New energy operation and maintenance field operation behavior intelligent identification and compliance monitoring system and method

Abstract

The invention relates to the technical field of new energy equipment operation and maintenance, in particular to a new energy operation and maintenance field operation behavior intelligent identification and compliance monitoring system and method, and the system comprises a multi-source sensing layer, an edge intelligent layer, a compliance reasoning layer, a command linkage layer and an evidence and auditing layer. The multi-source sensing layer adopts a multi-source acquisition device to monitor operation and maintenance field personnel, tools and environment states, and acquires and fuses multi-source data; the edge intelligent layer performs space-time alignment on the fused data, performs action recognition and interactive behavior modeling analysis, and outputs a result; the compliance reasoning layer performs compliance judgment based on the result to form a risk quantification result; the command linkage layer performs measures according to the risk quantification result and the risk level grading response; and the evidence and auditing layer forms an evidence chain and a compliance auditing log which cannot be tampered according to the evidence and auditing layer so as to realize evidence storage traceability.

Description

Technical Field

[0001] This invention relates to the field of new energy equipment operation and maintenance technology, specifically to a system and method for intelligent identification and compliance monitoring of on-site operation behaviors in new energy operation and maintenance. Background Technology

[0002] With the rapid expansion of new energy power generation, various power sources such as wind power, photovoltaics, and energy storage are increasingly being connected to the grid, making the on-site operation environment increasingly complex. Maintenance personnel need to perform various operations on-site, including inspection, debugging, and troubleshooting. These operations often involve working at heights, contact with high-voltage electrical equipment, and multi-person collaboration; even slight negligence can lead to safety accidents. Therefore, real-time monitoring and compliance management of on-site personnel's behavior has become a crucial aspect of ensuring the safe operation of new energy power systems.

[0003] Current operation and maintenance management mainly relies on manual inspections and video surveillance, with some scenarios incorporating image recognition-based behavior detection methods. However, these methods generally suffer from limitations such as single detection dimensions, insufficient recognition accuracy, and poor environmental adaptability. Especially in multi-person collaborative operations, the use of complex tools, and high-risk environments, existing systems often struggle to simultaneously and comprehensively assess multiple compliance factors, including actions, equipment wearing, and boundary violations, leading to high false alarm and false negative rates. Furthermore, traditional monitoring methods lack a unified mechanism for data storage and violation tracing, making it difficult to establish a complete chain of evidence in the event of an incident.

[0004] Based on the above situation, there is an urgent need for a technical solution in new energy operation and maintenance sites that can achieve multi-source data fusion, intelligent identification of personnel behavior, and quantitative assessment of compliance in complex environments. This solution aims to address the technical problems of existing technologies, such as low accuracy in behavior monitoring, delayed response, and insufficient evidence management. Summary of the Invention The technical problem to be solved by the present invention is to provide an intelligent identification and compliance monitoring system and method for on-site operation behavior of new energy operation and maintenance, which addresses the shortcomings of the prior art and solves the technical problems of low behavior monitoring accuracy, delayed response and insufficient evidence management in the prior art.

[0005] The objective of this invention is achieved through the following technical solutions: In a first aspect, the present invention provides an intelligent identification and compliance monitoring system for on-site operation behavior of new energy operation and maintenance, including a multi-source perception layer, an edge intelligence layer, a compliance reasoning layer, a command and linkage layer, and an evidence and audit layer with communication connection; The multi-source sensing layer is used to monitor the status of personnel, tools and equipment and the operation and maintenance environment at the operation and maintenance site using multi-source acquisition devices, and to collect multi-source data, and to fuse the multi-source data in a unified coordinate system to obtain fused data. The edge intelligence layer is used to perform spatiotemporal alignment on the fused data output by the multi-source perception layer, and to perform action recognition and interactive behavior modeling analysis on the spatiotemporally aligned fused data to obtain action recognition results and interactive behavior judgment results. The compliance reasoning layer is used to make compliance judgments based on action recognition results and interaction behavior judgment results, and to form risk quantification results based on the judgment results; The command and coordination layer is used to respond in a graded manner based on the risk quantification results and the corresponding risk level, and to execute the corresponding graded response measures. The evidence and audit layer is used to form an immutable chain of evidence and compliance audit log based on the risk quantification results, so as to realize the evidence preservation and traceability of the entire monitoring process.

[0006] As a further improvement of the present invention, the multi-source acquisition device includes at least a video acquisition device, a position and attitude acquisition device, an intelligent personal protective equipment detection device, a tool identification device, and an environmental status sensing device. The video acquisition device includes a fixed industrial camera or a pan-tilt camera, used to acquire images of the operator's movements; The position and attitude acquisition device includes a UWB positioning module and an IMU inertial measurement unit; the UWB positioning module is used to acquire the three-dimensional position of the operator in the work area in real time, and the IMU inertial unit is used to acquire the operator's attitude and motion trajectory. The intelligent personal protective equipment detection device is installed inside or on the surface of safety helmets, safety belts, and insulating gloves, and uses a detection device with tension and pressure sensors. The tool identification device identifies and detects the usage status of tools by attaching Bluetooth Low Energy tags to them and cross-referencing the data with video capture data. Environmental condition sensing devices include those deployed in high-risk work areas to acquire work environment parameters; After clock synchronization and signal calibration, each acquisition device achieves data acquisition under a unified time reference, and the joint estimation of position, attitude and image key points is achieved through the extended Kalman filter method.

[0008] As a further improvement of the present invention, the edge intelligence layer adopts a lightweight deep learning model and combines key point detection network with IMU posture data for cross-modal fusion to form a complete motion feature vector of the operator; the edge intelligence layer further introduces spatial prior factors through a multi-scale interactive attention mechanism, so that close-range collaborative actions are identified as positive interactions, while long-range interference actions are identified as abnormal behaviors.

[0009] As a further improvement of the present invention, the multi-scale interactive attention mechanism is as follows:

[0010] Among them, among them, Indicates the first The first worker and the first Interaction weights between individual operators; , These are the query vector and key vector representing the personnel characteristics, respectively. For feature dimensions; Spatial distance between people; It is a monotonically decreasing function, used to represent spatial decay; To adjust the parameters.

[0011] As a further improvement of the present invention, the compliance inference layer formally constrains the behavior and state sequences in the action recognition results through a probabilistic temporal logic rule engine. The formal constraints include: in high-altitude operations, when the personnel's working height exceeds a set threshold, the safety belt buckle action must be completed within a limited time window; in the closing operation scenario, the identities of the operator and the reviewer must be different; in the hot work scenario, the high-charge state and the hot work are defined as mutually exclusive.

[0012] As a further improvement of the present invention, the compliance reasoning layer is based on the reasoning behavior results and potential risks of the operation knowledge graph. The knowledge graph includes process nodes, tool nodes, qualification nodes and environment nodes. The nodes are connected by pre- and post-relationships, dependency relationships and mutual exclusion relationships, thereby realizing differentiated rule verification in different operation scenarios.

[0013] As a further improvement of the present invention, in the compliance reasoning layer, compliance determination is performed through a risk scoring mechanism, wherein the risk scoring mechanism is as follows:

[0014] in, Indicates risk score; Indicates the probability of a violation; Indicates the probability of abnormality in personal protective equipment; Indicates the probability of the region going out of bounds; Indicates environmental hazard factors; , , , Let be the weight coefficient, and satisfy... .

[0015] As a further improvement of the present invention, the command and control layer provides a prompt through a wearable display device or vibration device when the risk score is low, triggers an audible and visual alarm and a voice broadcast to warn when the risk score is medium, and triggers an emergency stop command for the work equipment, a power outage protection mechanism, or a remote security system to force intervention when the risk score is high, and synchronously transmits the handling record to the superior monitoring center.

[0016] As a further improvement of the present invention, the evidence and audit layer solidifies and stores the operation data through a segmented encapsulation method. Each data segment includes video keyframes, positioning trajectories, posture sequences, identification tags, and compliance judgment results. A unique verification value is calculated for each data segment, and adjacent data segments are connected sequentially through a chain hash method.

[0017] Secondly, the present invention provides a working method based on the above-mentioned intelligent identification and compliance monitoring system for on-site operation behavior in new energy operation and maintenance, comprising: The system monitors the personnel, tools, and environmental conditions at the maintenance site using video acquisition devices, position and attitude acquisition devices, intelligent personal protective equipment detection devices, tool identification devices, and environmental status sensing devices, and fuses the collected multi-source data in a unified coordinate system. The fused multi-source data is spatiotemporally aligned, and the spatiotemporally aligned data is used for action recognition and interaction behavior modeling analysis to obtain action recognition and interaction behavior judgment results. Based on the results of action recognition and interaction behavior judgment, a probabilistic temporal logic rule engine, a job knowledge graph, and a risk scoring mechanism are used to determine compliance and generate risk quantification results. Based on the risk quantification results and risk level classification response, implement prompts, warnings, and handling measures; Based on the risk quantification results, an immutable chain of evidence and compliance audit logs are formed to achieve evidence preservation and traceability throughout the entire process.

[0018] The beneficial effects of this invention are as follows: This invention provides an intelligent identification and compliance monitoring system for on-site operations in new energy operation and maintenance. Through a multi-source perception layer, it monitors the status of personnel, tools, and the environment at the operation and maintenance site and fuses the collected data under a unified coordinate system. An edge intelligence layer performs spatiotemporal alignment on the data output from the multi-source perception layer and obtains judgment results through action recognition and interactive behavior modeling analysis. A compliance reasoning layer utilizes a probabilistic temporal logic rule engine, an operation knowledge graph module, and a risk scoring mechanism to determine the compliance of the judgment results and form a risk quantification result. A command and linkage layer responds with prompts, warnings, and handling measures based on the risk quantification result and risk level classification. Finally, an evidence and audit layer forms an immutable evidence chain and compliance audit log based on the risk quantification result, achieving full-process evidence storage and traceability. Thus, under a unified architecture, it achieves real-time fusion of on-site operation and maintenance data, precise behavior analysis, quantitative processing of risk judgment, hierarchical execution of responses, and reliable recording of evidence. This effectively solves the technical problems of scattered on-site operation and maintenance monitoring, delayed compliance judgment, untimely risk response, and insufficient traceability, achieving the technical effects of improving monitoring accuracy, enhancing the real-time nature of risk response, ensuring the reliability of compliance management, and providing full-process traceability. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the structure of the intelligent identification and compliance monitoring system for on-site operation behavior of new energy operation and maintenance in an embodiment of the present invention. Detailed Implementation

[0021] To make the objectives and technical solutions of this invention clearer and easier to understand, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. The specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention.

[0022] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings and specific embodiments. The described embodiments are only some embodiments of the present invention, and not all embodiments.

[0023] Example 1 This embodiment provides an intelligent identification and compliance monitoring system for on-site operation behavior of new energy operation and maintenance, including a multi-source perception layer, an edge intelligence layer, a compliance reasoning layer, a command and linkage layer, and an evidence and audit layer with communication connection.

[0024] The multi-source sensing layer is used to monitor the status of personnel, tools, and the environment at the maintenance site through video acquisition devices, position and attitude acquisition devices, intelligent personal protective equipment detection devices, tool identification devices, and environmental status sensors. It also fuses the collected data under a unified coordinate system. The multi-source sensing layer includes fixed industrial cameras or pan-tilt-zoom cameras, UWB positioning modules, IMU inertial measurement units, personal protective equipment detection devices with sensors, and tool identification devices with attached Bluetooth Low Energy tags. Each acquisition device achieves data acquisition under a unified time reference after clock synchronization and signal calibration. Joint estimation of position, attitude, and image key points is achieved through an extended Kalman filter method. Image data is acquired by fixed industrial cameras or PTZ cameras, and position and attitude data are obtained by UWB positioning modules and IMU inertial measurement units. The status of equipment and tools is monitored by personal protective equipment detection devices with sensors and tool identification devices with Bluetooth Low Energy tags. All devices are synchronized by clock and calibrated by signal to achieve data acquisition under a unified time reference. The extended Kalman filter method is used to jointly estimate position, attitude and key points in the image, thereby solving the technical problems of inconsistent time references and isolated spatial information of multi-source heterogeneous data. This achieves the technical effect of improving the spatiotemporal correlation accuracy of multi-source sensing data and realizing the collaborative optimization of position, attitude and visual features.

[0025] The edge intelligence layer includes an edge computing host, a lightweight deep learning model, and a cross-modal fusion and inference module, which are used to perform action recognition and interaction behavior modeling on spatiotemporally aligned data; The edge intelligence layer employs a lightweight deep learning model and combines it with a keypoint detection network and IMU posture data for cross-modal fusion to form a complete motion feature vector for the worker. The edge intelligence layer further introduces spatial prior factors through a multi-scale interactive attention mechanism, enabling close-range collaborative actions to be recognized as positive interactions, while long-range interfering actions are identified as abnormal behaviors. By using a lightweight deep learning model combined with a keypoint detection network and IMU posture data for cross-modal fusion to generate a complete motion feature vector for the worker, and utilizing a multi-scale interactive attention mechanism to introduce spatial prior factors, close-range collaborative actions are recognized as positive interactions while long-range interfering actions are identified as abnormal behaviors. This solves the technical problems of incomplete single-modal motion feature representation and high misjudgment rate of interactive behaviors in complex scenarios, achieving the technical effects of enhancing the completeness of motion representation, improving the accuracy of close-range collaborative recognition, and effectively suppressing misjudgments of long-range interference.

[0026] Furthermore, the multi-scale interactive attention mechanism is as follows:

[0027] Among them, among them, Indicates the first The first worker and the first Interaction weights between individual operators; , These are the query vector and key vector representing the personnel characteristics, respectively. For feature dimensions; Spatial distance between people; It is a monotonically decreasing function, used to represent spatial decay; To adjust the parameters.

[0028] The compliance reasoning layer includes a probabilistic temporal logic rule engine, an operational knowledge graph module, and a risk scoring mechanism, which are used to determine the compliance of actions and behaviors and generate risk quantification results.

[0029] The compliance inference layer formalizes the work steps through probabilistic temporal logic. The constraints include: in high-altitude operations, when the personnel's working height exceeds the set threshold, the safety belt buckle action must be completed within a limited time window; in the closing operation scenario, the identities of the operator and the reviewer must be different; in the hot work scenario, the high-charge state and the hot work are defined as mutually exclusive.

[0030] The compliance reasoning layer establishes an operational knowledge graph, which includes process nodes, tool nodes, qualification nodes, and environment nodes. These nodes are connected through pre- and post-relationships, dependencies, and mutual exclusions, thereby enabling differentiated rule verification in different operational scenarios.

[0031] The risk scoring method for compliance assessment is as follows:

[0032] in, Indicates risk score; Indicates the probability of a violation; Indicates the probability of abnormality in personal protective equipment; Indicates the probability of the region going out of bounds; Indicates environmental hazard factors; , , , Let be the weight coefficient, and satisfy... .

[0033] The command and control layer includes a prompting submodule, a warning submodule, and a response submodule, used to respond to and execute prompts, warnings, and response measures according to the risk level. When the risk score is low, the command and control layer provides prompts via wearable display devices or vibration devices; when the risk score is medium, it triggers audible and visual alarms and voice broadcasts via electronic fences to issue warnings; and when the risk score is high, it triggers emergency stop commands for the work equipment, activates power-off protection, or forces intervention through remote security systems, and synchronously transmits the response records to the superior monitoring center.

[0034] The evidence and audit layer includes an evidence solidification unit, a timestamp and hash chain unit, and a violation log generation unit, which are used to form an immutable chain of evidence and a compliance audit log to achieve evidence preservation and traceability throughout the entire process.

[0035] The evidence and audit layer stores operational data in a segmented encapsulation manner. Each data segment includes video keyframes, positioning trajectories, attitude sequences, identification tags, and compliance judgment results. A unique verification value is calculated for each data segment, and adjacent data segments are connected sequentially using a chain hash method.

[0036] Furthermore, the evidence and auditing layer, while forming the hash chain, invokes an authoritative timestamp service to sign key nodes to ensure the time legitimacy and authority of the data, and quantifies the credibility of the evidence based on an integrity factor and a timestamp factor. The integrity factor is calculated based on the ratio of the number of data segments that pass hash verification to the total number of data segments, and the timestamp factor is calculated based on the ratio of the number of data segments that are timestamped to the total number of data segments. When the credibility score is greater than or equal to a preset threshold, the evidence is considered highly credible.

[0037] Example 2 This invention proposes a specific implementation of an intelligent identification and compliance monitoring system for on-site operations in new energy operation and maintenance. The system aims to address issues in existing technologies such as single-function operation identification, difficulty in accurately judging collaborative operations involving multiple personnel, static and rigid rule verification, and a lack of traceable audit mechanisms. By constructing an overall architecture of "multi-source perception—edge recognition—compliance reasoning—joint handling—evidence solidification," the system achieves comprehensive monitoring and compliance determination of on-site operation processes.

[0038] like Figure 1As shown, the system mainly comprises a multi-source perception layer, an edge intelligence layer, a compliance reasoning layer, a command and control layer, and an evidence and audit layer. The multi-source perception layer comprehensively acquires personnel location, posture, video images, tool status, and environmental parameters. The edge intelligence layer fuses multimodal data, extracts key human body points, and combines cross-modal feature alignment and interactive attention mechanisms to identify maintenance personnel actions and multi-person interaction behaviors. The compliance reasoning layer, based on probabilistic temporal logic and operational knowledge graphs, performs formal compliance verification on action sequences and dynamically optimizes rules through a self-correction mechanism. The command and control layer triggers multi-level response measures such as prompts, warnings, or emergency handling based on the compliance judgment results. The evidence and audit layer solidifies evidence through segmented hash chains and timestamp signatures, generating interpretable violation logs and forming a traceable safety supervision closed loop.

[0039] Through the coordinated operation of the above modules, the system of the present invention can not only identify personnel behavior in complex operation scenarios in real time, but also dynamically determine whether the operation process complies with the requirements of the regulations, and realize graded handling when the risk of violation occurs. At the same time, it provides a credible chain of evidence for subsequent accident investigation and compliance audit, thereby significantly improving the safety and standardization level of new energy operation and maintenance sites.

[0040] To facilitate a clearer understanding of the technical solution of this invention, the following will describe it in conjunction with the various functional modules of the system. Figure 1 The specific structure, working principle, and implementation of the intelligent identification and compliance monitoring system for on-site operations of new energy maintenance will be further explained.

[0041] The multi-source sensing layer, as the basic data acquisition unit of the system of this invention, is mainly used to comprehensively monitor the personnel, tools, and environmental conditions at the new energy operation and maintenance site. This layer includes video acquisition devices, position and attitude acquisition devices, intelligent personal protective equipment detection devices, tool identification devices, and environmental condition sensing devices, and is connected to the edge intelligent layer via wired or wireless means.

[0042] Specifically, video acquisition devices employ fixed industrial cameras or pan-tilt-zoom (PTZ) cameras, installed in key areas such as wind turbine tower bases, substation access roads, and energy storage battery compartments to capture images of workers' movements. Position and attitude acquisition devices include UWB positioning modules and IMU (Inertial Measurement Units). The UWB positioning module acquires the real-time three-dimensional position of workers within the work area, while the IMU acquires their posture and movement trajectory. Intelligent personal protective equipment (PPE) detection devices are installed inside or on the surface of safety helmets, safety belts, and insulating gloves, using tension sensors, pressure sensors, or dielectric strength detection elements to acquire information on the wearing status. Tool identification devices use Bluetooth Low Energy tags on tools and cross-reference them with video acquisition data to verify tool identity and detect usage status. Environmental condition sensors are deployed in high-risk work areas to collect environmental parameters such as temperature, humidity, gas concentration, noise, and wind speed.

[0043] To ensure the fusion of multi-source data in a unified coordinate system, this invention employs Extended Kalman Filter (EKF) to jointly estimate position, pose, and video keypoints. The update formula is as follows:

[0044] in, Indicates the time when personnel are The state vector includes position, velocity, and attitude; Indicates IMU input quantity; , These are the nonlinear state transition function and the observation function, respectively. , These represent process noise and observation noise, respectively. When using a linear approximation, the state transition matrix... With observation matrix Through the , The Jacobian expansion is obtained; in highly nonlinear scenarios, it can be replaced by unscented Kalman filtering (UKF) to improve stability.

[0045] The edge intelligence layer, serving as the core of data processing and intelligent recognition in this invention's system, is primarily used for real-time fusion and intelligent analysis of data collected by the multi-source perception layer, enabling the recognition of worker actions and the judgment of interactive behaviors. This layer consists of an edge computing host, a lightweight deep learning model, and a cross-modal fusion and inference module.

[0046] In terms of data processing, the edge intelligence layer first uses Kalman filtering to perform spatiotemporal alignment of video frames, UWB positioning signals, IMU pose data, and tool Bluetooth signals, ensuring consistency across different data sources under a unified time reference. Subsequently, a keypoint detection network is used to extract human keypoint information from the video data and perform cross-modal alignment with the IMU pose data to form a complete motion feature vector.

[0047] In terms of behavior recognition, the edge intelligence layer models the collaborative relationships among workers through a multi-scale interactive attention mechanism. The calculation formula is as follows:

[0048] in, Indicates the first The first worker and the first Interaction weights between individual operators; , These are the query vector and key vector representing the personnel characteristics, respectively. For feature dimensions; Spatial distance between people; It is a monotonically decreasing function, used to represent spatial decay; To adjust the parameters, this formula incorporates spatial priors into the attention calculation, enabling the system to distinguish between close-range cooperative relationships and long-range interference relationships, thereby improving the accuracy of interaction recognition.

[0049] Combined with the causal reasoning module, the system can dynamically model the causal relationships between action events, thereby detecting skipped steps, out-of-order operations, or non-compliant operational behaviors.

[0050] The compliance reasoning layer, serving as the core of rule determination and logical analysis in this invention system, is primarily used to verify the compliance of actions and behaviors identified by the edge intelligence layer and output risk assessment results. This layer includes a probabilistic temporal logic rule engine, a job knowledge graph module, and a rule self-correction mechanism.

[0051] In terms of rule modeling, the compliance inference layer uses probabilistic temporal logic to formally constrain work steps. The system measures the compliance of key operational behaviors by setting probability thresholds. For example, in the scenario of working at height, when the personnel's working height exceeds two meters, the mooring action must be identified within a specified time window, and the compliance probability must be no less than 95%; in the scenario of closing the switch, the system requires that the operator and the reviewer are not the same person, and the review probability must be no less than 90%.

[0052] To facilitate the quantification of risk levels, this invention proposes a multi-factor weighted risk scoring method, the specific formula of which is as follows:

[0053] in, Indicates risk score; Indicates the probability of a violation; Indicates the probability of abnormality in personal protective equipment; Indicates the probability of the region going out of bounds; Indicates environmental hazard factors; , , , Let be the weight coefficient, and satisfy... .

[0054] Among them, the probability of behavioral violations It can be synthesized from the violation probabilities of multiple rules:

[0055] in, For a set of rules, For the first The probability of rule violation is calculated. This method ensures that the system can reasonably synthesize and quantify situations where multiple rules are in effect simultaneously.

[0056] In terms of knowledge support, the compliance reasoning layer establishes an operational knowledge graph, which includes process nodes, tool nodes, qualification nodes, and environment nodes. These nodes are connected through pre- and post-requirement relationships, dependencies, and mutual exclusion. For example, power outage confirmation must precede power-on confirmation; hot work and high-charge states are mutually exclusive; and live-line work depends on specific personnel qualifications and protective equipment. Through this knowledge graph, the system can perform differentiated rule verification in different operational scenarios.

[0057] The command and control layer, serving as the execution and feedback module of this invention, is primarily used to implement tiered responses and handling measures based on the judgment results of the compliance reasoning layer, thereby ensuring real-time safety of on-site operations. This layer includes a prompting submodule, a warning submodule, and a handling submodule.

[0058] During the notification phase, the command and control layer provides intuitive prompts to operators through wearable HUD displays or AR glasses, and can also combine vibration signals for reminders. This is suitable for situations with low risk scores and ensures that personnel can adjust their operating behavior in a timely manner.

[0059] During the warning phase, when the risk score is in the medium range, the system triggers the electronic fence warning and voice broadcast functions, using on-site audio-visual alarms to alert all workers to the risk and prevent further escalation of violations. The trigger range of the electronic fence can be flexibly set according to the spatial boundaries of the work area to ensure accurate warning coverage.

[0060] During the response phase, when the risk score reaches a high-level threshold, the command and control layer immediately implements emergency measures, including triggering emergency stop commands for the operating equipment, activating power-off protection, or mandating intervention from remote security systems. The response operation record is simultaneously transmitted to the higher-level monitoring center, forming a dual linkage mechanism between on-site and remote operations.

[0061] The evidence and audit layer, serving as the recording and traceability module of this invention, is primarily used for the solidification, storage, and traceability management of data, events, and judgment results during the operation process. This layer includes an evidence solidification unit, a timestamp and hash chain unit, and a violation log generation unit.

[0062] Regarding evidence preservation, the system segments and encapsulates the data collected and processed by the multi-source perception layer and edge intelligence layer according to time segments. Each segment includes video keyframes, positioning trajectories, pose sequences, identification tags, and compliance judgment results. This segmented storage approach ensures data integrity during long-term operation and facilitates rapid retrieval.

[0063] In terms of security and trustworthiness, the evidence and audit layer uses a hash algorithm to calculate a unique check value for each data segment and chains the hash values ​​of adjacent segments to form an immutable chain of evidence. The head of the chain and key nodes are signed using an authoritative timestamp service to ensure the time legitimacy and authority of the data.

[0064] To quantify the completeness and credibility of evidence, this invention proposes a two-factor credibility calculation method based on hash integrity and timestamp validity, as shown in the following formula:

[0065] in, Indicates the credibility score of the evidence; The integrity factor representing the hash verification result; A validity factor representing timestamp verification; , These are the weighting coefficients, and Furthermore, the integrity factor and timestamp factor can be defined continuously:

[0066] in, This indicates the number of data segments that passed the hash check. This indicates the number of data segments verified by authoritative timestamps. This represents the total number of data segments. When... At that time, the evidence was deemed highly credible.

[0067] Regarding violation recording, the system generates structured logs through a violation log generation unit. These logs include the type of violation, the time of occurrence, the identification of the personnel and tools involved, the triggered rule number, the causal path, and the system's response measures. The logs can be used for daily inspections at the team level, as well as for incident investigations and compliance audits, providing strong evidence for accountability and improvement.

[0068] In summary, the intelligent identification and compliance monitoring system for on-site operations in new energy maintenance proposed in this invention achieves comprehensive monitoring of personnel, tools, and environmental conditions through a multi-source perception layer. Combined with real-time data processing and interactive attention mechanisms in the edge intelligence layer, it effectively identifies complex operational behaviors and their interactions. Through probabilistic temporal logic and risk scoring models in the compliance inference layer, it achieves quantitative assessment and dynamic judgment of operational compliance. Furthermore, by combining the hierarchical response strategy in the command and linkage layer, it can promptly take prompts, warnings, or disposal measures at different risk levels to ensure on-site operational safety. Finally, by leveraging the hash chain and credibility calculation methods in the evidence and audit layer, it achieves tamper-proof evidence storage of operational data and generation of violation logs, providing reliable support for accident tracing and compliance auditing.

[0069] This system is architecturally structured as a closed loop, from low-level perception to high-level reasoning, and then to execution feedback and evidence consolidation, achieving full-process coverage of intelligent identification, compliance monitoring, and audit traceability for new energy operation and maintenance. Compared with existing technologies, this invention not only improves real-time performance and accuracy but also proposes innovative solutions for data credibility and compliance assurance. It effectively addresses issues such as insufficient monitoring accuracy, slow response, and incomplete evidence chains in existing technologies, demonstrating strong application value and promising prospects for widespread adoption.

[0070] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0071] Example 3 Based on the intelligent identification and compliance monitoring system for on-site operation behavior of new energy operation and maintenance in Embodiments 1 and 2, this embodiment provides a related monitoring system working method, including the following steps: The system monitors the personnel, tools, and environmental conditions at the maintenance site using video acquisition devices, position and attitude acquisition devices, intelligent personal protective equipment detection devices, tool identification devices, and environmental status sensing devices, and fuses the collected multi-source data in a unified coordinate system. The fused multi-source data is spatiotemporally aligned, and the spatiotemporally aligned data is used for action recognition and interaction behavior modeling analysis to obtain action recognition and interaction behavior judgment results. Based on the results of action recognition and interaction behavior judgment, a probabilistic temporal logic rule engine, a job knowledge graph, and a risk scoring mechanism are used to determine compliance and generate risk quantification results. Based on the risk quantification results and risk level classification response, implement prompts, warnings, and handling measures; Based on the risk quantification results, an immutable chain of evidence and compliance audit logs are formed to achieve evidence preservation and traceability throughout the entire process.

[0072] The specific steps have been described in detail in the modules of Embodiments 1 and 2, and will not be repeated here.

Claims

1. A smart identification and compliance monitoring system for on-site operation behavior in new energy operation and maintenance, characterized in that, It includes a multi-source perception layer for communication connectivity, an edge intelligence layer, a compliance reasoning layer, a command and control layer, and an evidence and audit layer; The multi-source sensing layer is used to monitor the status of personnel, tools and equipment and the operation and maintenance environment at the operation and maintenance site using multi-source acquisition devices, and to collect multi-source data, and to fuse the multi-source data in a unified coordinate system to obtain fused data. The edge intelligence layer is used to perform spatiotemporal alignment on the fused data output by the multi-source perception layer, and to perform action recognition and interactive behavior modeling analysis on the spatiotemporally aligned fused data to obtain action recognition results and interactive behavior judgment results. The compliance reasoning layer is used to make compliance judgments based on action recognition results and interaction behavior judgment results, and to form risk quantification results based on the judgment results; The command and coordination layer is used to respond in a graded manner based on the risk quantification results and the corresponding risk level, and to execute the corresponding graded response measures. The evidence and audit layer is used to form an immutable chain of evidence and compliance audit log based on the risk quantification results, so as to realize the evidence storage and traceability of the entire monitoring process.

2. The intelligent identification and compliance monitoring system for on-site operation behavior of new energy operation and maintenance as described in claim 1, characterized in that, Multi-source acquisition devices include at least video acquisition devices, position and attitude acquisition devices, intelligent personal protective equipment detection devices, tool identification devices, and environmental status sensing devices. The video acquisition device includes a fixed industrial camera or a pan-tilt camera, used to acquire images of the operator's movements; The position and attitude acquisition device includes a UWB positioning module and an IMU inertial measurement unit; the UWB positioning module is used to acquire the three-dimensional position of the operator in the work area in real time, and the IMU inertial unit is used to acquire the operator's attitude and motion trajectory. The intelligent personal protective equipment detection device is installed inside or on the surface of safety helmets, safety belts, and insulating gloves, and uses a detection device with tension and pressure sensors. The tool identification device identifies and detects the usage status of tools by attaching Bluetooth Low Energy tags to them and cross-referencing the data with video capture data. Environmental condition sensing devices include those deployed in high-risk work areas to acquire work environment parameters; After clock synchronization and signal calibration, each acquisition device achieves data acquisition under a unified time reference, and the joint estimation of position, attitude and image key points is achieved through the extended Kalman filter method.

3. The intelligent identification and compliance monitoring system for on-site operation behavior of new energy operation and maintenance as described in claim 1, characterized in that, The edge intelligence layer employs a lightweight deep learning model and combines a key point detection network with IMU pose data for cross-modal fusion to form a complete motion feature vector of the operator. The edge intelligence layer further introduces spatial prior factors through a multi-scale interactive attention mechanism, so that close-range collaborative actions are identified as positive interactions, while long-range interference actions are identified as abnormal behaviors.

4. The intelligent identification and compliance monitoring system for on-site operation behavior of new energy operation and maintenance as described in claim 3, characterized in that, The multi-scale interactive attention mechanism is as follows: Among them, among them, Indicates the first The first worker and the first Interaction weights between individual operators; , These are the query vector and key vector representing the personnel characteristics, respectively. For feature dimensions; Spatial distance between people; It is a monotonically decreasing function, used to represent spatial decay; To adjust the parameters.

5. The intelligent identification and compliance monitoring system for on-site operation behavior of new energy operation and maintenance as described in claim 1, characterized in that, The compliance inference layer formally constrains the behavior and state sequences in the action recognition results through a probabilistic temporal logic rule engine. The formal constraints include: in high-altitude operations, when the personnel's working height exceeds a set threshold, the safety belt buckle action must be completed within a limited time window; in the closing operation scenario, the identities of the operator and the reviewer must be different; in the hot work scenario, the high-charge state and the hot work are defined as mutually exclusive.

6. The intelligent identification and compliance monitoring system for on-site operation behavior of new energy operation and maintenance according to claim 5, characterized in that, The compliance reasoning layer infers behavioral results and potential risks based on the work knowledge graph. The knowledge graph includes process nodes, tool nodes, qualification nodes, and environment nodes. These nodes are connected through pre- and post-relationships, dependencies, and mutual exclusions, thereby enabling differentiated rule verification in different work scenarios.

7. The intelligent identification and compliance monitoring system for on-site operation behavior of new energy operation and maintenance according to claim 6, characterized in that, In the compliance reasoning layer, compliance is determined through a risk scoring mechanism, which is as follows: in, Indicates risk score; Indicates the probability of a violation; Indicates the probability of abnormality in personal protective equipment; Indicates the probability of the region going out of bounds; Indicates environmental hazard factors; , , , Let be the weight coefficient, and satisfy... .

8. The intelligent identification and compliance monitoring system for on-site operation behavior of new energy operation and maintenance as described in claim 1, characterized in that, When the risk score is low, the command and control layer will provide a prompt through a wearable display device or vibration device; when the risk score is medium, it will trigger an audible and visual alarm and a voice broadcast through an electronic fence to issue a warning; and when the risk score is high, it will trigger an emergency stop command for the operating equipment, link the power-off protection, or force intervention through the remote security system, and simultaneously transmit the handling record to the superior monitoring center.

9. The intelligent identification and compliance monitoring system for on-site operation behavior of new energy operation and maintenance according to claim 1, characterized in that, The evidence and audit layer stores the operation data in a segmented encapsulation manner. Each data segment includes video keyframes, positioning trajectories, posture sequences, identification tags, and compliance judgment results. A unique verification value is calculated for each data segment, and adjacent data segments are connected sequentially using a chain hash method.

10. A working method for the intelligent identification and compliance monitoring system for on-site operation behavior of new energy operation and maintenance as described in any one of claims 1 to 9, characterized in that, include: The system monitors the personnel, tools, and environmental conditions at the maintenance site using video acquisition devices, position and attitude acquisition devices, intelligent personal protective equipment detection devices, tool identification devices, and environmental status sensing devices, and fuses the collected multi-source data in a unified coordinate system. The fused multi-source data is spatiotemporally aligned, and the spatiotemporally aligned data is used for action recognition and interaction behavior modeling analysis to obtain action recognition and interaction behavior judgment results. Based on the results of action recognition and interaction behavior judgment, compliance assessment is conducted and risk quantification results are generated; Based on the risk quantification results and risk level classification response, implement the corresponding graded response measures. Based on the risk quantification results, an immutable chain of evidence and compliance audit logs are formed to achieve evidence preservation and traceability throughout the monitoring process.

Images