Power metering on-site operation safety monitoring and alarm system

Abstract

The present invention provides a safety monitoring and alarm system for power metering field operations, designed in the field of monitoring and early warning technology. The system includes: an acquisition module for collecting an initial video stream of the power operation site based on the safety helmets worn by workers, and multiple sensors configured to sense the power and environmental parameters of the power operation site in real time; a compression module for compressing the collected data at each moment and transmitting it to a backend; a backend for setting a sliding window to lock the received compressed data at a specific moment, analyzing the hazard type and hazard trigger probability of the first processed data at the locked moment; and further for chronologically arranging all hazard types and hazard trigger probabilities, generating a reminder instruction, and issuing it to the alarm module of the corresponding safety helmet for reminder. Data compression and hazard analysis ensure the reliability of safety warnings, facilitating effective management and effectively ensuring personnel safety.

Description

Technical Field

[0001] The present invention relates to the technical field of monitoring and early warning, and in particular to a power metering on-site operation safety monitoring and alarm system. Background Art

[0002] Power operations are complex and fraught with hidden risks. Traditional safety measures for power operations rely heavily on manual experience and limited on-site monitoring equipment, making it difficult to achieve real-time, comprehensive, and accurate risk monitoring and early warning.

[0003] On the one hand, abnormal changes in power parameters, such as sudden voltage spikes and dips, and current overloads, can trigger equipment failures or even electrical accidents without warning, threatening the safety of workers and the stable operation of the power system. Environmental parameters also have a significant impact. For example, excessive humidity can cause electrical equipment to short-circuit, and strong winds can damage outdoor power facilities. Traditional methods struggle to capture subtle changes in these environmental factors and their potential impact on operations.

[0004] On the other hand, manual observation of operational processes is not only inefficient but also prone to overlooking critical risks due to negligence. For example, safety hazards such as improper operation at heights and improper placement of tools and equipment on-site are difficult to detect and correct in a timely manner. Furthermore, centralized management and real-time oversight are even more challenging when power operations are spread across large areas and with dispersed personnel.

[0005] Therefore, the present invention proposes a power metering on-site operation safety monitoring and alarm system. Summary of the Invention

[0006] The present invention provides a safety monitoring and alarm system for power metering on-site operations, which is used to monitor the operation of the workers themselves by setting up safety helmets and combining the monitoring of power parameters and environmental parameters to analyze the data. The reliability of safety warnings is then guaranteed through subsequent data compression and hazard analysis processing, which not only facilitates effective management but also effectively ensures personnel safety.

[0007] The present invention provides a power metering on-site operation safety monitoring and alarm system, comprising:

[0008] The acquisition module is used to collect the initial video stream of the power operation site based on the safety helmets worn by the workers, and at the same time, configure multiple sensors to sense the power parameters and environmental parameters of the power operation site in real time;

[0009] A compression module, configured to compress the initial video stream and the perceived power parameters and environmental parameters at each moment and transmit them to a backend;

[0010] The backend is used to set a sliding window to lock the received compressed processed data at a certain time, and analyze the danger type and danger trigger probability of the first processed data at the locking time;

[0011] The backend is also used to arrange all hazard types and hazard triggering probabilities in chronological order, generate reminder instructions, and send them to the alarm module of the corresponding safety helmet for reminder.

[0012] Preferably, the safety helmet comprises:

[0013] A high-definition camera, used to collect an initial video stream of the power operation site;

[0014] The camera bracket is arranged on the helmet body and is used for placing the high-definition camera and adjusting the shooting angle of the high-definition camera.

[0015] Preferably, the compression module includes:

[0016] a cycle determination unit, configured to determine the operation type of the power operation site and obtain a set collection cycle by matching the operation type from a type-cycle database;

[0017] a stamp assigning unit, configured to, when the actual acquisition time is consistent with the set acquisition period, set a timestamp for the initial video stream, power parameters, and environmental parameters and sort them in sequence to obtain sub-data at the same time;

[0018] an action recognition unit, configured to perform action recognition on the initial video stream based on an action recognition algorithm, assign an action state and an action detail description to each frame image at each moment, perform a first segmentation on the initial video stream according to the action state, determine action importance of the video in the first segmentation according to the action detail description, and determine an importance coefficient of each video in the first segmentation;

[0019] The compression and transmission unit is used to automatically compress and transmit the sub-data at each moment of the actual collection time according to the compression multiple.

[0020] Preferably, the behavior recognition unit includes:

[0021] A first calculation unit is used to calculate the action importance of the video corresponding to the first division based on the action state and the action detail description;

[0022]

[0023] A second calculation unit is used to calculate a corresponding importance coefficient based on the action importance;

[0024]

[0025] in, Represents the important coefficient of the video under the z-th first partition; represents the action importance of the video under the z-th first partition; represents the maximum importance of the action importance of all videos under the first division; ln represents the sign of the logarithmic function; represents the amount of information of the video under the z-th first partition; Represents the total information content of all videos under the first division; Indicates the total information storage capacity of the storage device installed on the helmet; Indicates the number of videos under the first partition; Indicates action-based state The state coefficient of is in the range of (0, 1); represents the number of action features involved in the action details description of the video under the z-th first partition; Represents the total number of action features of all videos under the first division; Indicates based on Analytical function of .

[0026] Preferably, it also includes:

[0027] A duration statistics module, configured to count the first time point at which the compression module sends the compressed data to the backend and the second time point at which the backend sends the arrival instruction to the acquisition module and the acquisition module receives the instruction, to obtain the round-trip communication duration;

[0028] A control module is configured to control the high-definition camera and various sensors to continue working when the acquisition module receives a reach instruction if the round-trip communication duration is less than the waiting duration corresponding to the set acquisition cycle;

[0029] Otherwise, after the waiting period is completed, the high-definition camera and the various sensors are controlled to continue working.

[0030] Preferably, the backend includes:

[0031] a window determination unit, configured to determine a first number of communication connections established by the backend with external helmets at the same time, and set a window size for each helmet based on a transmission rate of a transmission channel for communication between each helmet and the backend and a set scene weight of a power operation scene at the location of the helmet;

[0032]

[0033] in, Indicates the window size of the i-th helmet; represents the set size of the i-th helmet; Indicates the rounding symbol; Indicates the transmission data size of the i-th helmet; represents the first quantity; Indicates the maximum number of connections between the backend and the helmet; represents the set transmission rate of the i-th helmet; Indicates the maximum transmission rate of the backend; Indicates the warning threshold for the amount of data received by the backend for synchronous processing;

[0034] The time locking unit is used to lock the time of the compressed data received from the corresponding helmet according to the window size, wherein the window size is consistent with the time number.

[0035] Preferably, the backend further includes:

[0036] A first subclass determination unit is configured to determine all action states involved in the first processed data, and obtain a first hazard subclass corresponding to the first processed data by matching the action states from a combination state-type comparison table;

[0037] a second subclass determination unit, configured to determine a second dangerous subclass based on the monitoring data of the sensor involved in the first processed data and compare it with standard data;

[0038] The probability determination unit is configured to obtain the hazard type and hazard trigger probability matching the first hazard subclass and the second hazard subclass from a subclass-type-probability comparison table.

[0039] Preferably, the backend further includes:

[0040] A vector acquisition unit is used to arrange all hazard types and hazard trigger probabilities in chronological order to obtain an analysis vector;

[0041] The instruction acquisition unit is used to input the analysis vector into the vector analysis model to generate a reminder instruction.

[0042] Compared with the prior art, the present invention has the following advantages:

[0043] Starting from the staff themselves, safety helmets are set up for operation monitoring and combined with the monitoring of power parameters and environmental parameters to analyze the data. Then, data compression and hazard analysis are carried out to ensure the reliability of safety warnings. This not only facilitates effective management, but also effectively ensures personnel safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0045] Figure 1 This is a structural diagram of a power metering on-site operation safety monitoring and alarm system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0046] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0047] The present invention provides a power metering field operation safety monitoring and alarm system, such as Figure 1 Shown, including:

[0048] The acquisition module is used to collect the initial video stream of the power operation site based on the safety helmets worn by the workers, and at the same time, configure multiple sensors to sense the power parameters and environmental parameters of the power operation site in real time;

[0049] A compression module, configured to compress the initial video stream and the perceived power parameters and environmental parameters at each moment and transmit them to a backend;

[0050] The backend is used to set a sliding window to lock the received compressed processed data at a certain time, and analyze the danger type and danger trigger probability of the first processed data at the locking time;

[0051] The backend is also used to arrange all hazard types and hazard triggering probabilities in chronological order, generate reminder instructions, and send them to the alarm module of the corresponding safety helmet for reminder.

[0052] In this embodiment, multiple sensors are used to sense devices with different physical quantities, such as current sensors and voltage sensors to sense power parameters; temperature and humidity sensors and wind speed sensors to sense environmental parameters, for example, the current line current is 50A, the site temperature is 30°C, and the humidity is 60%RH.

[0053] In this embodiment, the safety helmet worn by the worker is a protective equipment worn on the head of the power worker, and is given the function of collecting the initial video stream to record the working site situation.

[0054] In this embodiment, the initial video stream is continuous video data captured and generated by a camera device on a hard hat without any processing, recording the worker's behavior of operating tools, climbing electric poles, etc., as well as the surrounding environment.

[0055] In this embodiment, power parameters are physical quantities used to describe the operating status of the power system, and common ones include voltage, current, power, frequency, etc.

[0056] In this embodiment, the environmental parameters are physical quantities that reflect the environmental conditions of the work site, such as temperature, humidity, wind speed, air pressure, etc.

[0057] In this embodiment, the backend generally refers to a remote data processing center or server that receives and processes data from the work site.

[0058] In this embodiment, the compression module compresses the initial video stream using a video compression algorithm such as H.264 and applies appropriate encoding compression methods to the power and environmental parameter data. For example, a one-minute initial video stream originally sized at 10MB is compressed to 2MB, and the power and environmental parameter data originally sized at 1KB is compressed to 500B. After compression, the compressed data is rapidly transmitted to a backend server via a 4G or 5G network.

[0059] In this embodiment, the sliding window is fixed to a certain time period.

[0060] In this embodiment, the types of hazards include but are not limited to electrical hazards (leakage, short circuit, etc.) and environmental hazards (risk of heat stroke due to high temperature, risk of objects falling due to strong winds, etc.).

[0061] In this embodiment, the danger trigger probability is an evaluation of the likelihood of a certain type of danger actually occurring at the current locking moment.

[0062] In this embodiment, a 5-second sliding window is set at the backend, and the window continues to slide as time goes by. When the window slides to the time period of 10 minutes and 30 seconds to 10 minutes and 35 seconds and is locked. At this time, the compressed processed data in the window is extracted as the first processed data, including several frames of video images within these 5 seconds, corresponding power parameters (such as the voltage suddenly drops below the warning value) and environmental parameters (such as the wind speed reaches level 8). The backend determines that there are electrical hazards (device failure may occur due to abnormal voltage) and environmental hazards (strong winds may cause tools at high altitudes to fall), and concludes that the probability of danger triggering is 60%.

[0063] In this embodiment, the alarm module can generate audible and visual alarms. For example, if the probability of a hazard detected at the 30th second is 60%, then the probability of a hazard detected at the 40th second is 30%, and so on. Based on this information, a warning instruction is generated, such as "Electrical hazard ahead, trigger probability 60%, please stop operation immediately and check the wiring." This instruction is transmitted wirelessly to the alarm module on the operator's helmet. Upon receiving the instruction, the alarm module emits a loud alarm and flashes a red light to remind the operator to pay attention to safety.

[0064] The beneficial effect of the above technical solution is: starting from the staff themselves, they set up safety helmets to monitor the work and combine the monitoring of power parameters and environmental parameters to analyze the data, and then subsequently use data compression and hazard analysis processing to ensure the reliability of safety warnings, which not only facilitates effective management, but also effectively ensures personnel safety.

[0065] The present invention provides a safety monitoring and alarm system for power metering on-site operations, wherein the safety helmet comprises:

[0066] A high-definition camera, used to collect an initial video stream of the power operation site;

[0067] The camera bracket is arranged on the helmet body and is used for placing the high-definition camera and adjusting the shooting angle of the high-definition camera.

[0068] In this embodiment, the camera bracket on the helmet adopts a rotatable and tiltable design and can be any existing structure.

[0069] In this embodiment, when a worker needs to observe overhead wiring connections, they can adjust the camera's angle upward by rotating it, for example, from the initial horizontal angle to a 45-degree angle. This allows them to clearly capture details such as the tightening of bolts at the wiring connections and the presence of electrical discharges. To check the placement of tools and the condition of the work platform below, they can tilt the camera downward to change the angle to capture the desired image.

[0070] The beneficial effect of the above technical solution is that by providing an adjustable camera bracket, the applicability of the high-definition camera at the power operation site and the comprehensiveness of data collection are greatly improved.

[0071] The present invention provides a power metering on-site operation safety monitoring and alarm system, wherein the compression module comprises:

[0072] a cycle determination unit, configured to determine the operation type of the power operation site and obtain a set collection cycle by matching the operation type from a type-cycle database;

[0073] a stamp assigning unit, configured to, when the actual acquisition time is consistent with the set acquisition period, set a timestamp for the initial video stream, power parameters, and environmental parameters and sort them in sequence to obtain sub-data at the same time;

[0074] an action recognition unit, configured to perform action recognition on the initial video stream based on an action recognition algorithm, assign an action state and an action detail description to each frame image at each moment, perform a first segmentation on the initial video stream according to the action state, determine action importance of the video in the first segmentation according to the action detail description, and determine an importance coefficient of each video in the first segmentation;

[0075] The compression and transmission unit is used to automatically compress and transmit the sub-data at each moment of the actual collection time according to the compression multiple.

[0076] Preferably, the behavior recognition unit includes:

[0077] A first calculation unit is used to calculate the action importance of the video corresponding to the first division based on the action state and the action detail description;

[0078]

[0079] A second calculation unit is used to calculate a corresponding importance coefficient based on the action importance;

[0080]

[0081] in, Represents the important coefficient of the video under the z-th first partition; represents the action importance of the video under the z-th first partition; represents the maximum importance of the action importance of all videos under the first division; ln represents the sign of the logarithmic function; represents the amount of information of the video under the z-th first partition; Represents the total information content of all videos under the first division; Indicates the total information storage capacity of the storage device installed on the helmet; Indicates the number of videos under the first partition; Indicates action-based state The state coefficient of is in the range of (0, 1); represents the number of action features involved in the action details description of the video under the z-th first partition; Represents the total number of action features of all videos under the first division; Indicates based on Analytical functions of Represents the factorial symbol.

[0082] In this embodiment, the operation types include but are not limited to substation equipment inspection, transmission line maintenance, power equipment installation, etc.

[0083] In this embodiment, the type-period database is a pre-established database that stores information correlating various power operation types with corresponding collection periods. This database is constructed based on factors such as past operational experience, safety regulations, and actual needs. For example, for substation equipment inspections, the collection period is set to 5 minutes. However, for certain operations with higher risks or complex operational procedures, a shorter collection period may be set.

[0084] In this embodiment, the sub-data is a data set integrating the initial video stream, power parameters, and environmental parameters collected at the same time.

[0085] In this embodiment, the behavior recognition algorithm is an algorithm based on computer vision and machine learning technology, which can identify the specific actions of the power worker from the initial video stream, such as bending, reaching, climbing, etc.

[0086] In this embodiment, the action status categorizes and describes the identified actions, such as normal operation, illegal operation, dangerous action, and other states.

[0087] In this embodiment, the action details description is to describe the specific details of the action, including the object of the action, the amplitude of the action, the speed of the action, etc. For example, "reach out to pick up the wrench, with the arm extended approximately 60 degrees."

[0088] In this embodiment, the initial video stream is segmented according to the action state, and video segments with the same action state are divided into a group.

[0089] In this embodiment, action importance: evaluates the importance of the action contained in each first partitioned video, with the higher the importance, the better.

[0090] In this embodiment, based on algorithm analysis, it is determined that the action state is "normal operation", and the action details are described as "right hand stretched out, slowly approaching the device housing, and the touch position is the upper left corner of the front of the device". As the video stream is continuously analyzed, all video clips with the "normal operation" action state are divided into a first division. For the videos in this first division, the first calculation unit calculates the action importance based on the action state and the action details description. For example, considering that this is a routine equipment inspection action, the action importance is relatively moderate, and the formula is used to calculate that the importance coefficient of the video in the first division is 0.6.

[0091] In this embodiment, a video compression algorithm (such as H.264) compresses the initial video stream, and appropriate encoding and compression methods are used for the power and environmental parameter data to reduce data storage space and transmission bandwidth. For example, a sub-data set containing a 5-minute operation video stream and related power and environmental parameters, originally 100MB in size, is reduced to 20MB after compression. After compression, this compressed sub-data is transmitted to a backend server via the 5G network. The backend server receives the data and can store and further analyze it, such as checking whether the operator's operation is standard and whether the power parameters are abnormal.

[0092] The beneficial effects of the above technical solution are: reducing data storage space and transmission bandwidth based on the determination of the compression multiple, effectively ensuring the simultaneous processing of multiple safety helmet data on the background end, realizing timely management of personnel safety and early warning, and more accurately quantifying the importance of the video based on the calculation of the importance coefficient, so as to make decisions in subsequent data processing.

[0093] The present invention provides a power metering on-site operation safety monitoring and alarm system, which also includes:

[0094] A duration statistics module, configured to count the first time point at which the compression module sends the compressed data to the backend and the second time point at which the backend sends the arrival instruction to the acquisition module and the acquisition module receives the instruction, to obtain the round-trip communication duration;

[0095] A control module is configured to control the high-definition camera and various sensors to continue working when the acquisition module receives a reach instruction if the round-trip communication duration is less than the waiting duration corresponding to the set acquisition cycle;

[0096] Otherwise, after the waiting period is completed, the high-definition camera and the various sensors are controlled to continue working.

[0097] In this embodiment, the first time point is the starting moment when the compression module starts to send compressed data to the backend; the second time point is the moment when the acquisition module receives the arrival instruction sent by the backend.

[0098] In this embodiment, after compressing the data, the compression module begins sending the compressed data to the backend at 10:05:00 AM (the first time point). After receiving and analyzing the data, the backend sends a "reach" command to the acquisition module at 10:05:10 AM. The acquisition module receives this command at 10:05:12 AM (the second time point). The duration statistics module then calculates the round-trip communication duration to be 12 seconds (from 10:05:00 to 10:05:12).

[0099] The beneficial effect of the above technical solution is that, by determining the first time point and the second time point, the orderly collection of information by the helmet is effectively guaranteed.

[0100] The present invention provides a power metering field operation safety monitoring and alarm system, the backend comprising:

[0101] a window determination unit, configured to determine a first number of communication connections established by the backend with external helmets at the same time, and set a window size for each helmet based on a transmission rate of a transmission channel for communication between each helmet and the backend and a set scene weight of a power operation scene at the location of the helmet;

[0102]

[0103] in, Indicates the window size of the i-th helmet; represents the set size of the i-th helmet; Indicates the rounding symbol; Indicates the transmission data size of the i-th helmet; represents the first quantity; Indicates the maximum number of connections between the backend and the helmet; represents the set transmission rate of the i-th helmet; Indicates the maximum transmission rate of the backend; Indicates the warning threshold for the amount of data received by the backend for synchronous processing;

[0104] The time locking unit is used to lock the time of the compressed data received from the corresponding helmet according to the window size, wherein the window size is consistent with the time number.

[0105] Preferably, the backend further includes:

[0106] A first subclass determination unit is configured to determine all action states involved in the first processed data, and obtain a first hazard subclass corresponding to the first processed data by matching the action states from a combination state-type comparison table;

[0107] a second subclass determination unit, configured to determine a second dangerous subclass based on the monitoring data of the sensor involved in the first processed data and compare it with standard data;

[0108] The probability determination unit is configured to obtain the hazard type and hazard trigger probability matching the first hazard subclass and the second hazard subclass from a subclass-type-probability comparison table.

[0109] Preferably, the backend further includes:

[0110] A vector acquisition unit is used to arrange all hazard types and hazard trigger probabilities in chronological order to obtain an analysis vector;

[0111] The instruction acquisition unit is used to input the analysis vector into the vector analysis model to generate a reminder instruction.

[0112] In this embodiment, power operation scenarios include but are not limited to substation maintenance, transmission line maintenance, and distribution room operations. Different scenarios vary in complexity, risk, and data volume requirements. For high-risk live-line operation scenarios, a higher scenario weight, such as 0.8, may be assigned; whereas for relatively low-risk equipment cleaning scenarios, a lower scenario weight, such as 0.4, may be assigned.

[0113] In this embodiment, for example, if the window size is 4, data at 4 moments are locked.

[0114] In this embodiment, due to the limitations of server performance and network bandwidth, the backend can establish connections with up to 50 safety helmets at the same time. The network bandwidth and server processing capacity of the backend determine its maximum transmission rate to be 1000Mbps. The backend sets the warning threshold to receive 800MB of data per second.

[0115] In this embodiment, the combination state-type mapping table is a pre-established database table that stores the mapping relationship between different action state combinations and corresponding hazard subcategories. For example, the action state combination of "reaching for a tool" and "near live equipment" may correspond to the "approaching live equipment" hazard subcategory in the mapping table.

[0116] In this embodiment, the second hazard subcategory, for example, where the monitored voltage exceeds the standard range, may correspond to the "abnormal voltage hazard subcategory." The voltage sensor on the helmet detects that the current device voltage is 240V, while the standard voltage range for this device is 220V±5V. The second subcategory determination unit compares the monitored data of 240V with the standard data and finds that the voltage exceeds the standard range, thus determining the second hazard subcategory as the "excessive voltage hazard subcategory."

[0117] In this embodiment, the subclass-type-probability comparison table is a database table that records the association between different hazard subclasses (first hazard subclass and second hazard subclass) and corresponding hazard types (such as electrical hazards, environmental hazards, etc.) and hazard trigger probabilities (a numerical value indicating the possibility of a hazard occurring, usually expressed as a percentage).

[0118] Assume that the first hazard subcategory determined by the first subcategory determination unit is "direct contact with live parts hazard subcategory," and the second hazard subcategory determined by the second subcategory determination unit is "overvoltage hazard subcategory." A lookup in the subcategory-type-probability comparison table reveals that the electrical hazard probability for the corresponding segments for "direct contact with live parts hazard subcategory" and "overvoltage hazard subcategory" is 30%.

[0119] In this embodiment, the analysis vector = {danger types and danger triggering probabilities arranged in chronological order}.

[0120] In this embodiment, the vector analysis model is obtained by training the neural network model based on the combination of different hazard types and hazard trigger probabilities and reminder instructions for the combination as samples. The training samples are greater than 1,000 examples, and the reminder instructions are such as "The current electrical hazard level has increased. Please pay attention to check the connections of electrical equipment and avoid staying in dangerous areas for a long time."

[0121] The beneficial effects of the above technical solution are: the window setting facilitates the analysis of whether the operation is standardized, whether the power parameters are abnormal, and whether the environmental conditions are safe. The constructed analysis vector can intuitively display the danger information at different time points, providing a data basis for subsequent analysis. The backend sends this reminder instruction to the corresponding safety helmet. After receiving the instruction, the operator can promptly understand the operation risks and take corresponding preventive measures.

[0122] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A power metering on-site operation safety monitoring and alarm system, characterized in that: include: The acquisition module is used to collect the initial video stream of the power operation site based on the safety helmets worn by the workers, and at the same time, configure multiple sensors to sense the power parameters and environmental parameters of the power operation site in real time; A compression module, configured to compress the initial video stream and the perceived power parameters and environmental parameters at each moment and transmit them to a backend; The backend is used to set a sliding window to lock the received compressed processed data at a certain time, and analyze the danger type and danger trigger probability of the first processed data at the locking time; The backend is also used to arrange all hazard types and hazard trigger probabilities in chronological order, generate reminder instructions, and send them to the alarm module of the corresponding helmet for reminder; The backend includes: a window determination unit, configured to determine a first number of communication connections established by the backend with external helmets at the same time, and set a window size for each helmet based on a transmission rate of a transmission channel for communication between each helmet and the backend and a set scene weight of a power operation scene at the location of the helmet; in, Indicates the window size of the i-th helmet; represents the set size of the i-th helmet; Indicates the rounding symbol; Indicates the transmission data size of the i-th helmet; represents the first quantity; Indicates the maximum number of connections between the backend and the helmet; represents the set transmission rate of the i-th helmet; Indicates the maximum transmission rate of the background end; Indicates the warning threshold for the amount of data received by the backend for synchronous processing; The time locking unit is used to lock the time of the compressed data received from the corresponding helmet according to the window size, wherein the window size is consistent with the time number.

2. The power metering on-site operation safety monitoring and alarm system according to claim 1 is characterized in that: The safety helmet comprises: A high-definition camera, used to collect an initial video stream of the power operation site; The camera bracket is arranged on the helmet body and is used for placing the high-definition camera and adjusting the shooting angle of the high-definition camera.

3. The power metering on-site operation safety monitoring and alarm system according to claim 1 is characterized in that: The compression module comprises: a cycle determination unit, configured to determine the operation type of the power operation site and obtain a set collection cycle by matching the operation type from a type-cycle database; a stamp assigning unit, configured to, when the actual acquisition time is consistent with the set acquisition period, set a timestamp for the initial video stream, power parameters, and environmental parameters and sort them in sequence to obtain sub-data at the same time; an action recognition unit, configured to perform action recognition on the initial video stream based on an action recognition algorithm, assign an action state and an action detail description to each frame image at each moment, perform a first segmentation on the initial video stream according to the action state, determine action importance of the video in the first segmentation according to the action detail description, and determine an importance coefficient of each video in the first segmentation; The compression and transmission unit is used to automatically compress and transmit the sub-data at each moment of the actual collection time according to the compression multiple.

4. The power metering on-site operation safety monitoring and alarm system according to claim 3 is characterized in that: The behavior recognition unit includes: A first calculation unit is used to calculate the action importance of the video corresponding to the first division based on the action state and the action detail description; A second calculation unit is used to calculate a corresponding importance coefficient based on the action importance; in, Represents the important coefficient of the video under the z-th first partition; represents the action importance of the video under the z-th first partition; represents the maximum importance of the action importance of all videos under the first division; ln represents the sign of the logarithmic function; represents the amount of information of the video under the zth first partition; Represents the total information content of all videos under the first division; Indicates the total information storage capacity of the storage device installed on the helmet; Indicates the number of videos under the first partition; Indicates action-based state The state coefficient of is in the range of (0, 1); represents the number of action features involved in the action details description of the video under the z-th first partition; Represents the total number of action features of all videos under the first division; Indicates based on Analytical function of .

5. The power metering on-site operation safety monitoring and alarm system according to claim 1 is characterized in that: Also includes: A duration statistics module, configured to count the first time point at which the compression module sends the compressed data to the backend and the second time point at which the backend sends the arrival instruction to the acquisition module and the acquisition module receives the instruction, to obtain the round-trip communication duration; A control module is configured to control the high-definition camera and various sensors to continue working when the acquisition module receives a reach instruction if the round-trip communication duration is less than the waiting duration corresponding to the set acquisition cycle; Otherwise, after the waiting period is completed, the high-definition camera and the various sensors are controlled to continue working.

6. The power metering on-site operation safety monitoring and alarm system according to claim 1 is characterized in that: The backend further includes: A first subclass determination unit is configured to determine all action states involved in the first processed data, and obtain a first dangerous subclass corresponding to the first processed data by matching from a combination state-type comparison table; a second subclass determination unit, configured to determine a second dangerous subclass based on the monitoring data of the sensor involved in the first processed data and compare it with standard data; The probability determination unit is configured to obtain the hazard type and hazard trigger probability matching the first hazard subclass and the second hazard subclass from a subclass-type-probability comparison table.

7. The power metering on-site operation safety monitoring and alarm system according to claim 6 is characterized in that: The backend further includes: A vector acquisition unit is used to arrange all hazard types and hazard trigger probabilities in chronological order to obtain an analysis vector; The instruction acquisition unit is used to input the analysis vector into the vector analysis model to generate a reminder instruction.

Images