CN-121979047-A - Multi-scene fusion fuel management centralized video monitoring and alarming method and related device

CN121979047ACN 121979047 ACN121979047 ACN 121979047ACN-121979047-A

Abstract

The invention discloses a fuel management centralized video monitoring and alarming method with multi-scene fusion and a related device, belonging to the technical field of coal-fired operation management; the method comprises the steps of collecting multidimensional environment data and operation behavior data in the operation process, carrying out fusion analysis on the collected multidimensional environment data and operation behavior data, identifying abnormal operation behaviors, and executing corresponding hierarchical early warning and equipment control instructions according to the identified abnormal operation behaviors. Based on the blockchain technology, the whole flow operation data and events from data acquisition to alarm response are stored, and a trusted traceability evidence chain is formed. The invention constructs a multi-dimensional sensing, full-chain analysis and intelligent linkage centralized monitoring system, and forms a closed-loop supervision capability covering the whole process of acquisition, transportation, sample preparation and assay.

Inventors

CHEN JIANPING
MA CHENXI
ZHANG CHUXIAN
LI YUXUAN
GUO JIAXI

Assignees

西安热工研究院有限公司
华能上海石洞口发电有限责任公司

Dates

Publication Date: 20260505
Application Date: 20260122

Claims (10)

1. The fuel management centralized video monitoring and alarming method for the multi-scene fusion is characterized by comprising the following steps of: collecting multidimensional environment data and operation behavior data in the operation process; Carrying out fusion analysis on the collected multidimensional environment data and operation behavior data, and identifying abnormal operation behaviors; and executing corresponding hierarchical early warning and equipment control instructions according to the identified abnormal operation behaviors.
2. The method for centralized video monitoring and alarming of fuel management of multi-scene fusion according to claim 1, wherein the step of collecting the multi-dimensional environmental data and the operation behavior data in the operation process comprises the following steps: Collecting image data of a target area through a visible light camera; acquiring temperature distribution data of a target area through an infrared thermal imager; acquiring point cloud data through a millimeter wave radar, and constructing a three-dimensional point cloud space field; Acquiring operation vibration spectrum data of key sample preparation equipment through a vibration sensor; and acquiring centimeter-level precision displacement track data of an operation tool and a person through UWB positioning.
3. The method for centralized video monitoring and alarming of multi-scene fusion fuel management according to claim 1, wherein the step of performing fusion analysis on the collected multi-dimensional environmental data and operation behavior data to identify abnormal operation behavior specifically comprises the following steps: Preprocessing the collected multidimensional environment data and operation behavior data; analyzing the joint rotation angle and micro-attitude characteristics of an operator through a binocular vision three-dimensional reconstruction technology based on the preprocessed operation behavior data to obtain first characteristic data; Based on the preprocessed multidimensional environment data, analyzing the continuous time sequence characteristics of the operation rhythm and the tool movement track by adopting a time convolution network to obtain second characteristic data; And constructing a space-time correlation map of personnel, equipment and materials by combining the first characteristic data and the second characteristic data, carrying out correlation analysis and causal inference on the multisource abnormal characteristics, and identifying abnormal operation behaviors.
4. The method for centralized video monitoring and warning of fuel management of a multi-scenario fusion according to claim 3, wherein the step of preprocessing the collected multi-dimensional environmental data and operation behavior data specifically comprises: when the environment dust concentration is monitored to exceed a preset threshold value, starting a dynamic spectrum compensation and non-uniform illumination enhancement algorithm for the image data; Aiming at the video jitter problem caused by mechanical vibration, carrying out electronic image stabilization processing on video stream data based on a six-axis inertial sensing-assisted electronic image stabilization mechanism; and carrying out space-time registration on the millimeter wave Lei Dadian cloud data and the data acquired by the infrared thermal imager so as to construct a fused three-dimensional motion trail model.
5. The method for centralized video monitoring and warning of fuel management of multi-scene fusion according to claim 1, wherein said step of executing corresponding hierarchical early warning and device control instructions according to said identified abnormal operation behavior specifically comprises: When the transient abnormality is identified, starting a first-level early warning, and projecting a visual correction guide to a field operator through the augmented reality equipment; When the associated multidimensional evidence confirms that the violation risk exists, a secondary response is started, and key operation parameters of the sample equipment are locked through an industrial control protocol; when the systematic fraud risk is judged to exist, a three-level alarm is started, a blockchain certification is triggered, an alarm event is pushed to an upper-level supervision platform, and meanwhile, the hard blocking operation of the equipment is executed.
6. The method for centralized video monitoring and alarming of fuel management of claim 1, further comprising verifying full-process operation data and events from data acquisition to alarming response based on blockchain technology to form a trusted traceability evidence chain.
7. The method for centralized video monitoring and alarming of fuel management of multi-scene fusion according to claim 6, wherein the step of verifying the whole flow operation data and events from data acquisition to alarm response based on the blockchain technology to form a trusted traceability evidence chain specifically comprises the following steps: Generating a unique digital fingerprint for each fuel sample; The biological characteristic data, the equipment working condition parameters, the environment data and the abnormal event label of the operator are packaged in a correlated manner with the digital fingerprint to form an operation evidence chain; establishing a quantization mapping model between a sample quality index and operation compliance; And when an abnormal result appears in the test link, performing reverse tracing on the abnormal result based on the quantitative mapping model and the operation evidence chain of the blockchain evidence, and positioning to a specific illegal operation link.
8. The utility model provides a fuel management centralized video monitoring and warning system of multiscreen fusion which characterized in that includes: The data acquisition module is used for acquiring multidimensional environment data and operation behavior data in the operation process; The abnormal identification module is used for carrying out fusion analysis on the collected multidimensional environment data and operation behavior data and identifying abnormal operation behaviors; and the alarm module is used for executing corresponding hierarchical early warning and equipment control instructions according to the identified abnormal operation behaviors.
9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor, when executing the computer program, implements the steps of a fuel management centralized video monitoring and alerting method of a multi-scenario fusion as claimed in any one of claims 1-7.
10. A computer readable storage medium storing a computer program, wherein the computer program when executed by a processor implements the steps of a fuel management centralized video monitoring and alerting method of any one of claims 1-7.

Description

Multi-scene fusion fuel management centralized video monitoring and alarming method and related device Technical Field The invention belongs to the technical field of coal burning operation management, and relates to a fuel management centralized video monitoring and alarming method with multi-scene fusion and a related device. Background In the energy industry, quality management of fuels (such as coal) is a key link for guaranteeing production benefits and cost control. The accuracy of the fuel collection, sample preparation and assay processes is crucial, and the trade settlement and combustion efficiency are directly related. However, in actual operation, the above links involve human operation and lack of effective supervision, so that there is a risk of fraud. To address this challenge, the prior art generally employs a video monitoring system and a preliminary intelligent analysis technology, but its application in a practical complex industrial scenario still has systematic defects. First, conventional monitoring systems rely mainly on video monitoring with visible light, which poses serious challenges in the core scenario of fuel management, such as sample preparation workshops. The environment is generally characterized by high dust concentration, strong mechanical vibration, dense metal equipment and the like. The existing single visible light monitoring scheme is poor in performance under the conditions that high dust causes light scattering, image blurring and target detail loss are caused, mechanical vibration causes continuous shaking of a video picture, key operation frames are caused to be lost, stable time sequence analysis is difficult to perform, and dense metal equipment interferes with wireless signal transmission to influence the reliability of sensor data. Secondly, in the aspect of behavior recognition and analysis, the prior art is mostly based on single-frame analysis of two-dimensional images or simple time sequence threshold judgment. The method has serious defects in analysis capability for operators to deliberately avoid monitoring behaviors (such as back to camera, sideways shielding, tool shielding and the like). The skeleton joint point detection algorithm based on monocular vision is easily affected by visual angle transformation, has limited analysis precision on micro-gestures such as shoulder joint rotation, trunk torsion and the like, and is difficult to distinguish compliance operation from disguised illegal actions. Meanwhile, the existing algorithm lacks deep semantic understanding of the continuous operation flow, and can not convert discrete operation actions (such as coal shoveling, dividing and bottling) into quantifiable node events which accord with the standard operation flow. Currently, individual monitoring subsystems for fuel management (e.g., sample point monitoring, transportation monitoring, sample room monitoring, laboratory data) typically employ an independent deployment, split-and-treat mode. This results in different system data standards, and a lack of efficient space-time alignment and correlation analysis mechanisms. Video stream data, equipment vibration data, tool trajectory data, and final assay result data are fractured from each other and a complete causal chain of "physical manipulation-equipment status-sample quality" cannot be constructed. When one link is abnormal, the system is difficult to trace to the problem source of upstream operation, and risk linkage early warning of the cross links cannot be performed. In addition, the existing system has a plurality of rigidifying structures, the intelligent analysis is lagged due to insufficient edge computing capability, the real-time response of the sub-second level cannot be realized, the cloud model is long in updating period, and novel cheating means are difficult to adapt quickly. Data barriers between various factory areas also prevent knowledge sharing and collaborative defense of violation patterns, leaving room for organized, systematic fraud. In addition, the monitoring data is disjointed with the business logic, the evidence content is mostly an original video or simple event label, and the association encapsulation of multidimensional information such as operation context, equipment state, environmental parameters and the like is lacking, so that the evidence chain fragmentation and the interpretation are weak. When disputes occur or need to be examined, it is difficult to provide clear, complete, non-tamperable and convincing technical evidence. In summary, the existing fuel management monitoring technology is limited by the singleness of the sensing means, the shallow layer of behavior analysis, the cracking property of the system architecture and the incompleteness of the traceability system, and is difficult to cope with increasingly hidden and intelligent fraud risks in complex industrial scenes. Therefore, there is an urgent need in the art for a job monitoring method that