Artificial Intelligence-Based Intelligent Management and Control Method and System for Team Operations

Abstract

The present invention discloses an intelligent control method and system for team operation based on artificial intelligence. The method includes: receiving service data from multiple independently deployed operation information systems, performing data correlation, merging and normalization processing to obtain an operation plan set; receiving first monitoring video data, identifying a first abnormal phenomenon based on the first monitoring video data, where the first abnormal phenomenon represents a static abnormal phenomenon in the power scene; retrieving second monitoring video data based on the operation plan, and identifying the execution progress of the operation plan, the first abnormal phenomenon and a second abnormal phenomenon based on the second monitoring video data, where the second abnormal phenomenon represents an abnormal phenomenon associated with the operation execution process. The present invention realizes unified management of operation plan information involved in various systems, avoids the problem of repeated input of team operation plan data, and at the same time realizes intelligent identification of potential safety hazards of violations at the operation site, and macroscopically controls the safety of the operation site.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power operation management, and particularly to an intelligent control method and system for team operation based on artificial intelligence. Background Art

[0002] In recent years, information technologies such as cloud computing, big data, Internet of Things, and mobile applications have developed rapidly, bringing new opportunities for the transformation of the production mode and management mode of power enterprises, and providing conditions for promoting technological transformation and innovative development in the power industry. However, there is still a certain gap in achieving the ultimate goals of online operation, data sharing, and intelligent support in the safety supervision team operation. For example, in multiple professional departments such as power transmission, substation, distribution, information and communication, infrastructure, and marketing of the company, due to different planning management dimensions and various requirements, there is a situation where operation plans are repeatedly entered into multiple business systems, further resulting in the inability to effectively conduct safety control at the operation site according to the requirements of the operation plan. Summary of the Invention

[0003] In view of the above problems existing in the prior art, the present invention provides an intelligent control method and system for team operation based on artificial intelligence, which uniformly manages the operation plan information involved in various systems, realizes the intelligent identification of potential safety hazards at the operation site, and macroscopically controls the safety of the operation site. The technical solution is as follows:

[0004] In the first aspect, the present invention provides an intelligent control method for team operation based on artificial intelligence, including the following steps:

[0005] S1: Receive the business data of multiple independently deployed operation information systems, perform data correlation merging and normalization processing, and obtain an operation plan set;

[0006] S2: Receive the first monitoring video data of the first preset monitoring point, and identify the first abnormal phenomenon based on the first monitoring video data, where the first abnormal phenomenon represents a static abnormal phenomenon in the power scene;

[0007] S3: Retrieve the second monitoring video data of the second monitoring point based on the operation plan, and identify the execution progress of the operation plan, the first abnormal phenomenon, and the second abnormal phenomenon based on the second monitoring video data, where the second abnormal phenomenon represents an abnormal phenomenon associated with the operation execution process.

[0008] Preferably, in step S2, it includes:

[0009] Based on the scene image of the first preset monitoring point, obtain the first characteristic pixel points in the scene image, where the first characteristic pixel points correspond to the targets in the scene image;

[0010] Based on the first characteristic pixel points and a preset first analysis algorithm, identify whether there is an abnormality in the target corresponding to the first characteristic pixel points.

[0011] Preferably, the identifying whether there is an abnormality in the target corresponding to the first characteristic pixel points based on the first characteristic pixel points and a preset first analysis algorithm includes:

[0012] Obtain a preset target area generation algorithm, and combine the target area generation algorithm and the first characteristic pixel points to obtain a scene image target area corresponding to the first characteristic pixel points;

[0013] Based on a preset second analysis algorithm and the scene image target area, determine whether there is an abnormality in the appearance of the target in the target area or determine whether there is an abnormality in the target in the target area based on the target type.

[0014] Preferably, the method for obtaining the first characteristic pixel points includes:

[0015] (1) Select a candidate position of the first characteristic pixel points in the first preset monitoring point scene image;

[0016] (2) Taking the candidate position of the first characteristic pixel points as the center, search for pixel points outward, judge the magnitude of the correlation between the peripheral pixel points and the candidate first characteristic pixel points, and stop searching until the peripheral pixel points with the correlation less than a preset correlation threshold are found;

[0017] (3) Judge whether there are pixel points that have not been searched. If so, select a candidate position of the first characteristic pixel points from the pixel points that have not been searched;

[0018] (4) Repeat steps (2) and (3) until all pixel points in the first preset monitoring point scene image have been searched, and obtain the first characteristic pixel points based on all candidate first characteristic pixel points.

[0019] Preferably, when obtaining the first characteristic pixel points, it includes:

[0020] Analyze the neighborhood pixels of the candidate first characteristic pixel points, compare the pixel point features of the neighborhood pixels with the pixel point features of the candidate first characteristic pixel points. If the distribution difference of the comparison results between the first characteristic pixel points and the neighborhood pixels is greater than a preset threshold, select a pixel point from the neighborhood pixels as a new candidate first characteristic pixel point. If the distribution difference of the comparison results is less than or equal to the preset threshold, determine to retain the candidate first characteristic pixel points.

[0021] Preferably, determining whether there is an abnormality in the presence of the target in the target area based on the preset second analysis algorithm and the target area of the scene image, or determining whether there is an abnormality in the target in the target area based on the target type, includes:

[0022] Based on the target area of the scene image, determine the target type of the target in the target area;

[0023] Based on the mutual distance relationship between the target type of the target area and the geographical location of the target in the target area, determine whether there is an abnormality in the presence of the target in the target area or determine whether there is an abnormality in the target in the target area based on the target type.

[0024] Preferably, in the step S3, it includes:

[0025] Based on the second monitoring video data of the second monitoring point, obtain continuous image frames;

[0026] Based on the continuous image frames, perform image analysis separately in terms of space and time to determine static abnormal phenomena in a single image frame and dynamic abnormal phenomena in continuous image frames.

[0027] Preferably, the performing image analysis separately in terms of space and time based on the continuous image frames to determine static abnormal phenomena in a single image frame and dynamic abnormal phenomena in continuous image frames includes:

[0028] Locate the working area of the workers in the working scene image and determine the number of personnel in the area;

[0029] Determine the type of violation behavior to be detected in the working area according to the work plan;

[0030] Based on the type of violation behavior to be detected, determine the image analysis area for each worker in the working area;

[0031] Perform a summary analysis on the sub - regions of the images to be analyzed of each worker corresponding to the same type of violation behavior to be detected, and determine whether an abnormal phenomenon occurs and the workers with abnormal phenomena according to the summary result.

[0032] Preferably, the performing a summary analysis on the sub - regions of the images to be analyzed of each worker corresponding to the same type of violation behavior to be detected includes:

[0033] Extract the first feature parameters from the sub - regions of the images to be analyzed of each worker obtained based on the same type of violation behavior to be detected, and the first feature parameters characterize the pixel distribution characteristics in the sub - regions of the images;

[0034] Cluster based on each first feature parameter to obtain candidate violation operation personnel corresponding to the category of first feature parameters indicating the occurrence of violation behaviors in the image sub-region.

[0035] Preferably, the step of clustering based on each first feature parameter to obtain candidate violation operation personnel corresponding to the category of first feature parameters indicating the occurrence of violation behaviors in the image sub-region further includes:

[0036] Based on the clustering result of the first feature parameter, obtain each image sub-region in the category of first feature parameters indicating the occurrence of violation behaviors in the image sub-region;

[0037] Based on each obtained image sub-region in the category of first feature parameters indicating the occurrence of violation behaviors in the image sub-region, perform image sub-region behavior recognition to determine whether a violation behavior occurs in the image sub-region.

[0038] In a second aspect, the present invention provides an intelligent control system for team operation based on artificial intelligence, including:

[0039] An operation plan association and merging module, configured to receive service data of multiple independently deployed operation information systems, perform data association, merging and normalization processing to obtain an operation plan set;

[0040] A first abnormal phenomenon monitoring module, configured to receive first monitoring video data of a first preset monitoring point, and identify a first abnormal phenomenon based on the first monitoring video data, where the first abnormal phenomenon represents a static abnormal phenomenon in the power scene;

[0041] A second abnormal phenomenon monitoring module, configured to retrieve second monitoring video data of a second monitoring point based on the operation plan, and identify the execution progress of the operation plan, a first abnormal phenomenon and a second abnormal phenomenon based on the second monitoring video data, where the second abnormal phenomenon represents an abnormal phenomenon associated with the operation execution process.

[0042] The intelligent control method and system for team operation based on artificial intelligence of the present invention has the following beneficial effects: For the service data of multiple independently deployed operation information systems, perform data association and merging to obtain a unified operation plan set, and provide effective data for safety control management in the subsequent operation execution process based on the unified operation plan set. At the same time, it realizes the intelligent identification of various safety hazards at the operation site and macroscopically controls the safety of the operation site. Description of the Drawings

[0043] Figure 1 is a flowchart of the intelligent control method for team operation based on artificial intelligence in an embodiment of the present application;

[0044] Figure 2It is the structural diagram of the intelligent control system for team operation based on artificial intelligence in the embodiments of the present application. Specific Embodiments

[0045] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0046] The embodiments of the present application provide an intelligent control method for team operation based on artificial intelligence, including the following steps:

[0047] S1: Receive the business data of multiple independently deployed operation information systems, perform data association merging and normalization processing, and obtain an operation plan set;

[0048] S2: Receive the first monitoring video data of the first preset monitoring point, and identify the first abnormal phenomenon based on the first monitoring video data, where the first abnormal phenomenon represents a static abnormal phenomenon in the power scene;

[0049] S3: Retrieve the second monitoring video data of the second monitoring point based on the operation plan, and identify the execution progress of the operation plan, the first abnormal phenomenon and the second abnormal phenomenon based on the second monitoring video data, where the second abnormal phenomenon represents an abnormal phenomenon associated with the operation execution process.

[0050] For multiple professional departments such as power transmission, substation, distribution, information and communication, infrastructure, and marketing in the company, due to different plan management dimensions and requirements, there is a situation where operation plans are repeatedly entered in multiple business systems. In the embodiments of the present application, for the business data of multiple independently deployed operation information systems, data association merging is performed to obtain a unified operation plan set, and effective data is provided for safety control management in the subsequent operation execution process based on this unified operation plan set.

[0051] Further, in the embodiments of the present application, for non-real-time operation areas, mainly analyze whether there is a first abnormal phenomenon in the equipment in the power scene, such as line foreign objects, smoke, wildfires, cranes, tower cranes, excavators, etc. For real-time operation areas, mainly analyze and detect the illegal and dangerous behaviors of operators, such as the correct wearing of safety helmets, safety belts, work clothes, smoking, illegal behaviors such as personnel gathering, etc.

[0052] The embodiments of the present application realize the unified management of the operation plan information involved in various systems, avoid the problem of repeated entry of team operation plan data, and at the same time realize the intelligent identification of potential safety hazards of violations at the operation site, and macroscopically control the safety of the operation site.

[0053] Further, in the above step S2, it includes:

[0054] Based on the scene image of the first preset monitoring point, obtain the first feature pixel points in the scene image, where the first feature pixel points correspond to the target in the scene image;

[0055] Based on the first feature pixel points and a preset first analysis algorithm, identify whether there is an abnormality in the target corresponding to the first feature pixel points.

[0056] In the embodiments of the present application, by first determining the first feature pixel points and further determining whether there is an abnormality in the target corresponding to the first feature pixel points, it is avoided that in the case where the number of targets in the scene image is uncertain, the positions of the targets are uncertain, and the types of the targets are uncertain, directly using a variety of target detection algorithms and a variety of abnormal phenomenon analysis algorithms to repeatedly identify the scene image. Based on the number and positions of the first pixel points, the number of targets in the scene image and the approximate position of each target can be determined simultaneously. Further, based on the first feature pixel points and a preset first analysis algorithm, the target region of each target composed of multiple pixel points belonging to the same target as the first feature pixel points in the scene image can be determined. Based on the pixel features of the multiple pixel points belonging to the same target as the first feature pixel points, the target type can be determined and further determine whether there is an abnormal phenomenon in the target.

[0057] Further, the above-mentioned identifying whether there is an abnormality in the target corresponding to the first feature pixel points based on the first feature pixel points and a preset first analysis algorithm includes:

[0058] Obtain a preset target region generation algorithm, and combine the target region generation algorithm and the first feature pixel points to obtain the scene image target region corresponding to the first feature pixel points;

[0059] Based on a preset second analysis algorithm and the scene image target region, determine whether there is an abnormality in the appearance of the target in the target region or determine whether there is an abnormality in the target in the target region based on the target type.

[0060] Further, the method for obtaining the above-mentioned first feature pixel points includes:

[0061] (1) Select a candidate position of the first feature pixel point in the scene image of the first preset monitoring point;

[0062] (2) Centering on the candidate position of the first feature pixel point, search for pixel points outward, judge the magnitude of the correlation between the outer pixel points and the candidate first feature pixel point, and stop searching until the outer pixel points with a correlation less than a preset correlation threshold are found;

[0063] (3) Determine whether there are pixel points that have not been searched. If so, select a candidate first feature pixel point position from the pixel points that have not been searched.

[0064] (4) Repeat steps (2) and (3) until all pixel points in the first preset monitoring point scene image have been searched, and obtain the first feature pixel points based on all candidate first feature pixel points.

[0065] In the embodiment of the present application, in the acquisition of the first feature pixel points, taking the position of the candidate first feature pixel point as the center, search for pixel points outward. When the peripheral pixel points with the associated relationship less than the preset association threshold are searched, the cut-off area of the target area corresponding to the first feature pixel point can be determined. The pixel points obtained by continuing to search outward are the pixel points in the target area of another target. That is, the pixel points obtained by continuing to search outward in the cut-off area of the target area can be another candidate first feature pixel point. In the embodiment of the present application, the disadvantages of directly using multiple target detection algorithms to repeatedly detect and identify the targets in the scene image are further avoided, and the recognition and detection process of multiple targets of multiple target types in the scene image is reduced.

[0066] Further, when obtaining the above first feature pixel points, it includes:

[0067] Analyze the neighborhood pixels of the candidate first feature pixel point, compare the pixel point features of the neighborhood pixels with the pixel point features of the candidate first feature pixel point. If the distribution difference between the comparison results of the first feature pixel point and the neighborhood pixels is greater than the preset threshold, select a pixel point from the neighborhood pixels as the new candidate first feature pixel point. If the distribution difference of the comparison results is less than or equal to the preset threshold, determine to retain the candidate first feature pixel point.

[0068] In the embodiment of the present application, after obtaining the candidate first feature pixel points through the above steps (1)-(4), perform position analysis on the candidate first feature pixel points to analyze whether the candidate first feature pixel point is at the boundary position of the target area. If so, select a pixel point in its neighborhood pixels to replace the candidate first feature pixel point to ensure that the first feature pixel point corresponds to the target in the scene image. Specifically, when the distribution of the comparison results of the first feature pixel point and the neighborhood pixels is relatively similar, it is determined that the position where the first feature pixel point is located is not the boundary position of the target area. When the distribution difference of the comparison results of the first feature pixel point and the neighborhood pixels is large, it is determined that the position where the first feature pixel point is located is the boundary position of the target area. Among them, the comparison result between the first feature pixel point and the neighborhood pixels can be the feature difference loss value between the first feature pixel point and the neighborhood pixels, which can be calculated based on the loss function.

[0069] Further, determining whether there is an abnormality in the presence of the target in the target area or determining whether there is an abnormality in the target in the target area based on the target type, based on the preset second analysis algorithm and the target area of the scene image, includes:

[0070] Based on the target area of the scene image, determine the target type of the target in the target area;

[0071] Based on the distance relationship between the target type of the target area and the geographical location of the target in the target area, determine whether there is an abnormality in the presence of the target in the target area or determine whether there is an abnormality in the target in the target area based on the target type.

[0072] In the embodiments of the present application, the analysis of the first abnormal phenomenon includes the analysis of line foreign objects, smoke, wildfires, cranes, tower cranes, and excavators. After determining the line target area and line position in the scene image, analyze the distance relationship between other targets such as cranes, tower cranes, and excavators and the line in the scene image, and then determine whether there are line hidden dangers. Analyze whether other targets in the scene image are line foreign objects, smoke, or wildfires. If so, determine that there are line hidden dangers.

[0073] In the embodiments of the present application, the target type of the target area can be determined using a target detection algorithm, which is not limited herein.

[0074] Further, step S3 above includes:

[0075] Based on the second monitoring video data of the second monitoring point, obtain consecutive image frames;

[0076] Based on the consecutive image frames, perform image analysis separately in terms of space and time to determine static abnormal phenomena in a single image frame and dynamic abnormal phenomena in consecutive image frames.

[0077] In the embodiments of the present application, the analysis of the second abnormal phenomenon can be based on analyzing static abnormal phenomena in a single image frame and dynamic abnormal phenomena in consecutive image frames based on consecutive image frames. For example, based on a consecutive number of image frames, if it is determined that there is a phenomenon of workers gathering, it can be determined that this abnormal phenomenon occurs in a single image frame. For example, based on a consecutive number of image frames, if the trajectory of the workers is determined, it can be determined whether there is an abnormal phenomenon in the trajectory of the workers, that is, it can be determined whether there is a dynamic abnormal phenomenon in consecutive image frames.

[0078] Further, the above-mentioned performing image analysis separately in terms of space and time based on consecutive image frames to determine static abnormal phenomena in a single image frame and dynamic abnormal phenomena in consecutive image frames includes:

[0079] Locate the working area of the workers in the working scene image and determine the number of people in the area;

[0080] Determine the type of violation behavior to be detected in the operation area according to the operation plan;

[0081] Determine the image analysis area for each operator in the operation area based on the type of violation behavior to be detected;

[0082] Perform a summary analysis on the sub-regions of the images to be analyzed for each operator corresponding to the same type of violation behavior to be detected, and determine whether an abnormal phenomenon occurs and the operator with the abnormal phenomenon according to the summary result.

[0083] Among them, determine the type of violation behavior to be detected in the operation area according to the operation plan. For example, if there is a high-altitude operation in the operation plan, it can be determined that the types of violation behavior to be detected include the possible violation behaviors in high-altitude operations, such as whether the safety belt is worn correctly, etc. For all operators in the operation plan, personnel detection and positioning can be performed, and it can be determined that the types of violation behavior to be detected include whether the safety helmet is worn correctly, whether the work clothes are worn correctly, whether smoking occurs, whether personnel gather, etc.

[0084] Perform a summary analysis on the sub-regions of the images to be analyzed for each operator corresponding to the same type of violation behavior to be detected, and determine whether an abnormal phenomenon occurs and the operator with the abnormal phenomenon according to the summary result. For example, for the detection of safety helmet wearing of operators, the sub-region of the head area image of each operator can be located. Based on the summary analysis of the sub-regions of the head area images of all operators, it can be understood that in one case, the total number of personnel who need to wear safety helmets is determined based on all operators in the operation plan. When all operators who need to wear safety helmets wear safety helmets, the summary result of the sub-regions of the head area images of all operators corresponds to a theoretical summary image feature parameter. When performing a summary analysis on the sub-regions of the head area images of all operators, if the difference between the summary analysis result and the theoretical summary image feature parameter is greater than the preset difference threshold, it can be determined that there is an abnormal phenomenon in the safety helmet wearing of the operator. In the embodiments of the present application, the cumbersome calculation of directly detecting the safety helmet wearing of all operators one by one is avoided.

[0085] Further, the above-mentioned summary analysis of the sub-regions of the images to be analyzed for each operator corresponding to the same type of violation behavior to be detected includes:

[0086] Extract the first feature parameter from the sub-regions of the images to be analyzed for each operator obtained based on the same type of violation behavior to be detected, and the first feature parameter characterizes the pixel distribution feature in the image sub-region;

[0087] Cluster based on each first feature parameter to obtain candidate violation operation personnel corresponding to the category of first feature parameters indicating the occurrence of violation behaviors in the image sub-regions.

[0088] In the embodiments of the present application, for the sub-regions of the images to be analyzed of each operator obtained for the same type of violation behavior to be detected, the first feature parameters are extracted, and the first feature parameters are clustered. It can be understood that there are obvious differences between the image sub-regions of the operators who have committed violation behaviors and those of the operators who have not committed violation behaviors. By clustering the first feature parameters of the image sub-regions, a class set of the image sub-regions determined to have committed violation behaviors can be obtained. The probability that the operators corresponding to the image sub-regions in this class set have committed violation behaviors is relatively high, that is, the probability of committing a violation behavior is greater than that of not committing a violation behavior.

[0089] Furthermore, the above-mentioned clustering based on each first feature parameter to obtain candidate violation operation personnel corresponding to the category of first feature parameters indicating the occurrence of violation behaviors in the image sub-regions further includes:

[0090] Based on the clustering result of the first feature parameters, obtain each image sub-region in the category of first feature parameters indicating the occurrence of violation behaviors in the image sub-regions;

[0091] Based on each image sub-region obtained in the category of first feature parameters indicating the occurrence of violation behaviors in the image sub-regions, perform image sub-region behavior recognition to determine whether a violation behavior has occurred in the image sub-region.

[0092] In the embodiments of the present application, image analysis is performed separately in terms of space and time based on consecutive image frames to determine static abnormal phenomena in a single image frame and dynamic abnormal phenomena in consecutive image frames. Based on the spatial pixel distribution characteristics of the single image frame, it is determined whether there are static abnormal phenomena in the single image frame. Based on consecutive image frames, the position changes of the target regions of the same target in consecutive sequential image frames are located and tracked, and multiple target region images of the same target in consecutive sequential image frames are obtained. Based on the multiple target region images of the same target, it is analyzed whether the target has dynamic abnormal phenomena.

[0093] In the embodiments of the present application, in order to avoid misjudging the violation behaviors of the operators corresponding to the category of first feature parameters indicating the occurrence of violation behaviors in the image sub-regions in the clustering result, further behavior recognition is performed on the image sub-regions in the class set of the image sub-regions determined to have committed violation behaviors to determine whether a violation behavior has occurred in the image sub-region.

[0094] The embodiments of the present application also provide an intelligent control system for team operations based on artificial intelligence, including:

[0095] The job plan association and merging module is used to receive the business data of multiple independently deployed job information systems, perform data association, merging and normalization processing, and obtain a job plan set;

[0096] The first abnormal phenomenon monitoring module is used to receive the first monitoring video data of the first preset monitoring point, and identify the first abnormal phenomenon based on the first monitoring video data. The first abnormal phenomenon represents a static abnormal phenomenon in the power scenario;

[0097] The second abnormal phenomenon monitoring module is used to retrieve the second monitoring video data of the second monitoring point based on the job plan, and identify the execution progress of the job plan, the first abnormal phenomenon and the second abnormal phenomenon based on the second monitoring video data. The second abnormal phenomenon represents an abnormal phenomenon associated with the job execution process.

[0098] In some embodiments, the intelligent control system for team jobs provided in the embodiments of the present application can be implemented in a combination of software and hardware. As an example, the intelligent control system for team jobs provided in the embodiments of the present invention can be directly embodied as a software module combination executed by a processor. The software module can be located in a storage medium, and the storage medium is located in the memory. The processor reads the executable instructions included in the software module in the memory and combines the necessary hardware (for example, including a processor and other components connected to the bus) to complete the intelligent control method for team jobs provided in the embodiments of the present application.

[0099] It should be noted that: the intelligent control system for team jobs provided in this embodiment and the embodiment of the intelligent control method for team jobs provided in the above embodiment belong to the same concept. For the specific implementation process, please refer to the embodiment of the intelligent control method for team jobs, which will not be elaborated here.

[0100] The present invention is not limited to the above specific implementation manners. Those of ordinary skill in the art starting from the above concept and making various changes without creative labor fall within the protection scope of the present invention.

Claims

1. An intelligent control method for team operations based on artificial intelligence, characterized in that, Including: S1: Receive the business data of multiple independently deployed job information systems, perform data correlation and merging and normalization processing to obtain a job plan set; S2: Receive the first monitoring video data of the first preset monitoring point, and identify the first abnormal phenomenon based on the first monitoring video data, where the first abnormal phenomenon represents a static abnormal phenomenon in the power scenario; S3: Retrieve the second monitoring video data of the second monitoring point based on the job plan, and identify the execution progress of the job plan, the first abnormal phenomenon, and the second abnormal phenomenon based on the second monitoring video data, where the second abnormal phenomenon represents an abnormal phenomenon associated with the job execution process; In the step S2, it includes: obtaining a first feature pixel point in the scene image based on the scene image of the first preset monitoring point, where the first feature pixel point corresponds to the target in the scene image; identifying whether there is an abnormality in the target corresponding to the first feature pixel point based on the first feature pixel point and a preset first analysis algorithm; the identifying whether there is an abnormality in the target corresponding to the first feature pixel point based on the first feature pixel point and a preset first analysis algorithm includes: obtaining a preset target area generation algorithm, and combining the target area generation algorithm and the first feature pixel point to obtain a scene image target area corresponding to the first feature pixel point; determining whether there is an abnormality in the appearance of the target in the target area or determining whether there is an abnormality in the target in the target area based on the target type based on a preset second analysis algorithm and the scene image target area; the method for obtaining the first feature pixel point includes: (1) selecting a candidate first feature pixel point position in the scene image of the first preset monitoring point; (2) taking the candidate first feature pixel point position as the center, searching for pixel points outward, judging the size of the association relationship between the peripheral pixel points and the candidate first feature pixel point, and stopping when the association relationship less than a preset association threshold is searched; (3) judging whether there are pixel points that have not been searched, if so, selecting a candidate first feature pixel point position from the pixel points that have not been searched; (4) repeating steps (2) and (3) until all pixel points in the scene image of the first preset monitoring point are searched, and obtaining the first feature pixel point based on all candidate first feature pixel points; when obtaining the first feature pixel point, it includes: analyzing the neighborhood pixels of the candidate first feature pixel point, comparing the pixel point features of the neighborhood pixels with the pixel point features of the candidate first feature pixel point, if the distribution difference of the comparison result between the first feature pixel point and the neighborhood pixels is greater than a preset threshold, then selecting a pixel point from the neighborhood pixels as a new candidate first feature pixel point, if the distribution difference of the comparison result is less than or equal to the preset threshold, then determining to retain the candidate first feature pixel point; the determining whether there is an abnormality in the appearance of the target in the target area or determining whether there is an abnormality in the target in the target area based on the target type based on a preset second analysis algorithm and the scene image target area includes: determining the target type of the target in the target area based on the scene image target area; determining whether there is an abnormality in the appearance of the target in the target area or determining whether there is an abnormality in the target in the target area based on the target type based on the mutual distance relationship between the target type of the target area and the geographical location of the target in the target area; In step S3, it includes: obtaining continuous image frames based on the second monitoring video data of the second monitoring point; performing image analysis on the continuous image frames respectively in terms of space and time to determine static abnormal phenomena in a single image frame and dynamic abnormal phenomena in the continuous image frames; the performing image analysis on the continuous image frames respectively in terms of space and time to determine static abnormal phenomena in a single image frame and dynamic abnormal phenomena in the continuous image frames includes: positioning the working areas of the workers in the working scene image and determining the number of personnel in the area; determining the types of violation behaviors to be detected in the working area according to the work plan; determining the image analysis areas for each worker in the working area based on the types of violation behaviors to be detected; performing summary analysis on the sub-image areas to be analyzed for each worker corresponding to the same type of violation behavior to be detected, and determining whether abnormal phenomena occur and the workers with abnormal phenomena according to the summary result; the performing summary analysis on the sub-image areas to be analyzed for each worker corresponding to the same type of violation behavior to be detected includes: extracting first feature parameters from the sub-image areas to be analyzed for each worker obtained based on the same type of violation behavior to be detected, where the first feature parameters represent the pixel distribution characteristics in the sub-image areas; performing clustering based on each first feature parameter to obtain candidate violating workers corresponding to the first feature parameter category indicating the occurrence of violation behaviors in the sub-image areas.

2. The intelligent control system for team operation based on artificial intelligence is characterized in that It includes: A work plan association and merging module, configured to receive the service data of multiple independently deployed and dispersed work information systems, perform data association, merging and normalization processing, and obtain a work plan set; A first abnormal phenomenon monitoring module, configured to receive the first monitoring video data of the first preset monitoring point, and identify a first abnormal phenomenon based on the first monitoring video data, where the first abnormal phenomenon represents a static abnormal phenomenon in the power scene; A second abnormal phenomenon monitoring module, configured to retrieve the second monitoring video data of the second monitoring point based on the work plan, and identify the execution progress of the work plan, the first abnormal phenomenon and the second abnormal phenomenon based on the second monitoring video data, where the second abnormal phenomenon represents an abnormal phenomenon associated with the work execution process; In the first abnormal phenomenon monitoring module, the execution steps include: based on the scene image of the first preset monitoring point, obtaining the first characteristic pixel points in the scene image, where the first characteristic pixel points correspond to the target in the scene image; based on the first characteristic pixel points and a preset first analysis algorithm, identifying whether there is an abnormality in the target corresponding to the first characteristic pixel points; the identifying whether there is an abnormality in the target corresponding to the first characteristic pixel points based on the first characteristic pixel points and the preset first analysis algorithm includes: obtaining a preset target area generation algorithm, combining the target area generation algorithm and the first characteristic pixel points to obtain the scene image target area corresponding to the first characteristic pixel points; based on a preset second analysis algorithm and the scene image target area, determining whether there is an abnormality in the appearance of the target in the target area or determining whether there is an abnormality in the target in the target area based on the target type; the method for obtaining the first characteristic pixel points includes: (1) selecting a candidate position of the first characteristic pixel point in the scene image of the first preset monitoring point; (2) taking the candidate position of the first characteristic pixel point as the center, searching outward to obtain pixel points, judging the magnitude of the correlation relationship between the peripheral pixel points and the candidate first characteristic pixel point, and stopping when the peripheral pixel points with a correlation relationship less than a preset correlation threshold are searched; (3) judging whether there are pixel points that have not been searched, if so, selecting a candidate position of the first characteristic pixel point from the pixel points that have not been searched; (4) repeating steps (2) and (3) until all pixel points in the scene image of the first preset monitoring point are searched, and obtaining the first characteristic pixel points based on all candidate first characteristic pixel points; when obtaining the first characteristic pixel points, it includes: analyzing the neighborhood pixels of the candidate first characteristic pixel points, comparing the pixel point characteristics of the neighborhood pixels with the pixel point characteristics of the candidate first characteristic pixel points, if the distribution difference of the comparison results between the first characteristic pixel points and the neighborhood pixels is greater than a preset threshold, then selecting a pixel point from the neighborhood pixels as a new candidate first characteristic pixel point, if the distribution difference of the comparison results is less than or equal to the preset threshold, then determining to retain the candidate first characteristic pixel point; the determining whether there is an abnormality in the appearance of the target in the target area or determining whether there is an abnormality in the target in the target area based on the target type based on the preset second analysis algorithm and the scene image target area includes: determining the target type of the target in the target area based on the scene image target area; based on the mutual distance relationship between the target type of the target area and the geographical location of the target in the target area, determining whether there is an abnormality in the appearance of the target in the target area or determining whether there is an abnormality in the target in the target area based on the target type; In the second abnormal phenomenon monitoring module, the execution steps include: obtaining consecutive image frames based on the second monitoring video data of the second monitoring point; performing image analysis on the consecutive image frames separately in terms of space and time to determine static abnormal phenomena in a single image frame and dynamic abnormal phenomena in the consecutive image frames; the performing image analysis on the consecutive image frames separately in terms of space and time to determine static abnormal phenomena in a single image frame and dynamic abnormal phenomena in the consecutive image frames includes: locating the working area of the workers in the working scene image and determining the number of workers in the area; determining the type of violation behavior to be detected in the working area according to the work plan; determining the image analysis area for each worker in the working area based on the type of violation behavior to be detected; performing summary analysis on the sub-image areas to be analyzed of each worker corresponding to the same type of violation behavior to be detected, and determining whether an abnormal phenomenon occurs and the workers with abnormal phenomena according to the summary result; the performing summary analysis on the sub-image areas to be analyzed of each worker corresponding to the same type of violation behavior to be detected includes: extracting a first feature parameter from the sub-image areas to be analyzed of each worker obtained based on the same type of violation behavior to be detected, where the first feature parameter characterizes the pixel distribution feature in the sub-image area; performing clustering based on each first feature parameter to obtain candidate violation workers corresponding to the first feature parameter category indicating the occurrence of violation behavior in the sub-image area.

Images