter vector is used as the optimal parameter vector after neighborhood exploration.

6. The crane monitoring method based on artificial intelligence according to claim 5 is characterized in that: For other parameter vectors except the optimal parameter vector, the cosine exploration strategy is used to make all parameter vectors jointly explore around the optimal parameter vector, and the parameter vector after joint exploration is obtained, including: in, represents the mth parameter vector in the tth learning process, Represents the parameter vector after joint exploration m=1,2,…,M, M represents the total number of other parameter vectors, C represents the step size control factor, π represents the circumference of a circle, and R1 represents a random number between (-1,1); the step size control factor is: C0 represents the maximum step size, e represents a natural constant, and iter represents the maximum number of learning times.

7. The crane monitoring method based on artificial intelligence according to claim 6 is characterized in that: For the parameter vector after joint exploration, an interpolation exploration strategy that weakens with the number of learning times is used to perform multi-level exploration on the parameter vector, and the parameter vector after multi-level exploration is obtained, including: Obtain the error value of the parameter vector after each joint exploration, and arrange them in ascending order of error value to obtain the arranged parameter vector; According to the parameter vector after arrangement, multi-level exploration is performed on the parameter vector, and the parameter vector after multi-level exploration is obtained as follows: In the formula, represents the nth parameter vector in the tth learning process, n=1,2,3,…,N, N represents the total number of parameter vectors, Represents the parameter vector after multi-level exploration represents the nth parameter vector in the t-1th learning process, represents the nth parameter vector in the t-2th learning process, represents the nth parameter vector in the t-3th learning process, represents the n-1th parameter vector in the tth learning process, and when n=1, represents the optimal parameter vector, v represents the updated control variable, and 8. The crane monitoring method based on artificial intelligence according to claim 7 is characterized in that: For the parameter vector after multi-level exploration, the mutation greedy exploration strategy is used to perform mutation exploration on the parameter vector to obtain the parameter vector after mutation exploration, including: For the parameter vector after multi-level exploration, determine the mutation value corresponding to each parameter vector as follows: in, represents the parameter vector after the k-th multi-level exploration in the t-th learning process, represents the variation value corresponding to the parameter vector after the k-th multi-level exploration, α represents the proportional factor, φ represents the intermediate parameter; μ represents a random number, and μ~N(0,1) obeys the normal distribution, N represents the normal distribution; υ represents a random number, and υ~N(0,1); β represents a random number between (0,2), represents the center of all parameter vectors, γ represents the standard gamma function, and π represents pi; Determine whether the error value corresponding to the variation value corresponding to the parameter vector is reduced. If so, the variation value corresponding to the parameter vector is used as the parameter vector after variation exploration. Otherwise, the original parameter vector after multi-level exploration is directly used as the parameter vector after variation exploration.

9. The crane monitoring method based on artificial intelligence according to claim 7, characterized in that: Determine whether the learning end conditions are met, including: Determine whether the current learning times reaches the maximum learning times or whether the error value of the optimal parameter encoding is less than a preset threshold. If so, determine that the learning end condition is met, otherwise determine that the learning end condition is not met.

10. The crane monitoring method based on artificial intelligence according to claim 1, characterized in that: When the crane operation monitoring result is an abnormal operator behavior category, a crane safety operation warning is issued, including: When the crane operation monitoring result is an abnormal operator behavior category, a text warning message is generated and the text warning message is transmitted to a device designated by the staff or an audible and visual alarm is issued through an audible and visual alarm device installed at the crane site.

Images