Multi-compartment NVR (Network Video Recorder) state reporting and cooperative control system based on multicast communication

Abstract

The invention relates to the technical field of cooperative control, in particular to a multi-compartment NVR (Network Video Recorder) state reporting and cooperative control system based on multicast communication, which comprises a state sensing aggregation module, a communication load evaluation module, a cache scheduling release module, a synchronous boundary correction module and a cooperative control configuration module. According to the invention, the node group in the frequently activated state can be identified through the normalization quantization processing of the starting frequency of the video channel in the carriage, the reading of the temperature sensor and the writing rate of the hard disk, and the load evaluation is carried out by combining the message data volume and the channel transmission capability of the node in the historical multicast period. A conflict early warning mark can be formed before a state conflict occurs, correction of time migration among multiple nodes is completed based on comparison of a node receiving timestamp difference value and a clock synchronization tolerance, a control instruction set for time alignment is constructed based on a unified time correction result, a trigger control task is broadcast and executed in a time window packaging mode, and the time correction efficiency is improved. And the communication resource use efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of collaborative control technology, and in particular to a multi-carriage NVR status reporting and collaborative control system based on multicast communication. Background Art

[0002] The field of collaborative control technology focuses on achieving overall system coordination, collaboration, and global performance optimization through information exchange and control strategy design between individuals in multi-agent systems. This field covers multiple aspects, including system architecture design, information communication mechanisms, control algorithm construction, and stability analysis. It is widely used in scenarios such as automated train systems, intelligent transportation, distributed robotics, multi-drone formations, and industrial process control. In a collaborative control system, each node typically possesses autonomous perception, computing, and decision-making capabilities. Relying on information exchange under limited bandwidth, it achieves the collaborative execution of target tasks such as synchronous control, formation maintenance, and resource scheduling, thereby improving the overall robustness and efficiency of the system.

[0003] The multi-carriage NVR status reporting and collaborative control system is a specific application of collaborative control in multi-node equipment management on rail transit vehicles. It aims to enable information sharing, unified status monitoring, and coordinated command control among network video recording devices in multiple carriages of a train. Through an information exchange protocol between the master and slave carriages, the system enables each carriage's NVR device to automatically complete functions such as status reporting, fault response, parameter adjustment, and recording task scheduling during train operation, enhancing the overall stability and management efficiency of the train monitoring system.

[0004] In the information exchange between the NVRs of multiple carriages, the traditional control system uses a single node to report and broadcast step by step to synchronize status and transmit task instructions. It does not distinguish and count the status of frequently active nodes, which leads to data channel congestion during information-intensive periods, increased message queuing delays and lack of priority control, and reduced real-time status reporting and the timeliness of key node responses. In addition, the clock synchronization offset of each carriage equipment is not dynamically monitored and corrected, and the execution time of control instructions between different nodes drifts, affecting the consistency and integrity of the synchronized recording tasks. For example, when an emergency occurs during train operation and recording needs to be started synchronously, the response time of each carriage equipment varies, affecting the accuracy of unified scheduling and the continuity of event recording. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and propose a multi-car NVR status reporting and collaborative control system based on multicast communication.

[0006] In order to achieve the above objectives, the present invention adopts the following technical solution: a multi-carriage NVR status reporting and collaborative control system based on multicast communication, the system comprising: The state perception aggregation module obtains the video channel activation flag, temperature sensor reading, and hard disk write rate in the carriage NVR, calculates the load index score, combines the channel activation frequency, identifies the high-frequency state source group, and generates an active state node index set; The communication load evaluation module calls the total number of message data bytes of the corresponding node in the previous multicast cycle and the maximum transmission unit of the multicast channel based on the active state node index set, determines whether it has entered the warning interval, and generates a status reporting conflict risk mark; The cache scheduling and release module calls the status message residence time, cache priority flag and hard disk health status value of the active state node in the cache queue according to the status reporting conflict risk mark, filters the message data group whose residence time exceeds the set window, sorts them in descending order according to the hard disk health status value, and generates a multicast status release scheduling list; The synchronization boundary correction module calls the multicast state release scheduling list, compares the timestamp difference of each node in the same group with the set clock synchronization tolerance threshold, corrects it based on the timestamp of the network time protocol master time source, and generates cross-node synchronization window alignment parameters.

[0007] The present invention has the following improvements: the active state node index set includes a channel index sequence number, an activation frequency label, a load score label and a node active identifier; the status reporting conflict risk mark is specifically a communication warning label, a node load level, a cycle occupancy ratio and a conflict judgment result; the multicast state release scheduling list includes a message priority index, a release sorting sequence, a cache residency flag and a health status label; the cross-node synchronization window alignment parameters include a node time offset, a synchronization adjustment amount, a calibration reference value and a tolerance judgment label.

[0008] The present invention is improved in that the state awareness aggregation module includes: The channel activation identification submodule obtains the video channel activation flag in the carriage NVR, classifies and counts each activation flag according to the channel sequence, records the number of times the same channel number is activated in a unit period, and constructs an activation frequency sequence to obtain channel activation frequency distribution information; The load index conversion submodule calls the temperature sensor readings and hard disk write rates in the carriage NVR, performs minimum-maximum normalization conversion on the temperature sensor readings and hard disk write rates, calculates and obtains the node load index score value, and indexes and maps the score value with the channel number to obtain the corresponding load score sequence; The status node screening submodule calls the corresponding value difference of the two sets of data under the same channel number dimension according to the channel activation frequency distribution information and the corresponding sequence of the load score, sorts the difference value sequence by absolute value and screens the channel numbers of the sorting target number, combines the channel index identifier and the node position label to construct a screening result set, and generates an active status node index set.

[0009] The present invention is improved in that the communication load evaluation module includes: The message traffic collection submodule calls the total number of bytes of status message data of the corresponding node in the previous multicast cycle based on the active state node index set, records the corresponding node identifier and time index, counts the cumulative transmission data size of all nodes, and generates node message byte statistics; The channel load judgment submodule calls the node message byte statistical information, and calculates the channel load risk value corresponding to the node with the normalized value of the maximum transmission unit, the normalized value of the broadcast channel rate benchmark, and the normalized upper limit of the uplink occupancy ratio in the carriage multicast channel configuration parameters, and compares the load risk value with the communication load threshold value. The node index outside the threshold range is identified and the communication load judgment result is generated; The early warning status screening submodule calls the node index based on the communication load judgment result, performs cluster identification encoding on the index sequence and channel mapping according to the index set marked as exceeding the threshold node, writes the encoded data into the monitoring output stack, and generates a status reporting conflict risk mark.

[0010] The present invention is improved in that the cache scheduling release module includes: The residence time screening submodule obtains the residence time of the state message corresponding to the active state node in the cache queue and the message residence time threshold of the current cycle according to the state reporting conflict risk mark, judges the size of the residence time of each message and the threshold, filters the message data group whose residence time exceeds the threshold, and obtains the timeout message screening result; The health value sorting submodule, based on the timeout message screening result, calls the hard disk health status value and cache priority flag of each screened message corresponding to the NVR device, sorts the health status values ​​from large to small, and adds the priority flag as an auxiliary tag to the sorted message data to obtain the health priority sorting result; The release list generation submodule extracts the sorted status message identifiers and corresponding sending orders according to the health priority sorting result, performs index rearrangement and multicast encapsulation order writing operations on the identifiers, and generates a multicast status release scheduling list.

[0011] The present invention is improved in that the synchronization boundary correction module includes: The time difference calculation submodule calls the multicast state release scheduling list, collects the reception timestamp values ​​of the remaining nodes in the same multicast frame according to the reception timestamp corresponding to each message, and performs time difference calculation based on the reception timestamp of the first node as the benchmark to generate a sequence of node time offset absolute values, thereby obtaining a set of time offset differences between nodes; The offset target identification submodule obtains the set clock synchronization tolerance threshold based on the time offset difference set between the nodes, compares each offset value with the threshold, and constructs a node evaluation index sequence based on the receiving time stability and historical calibration times of each node. It calculates and obtains the node drift evaluation coefficient, compares the evaluation coefficient with the synchronization control reference value, writes the time offset identification set to the node that exceeds the limit, and generates the node drift evaluation result; The unified timing correction submodule obtains the node number list and the network time protocol main time source timestamp corresponding to the node drift evaluation results that exceed the evaluation interval, corrects the difference between the current timestamp and the time source timestamp of each exceeding node, replaces the time index record, and generates the cross-node synchronization window alignment parameters.

[0012] The present invention is improved in that the system further comprises: The collaborative control configuration module generates a control instruction frame structure encapsulated by a time window based on the cross-node synchronization window alignment parameters and the channel ID involved in the active state node index set, performs multicast broadcast in the set frame structure encapsulation order, and generates a synchronous trigger recording control command group; The synchronous trigger video recording control command group specifically refers to a control frame structure identifier, a channel control instruction index, a trigger execution time point, and a node group target address.

[0013] The present invention is improved in that the collaborative control configuration module includes: The synchronization time point generation submodule extracts the timestamp correction offset of the NVR in each carriage according to the cross-node synchronization window alignment parameter, averages all the offset values ​​and aligns them with the network time protocol time source, selects the minimum available execution time point that meets the synchronization tolerance benchmark on the corrected time axis, and generates unified instruction execution time point value information; The channel instruction encapsulation submodule calls the unified instruction execution time value information, combines the channel ID set recorded in the active state node index set, groups the channel IDs according to the car to which the node belongs, matches the video channel control instruction template to each group of channel IDs and binds them to the unified execution time, builds a complete control instruction structure within each group, and generates time window encapsulation structure parameters; The multicast command broadcast submodule extracts the control content, target channel mapping sequence and execution time field of each instruction structure according to the time window encapsulation structure parameters, writes the frame header, address index and payload information in combination with the multicast control protocol frame format specification, writes all instruction groups into the broadcast queue in the encapsulation order, and generates a synchronous trigger recording control command group.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, by normalizing and quantifying the activation frequency of the video channel in the car, the temperature sensor reading and the hard disk write rate, it is possible to identify the node group in the frequently activated state, realize the dynamic identification of high-load sources and the construction of index sets, and perform load evaluation based on the message data volume and channel transmission capacity of the node in the historical multicast cycle. A conflict warning mark can be formed before the state conflict occurs. By giving priority to processing data with too long a residence time and low hard disk health value, the optimized scheduling of the state message release is realized. Based on the comparison of the node receiving timestamp difference and the clock synchronization tolerance, the time offset between multiple nodes is corrected, and a time-aligned control instruction set is constructed based on the unified time calibration result. The trigger control task is broadcast in a time window encapsulation manner, which improves the efficiency of communication resource utilization, realizes the unified scheduling of multi-node video tasks, execution rhythm synchronization and control response consistency, and improves the stability and timeliness of cross-node state interaction and control execution. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a system flow chart of the present invention; Figure 2 This is a flow chart of the state awareness aggregation module of the present invention; Figure 3 is a flow chart of the communication load evaluation module of the present invention; Figure 4 This is a flow chart of the cache scheduling and release module of the present invention; Figure 5 This is a flow chart of the synchronization boundary correction module of the present invention; Figure 6 This is a flow chart of the collaborative control configuration module of the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0017] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings and are only for the convenience of describing the present invention and simplifying the description. They do not indicate or imply that the devices or elements referred to must have a specific direction, be constructed and operate in a specific direction, and therefore should not be understood as limiting the present invention. In addition, in the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0018] See also Figure 1 , the present invention provides a technical solution: a multi-carriage NVR status reporting and collaborative control system based on multicast communication, the system includes a state perception aggregation module, a communication load evaluation module, a cache scheduling release module, a synchronization boundary correction module and a collaborative control configuration module; The state perception aggregation module obtains the video channel activation flag, temperature sensor reading, and hard disk write rate in the carriage NVR, accumulates and counts the activation flag, obtains the channel activation frequency of each carriage, normalizes the temperature sensor reading and hard disk write rate, calculates the load index score, and combines the channel activation frequency to identify the high-frequency state source group and generate the active state node index set; The communication load assessment module uses the active node index set to call the total number of message data bytes of the corresponding node in the previous multicast cycle and the maximum transmission unit of the multicast channel, calculates the ratio of the two, and compares it with the set communication load threshold to determine whether it has entered the warning range and generate a status reporting conflict risk flag; The cache scheduling and release module uses the status report conflict risk flag, the cache message residence time, cache priority flag and hard disk health status value of the active state node in the cache queue to filter out message data groups whose residence time exceeds the set window, and uses the hard disk health status value as the descending order to output the release group sequence list and generate the multicast status release scheduling list; The synchronization boundary correction module calls the multicast state release scheduling list and obtains the reception timestamp values ​​of the remaining nodes in the same multicast frame based on the reception timestamp corresponding to the message. It compares the timestamp difference of each node in the same group with the set clock synchronization tolerance threshold. If the drift value is greater than the threshold, it is recorded as the offset target and corrected based on the timestamp of the main time source of the Network Time Protocol to generate the cross-node synchronization window alignment parameters. The collaborative control configuration module determines the unified execution time point of the recording channel control instruction set after the NVR device status of each carriage is synchronized based on the cross-node synchronization window alignment parameters. It also combines the channel IDs involved in the active state node index set to generate a control instruction frame structure encapsulated by the time window, and multicasts it in the set frame structure encapsulation order to generate a synchronous trigger recording control command group. The active state node index set includes the channel index sequence number, activation frequency label, load score label and node activity identifier. The status reporting conflict risk mark is specifically the communication warning label, node load level, cycle occupancy ratio and conflict judgment result. The multicast state release scheduling list includes the message priority index, release sorting sequence, cache residency flag and health status label. The cross-node synchronization window alignment parameters include node time offset, synchronization adjustment amount, calibration reference value and tolerance judgment label. The synchronous trigger recording control command group specifically refers to the control frame structure identifier, channel control instruction index, trigger execution time point and node group target address.

[0019] See also Figure 2 , the state awareness aggregation module includes: The channel activation identification submodule obtains the video channel activation flag in the carriage NVR, classifies and counts each activation flag according to the channel sequence, records the number of times the same channel number is activated in a unit period, and constructs an activation frequency sequence to obtain channel activation frequency distribution information; When obtaining the recording channel activation flag in each carriage NVR, the system reads the result based on the real-time working status of each node NVR, extracts the Boolean flag of whether it is in the enabled state, and aggregates the channel numbers of all status frames marked as enabled within the unit control cycle (such as every 300 seconds). The aggregated numbers are sorted in ascending order and then assigned to their corresponding number indexes. For example, in the 4 carriages of a train, the channel activation flags of nodes N1~N5 are [1,1,0,1,1], and the corresponding enabled channel numbers are [101,102,103,104,105]. The channel corresponding to number 103 is not enabled, so the number corresponding to this number is 0 when it is included in the activation frequency sequence. The cumulative number of each number in one cycle is summarized as: {101:3,102:3,103:0,104:3,105:3}. For different channel numbers, the system records their The activity level in each sampling is set and the activation frequency statistical benchmark interval is set to 10 times the total number of sampling times in the unit cycle. The statistical results are converted into percentages with the benchmark. For example, the activation frequency of channel 101 is 3, and the activation frequency percentage is 30%. The activation frequency benchmark value of 10 times comes from the system-set sampling interval of once every 30 seconds. The total number of sampling times in a unit cycle of 300 seconds is 10 times. This value does not change with the number of nodes and only depends on the two system-set parameters of period and sampling frequency. Therefore, the benchmark value is fixed. If the frequency is less than 20%, it is marked as a non-master active channel. If the frequency exceeds 60%, it is marked as a core recording channel. The interval judgment standard is derived from the previous engineering debugging data, and the distinction boundary is obtained by comparing the abnormal state and the recording failure rate. Finally, the channel activation frequency distribution information generated according to the frequency ratio is used for subsequent combined judgment with the load coupling degree score. The load index conversion submodule calls the temperature sensor reading and hard disk write rate in the carriage NVR, and performs minimum-maximum normalization conversion on the temperature sensor reading and hard disk write rate respectively, using the formula: ; Obtain the node load index score value through operation, perform index mapping between the score value and the channel number, and obtain the corresponding sequence of load scores; in, represents the load index score value, Represents the normalized value of the temperature sensor reading, Indicates the normalized value of the hard disk write rate. Indicates the normalized value of the minimum reading of the temperature sensor, Indicates the normalized value of the maximum reading of the temperature sensor, Indicates the normalized value of the minimum hard disk write rate. Indicates the normalized value of the maximum hard disk write rate. is the normalized value of the total hard disk operation time; After calling the raw data of the temperature sensor reading and the hard disk write rate, the system performs a minimum-maximum normalization conversion on each parameter. The temperature reading range is extracted from Table 1 below. The minimum value is 55.7 and the maximum value is 62.1. For example, the normalized value corresponding to the temperature of node N1 of 58.4 is: , the hard disk rate normalization method is the same. The rate of node N1 is 110MB / s, corresponding to the minimum value 108 and the maximum value 120. After normalization, it is: , and read the hard disk running time at the same time, normalized to node N1: , bring all parameters into the improved formula: ; The numerator part of the formula Indicates the combined activity of the node's temperature and rate in the current cycle. A larger value indicates that the node is in a higher operating load state in the current cycle. The denominator introduces the sum of the temperature and rate ranges and the logarithm of the normalized operating time. This term is used to adjust the intensity of the numerator output. As the hard disk's continuous operating time increases, the corresponding natural logarithm term increases, resulting in a decrease in the score value, thereby penalizing devices with continuously high loads. This helps to highlight newly surged load nodes and enhance state recognition sensitivity. The logarithmic function in this mechanism uses the natural logarithm form to ensure that the term converges to 0 when D approaches 0, thereby avoiding artificially lowering the score value in short-term loads. The result of the formula calculation is the node load index score L. The unit interval range is between 0 and 1 and is bound to the channel number.

[0020] Table 1 Channel node operating parameters:

[0021] As shown in Table 1, the data comes from the sensors and operation monitoring logs collected during the N1-N5 node period. This table provides the data basis for temperature, rate, and duration for the formula.

[0022] Substituting the values: , , , , then the score of node N1 is: ; The normalized baseline value for hard drive runtime is the difference between the minimum and maximum runtimes collected during the current cycle, which is 450 hours. The fixed value range is affected by the current monitoring system's 24-hour sampling policy. Each node updates its runtime hours once a day, so the normalized range doesn't fluctuate dramatically. The system binds the L value to the channel number to form a load score sequence for subsequent difference determination. The status node screening submodule uses the channel activation frequency distribution information and the corresponding sequence of load scores to call the corresponding value difference of the two sets of data under the same channel number dimension, sorts the difference value sequence by absolute value, and screens the channel numbers of the target number of sorting. It combines the channel index identifier and the node position label to construct a screening result set and generate an active status node index set; According to the channel activation frequency distribution information and the load score corresponding sequence, the system aligns the two sequences with the channel number as the index. Each channel number corresponds to two values: the activation frequency percentage F and the load score value L. The absolute value of the difference between the two is calculated and executed. For example, if number 101 corresponds to F=0.3 and L=0.0295, the difference is 0.2705. The system sorts all difference sequences from large to small and extracts the top 3 channel numbers. If the system sets the retention threshold as channels with a difference ≥ 0.25, channels 101, 104, and 105 are filtered. The screening threshold of 0.25 is set based on the upper and lower limits of the activation frequency standard and the load score theoretical value fluctuation. When the difference between the two reaches 0.25 or more, it means that the node load is separated from its activation status, which is an unbalanced node. If this threshold is reached multiple times within the sampling period, it will be included in the central control alarm information pool. The channel numbers that meet the conditions are combined with their physical layout information and logical hierarchical structure to form a structured form record. The output structure constitutes the active state node index set, which is used to identify the target channel in the synchronization configuration and control task issuance stage.

[0023] See also Figure 3 , the communication load evaluation module includes: The message traffic collection submodule, based on the active node index set, calls the total number of bytes of status message data of the corresponding node in the previous multicast cycle, records the corresponding node identifier and time index, counts the cumulative transmission data size of all nodes, and generates node message byte statistics; Obtain the total number of bytes of status message data for each node in the active node index set in the previous multicast cycle. The node set is derived from the synchronous identification results of the upper-layer module. Assume that there are nodes N1 to N5 in this cycle, numbered 01 to 05 respectively. The system detection module sends a timing trigger frame at the cycle boundary time point. Each node records the cumulative bytes and continuous sending time of the status message from the start to the end of the timing. The cumulative reading of node N1 is 245kB, and the sending duration in this cycle is 58 seconds. Node N2 reads 198kB, which lasts for 54 seconds. The remaining nodes are collected and recorded in turn. After the sampling is completed, the system sums the unified byte amount of all nodes and constructs a data structure according to the node sequence number. The total amount of transmitted data is counted and the identification time index and node ID index are normalized to build a unified cross-node data index structure for participating in bandwidth occupancy calculation in the subsequent communication load judgment module.

[0024] Table 2 Statistics of node message transmission characteristics

[0025] As shown in Table 2, the total byte value and corresponding time index data of each node message have been completely collected and mapped by the system cache metering module. After statistics, node message byte statistics are generated for subsequent modules to call.

[0026] The channel carrying judgment submodule calls the node message byte statistics information, and combines it with the normalized maximum transmission unit value, the normalized broadcast channel rate benchmark value, and the normalized upper limit of the uplink occupancy ratio in the carriage multicast channel configuration parameters, using the formula: ; Obtain the channel load risk value corresponding to the node through calculation, perform a single value comparison between the load risk value and the communication load threshold, identify the node index outside the threshold range, and generate the communication load judgment result; in, represents the channel load risk value, Indicates the normalized value of the node message byte statistics. Indicates the normalized value of the maximum transmission unit of the multicast channel, Indicates the normalized value of the accumulated duration of the node status message. Indicates the normalized value of the channel broadcast rate reference value, Indicates the normalized value of the node uplink occupancy ratio, Indicates the normalized value of the uplink occupancy ratio threshold; Based on the node message byte statistics, the system calls the multicast channel configuration parameters corresponding to the car to which each node belongs, extracts the maximum transmission unit setting value (in Byte), the broadcast channel rate baseline value (in MB / s), and the system monitoring value of the current uplink occupancy ratio (without unit) from the configuration table, and performs minimum and maximum normalization processing on these parameters and the node message byte value and transmission duration collected in the previous section, and converts them into dimensionless variables: byte normalization value , Maximum transmission unit normalized value , duration normalized value , rate normalization value , normalized value of uplink occupancy ratio , Uplink occupancy ratio threshold normalized value , the normalization process is as follows: For example, if the maximum transmission unit range is 1000B to 1600B, then the node N1 corresponds to 1200B, then the normalization ; The write rate ranges from 50MB / s to 100MB / s. If N1 corresponds to 80MB / s, it is normalized to The system preset threshold for uplink occupancy is 60%. The current N1 measurement is 64%. After normalization, , .

[0027] Afterwards, the formula is used: ; Substituting all the above normalized values, we can calculate N1 as follows: , , ; The calculation logic of this formula is as follows: the first term is the superposition of the single-node byte transmission pressure and the time dimension bandwidth pressure, and the square root of the second term quantifies the deviation between the current actual uplink occupancy status and the set benchmark, which expresses the critical pressure of the channel resources as a whole. The coefficient term does not need to be adjusted externally, and the result magnitude is directly controlled by the normalized data alone.

[0028] The parameter settings are as follows: The normalized range of the maximum transmission unit is set to 1000B to 1600B, which is fixed in the corresponding carriage hardware module and comes from the protocol MTU configuration; the broadcast rate benchmark is 50MB / s to 100MB / s, and the equipment configuration is fixed; the uplink occupancy ratio threshold is set to U=0.60, and the average uplink transmission reserved channel ratio is set according to the system multicast design, and the fluctuation tolerance is allowed to vary in the range of 5% to 10%.

[0029] After operation, all nodes The value is calculated and compared with the communication load threshold (for example, the system is uniformly preset to 0.45). If a node , that is, it is judged as a node with channel load exceeding the limit, its corresponding node number is identified, and the communication load judgment result is established.

[0030] The early warning status screening submodule calls the node index based on the communication load determination result, performs cluster identification encoding on the index sequence and channel mapping according to the index set marked as exceeding the threshold node, writes the encoded data into the monitoring output stack, and generates a status reporting conflict risk mark; Based on the communication load judgment result, the index set of risk nodes is extracted. For example, nodes N3 and N4 are marked as over-threshold nodes. Their node numbers are sorted in ascending order and mapped to the car numbers, and the physical multicast channel numbers assigned to them are read. For example, N3 is channel A and N4 is channel B. The system constructs a two-dimensional mapping pair for the node index and channel number, and uses the system-defined clustering interval value group to cluster and merge according to the node index step of 1. For example, if N3 and N4 are adjacent, they are clustered into the same group G1, and the cluster identifier is G1. The mapping results are encoded in the cluster order and written into the status conflict monitoring output stack. The compressed data index amount is mapped through clustering focus. After writing into the stack structure, a status reporting conflict risk mark is established for the subsequent cache scheduling release module to dynamically adjust the release control strategy for the risk node.

[0031] See also Figure 4 , the cache scheduling and release module includes: The residence time screening submodule obtains the residence time of the state message corresponding to the active state node in the cache queue and the message residence time threshold of the current cycle according to the state reporting conflict risk mark, judges the size of the residence time of each message against the threshold, and filters the message data group whose residence time exceeds the threshold to obtain the timeout message screening result; According to the node index data in the status report conflict risk mark, the system obtains the status message corresponding to the active node marked as conflict state from the cache queue one by one, and obtains the residence time of the message by comparing its cache entry time with the current cycle reference time, and compares it with the message residence time threshold set in the current scheduling cycle. In this embodiment, the message residence time is calculated by the formula It is concluded that is the dwell time, is the current scheduling cycle time, The time for the message cache to be queued. Set the cycle threshold The value is 12 seconds, which is derived from the system scheduling parameters. Considering that the master control scheduling cycle is 200ms and the maximum number of frames that can be accommodated in the cache window is 600 frames, the estimated retention threshold is Due to the compression of the current cycle limit, this value is standardized and reduced to 12s. The message cache entry time and the current system cycle time of nodes N1 to N5 are obtained, and the residence time of each message is calculated accordingly. The specific values ​​are shown in Table 3: Table 3 Comparison of message residence time and threshold

[0032] The system For the judgment condition, the nodes corresponding to the timeout message are identified as N2 and N5, forming the timeout message screening result. This result will serve as the basis for message sorting and scheduling in the next processing link.

[0033] The health value sorting submodule is based on the timeout message filtering results. It calls the hard disk health status value and cache priority flag of each filtered message corresponding to the NVR device, sorts the health status values ​​from large to small, and adds the priority flag as an auxiliary tag to the sorted message data to obtain the health priority sorting result; According to the message list identified in the timeout message screening results, the system calls the status index of the NVR device to which each message belongs, and obtains its current hard disk health status value and the cache priority flag recorded in its cache management module from the device health monitoring module. The priority is represented by a 2-bit binary value, where "00" is the lowest and "11" is the highest. Taking nodes N2 and N5 as examples, their health values ​​are 87 and 65 respectively, and the priority flags are "10" and "11" respectively. The system sorts in descending order based on the health value. When the health values ​​are equal in the sorting, the priority is determined by reference. The system adopts a double sorting logic: first, the health status value is used to sort the nodes, and then the priority is used to sort the nodes. Sort by priority bit code A stable sort is performed, and the sorted node order is bound to the message index structure to construct a sort table. This sorting process does not require external sorting methods; it is accomplished simply by rearranging the message record list structure based on a combination of health values ​​and priorities. After sorting is complete, the system writes all sorted fields into the structure table, generating a health-prioritized sort result for subsequent scheduling priority decisions.

[0034] The release list generation submodule extracts the sorted status message identifiers and the corresponding sending order according to the health priority sorting results, re-indexes the identifiers and writes the multicast encapsulation order to generate a multicast status release scheduling list; According to the health priority sorting results, the system extracts the corresponding status message unique identifier, sorting index number and target sending node ID in each record in turn, and constructs the message multicast scheduling structure according to the sorting index. The system adopts a structural encapsulation rule: with the message index as the main field, each message is encapsulated into the multicast frame structure according to the sequence number. During the process, the sorting index, channel ID mapping value and sending timestamp are synchronously recorded and written into the frame header information structure table. The index rearrangement in this process is implemented through the field position index writing strategy and does not rely on other external algorithms. For the status messages of N2 and N5 involved in this scheduling, their sorting indexes are 1 and 2 respectively, and the system encapsulates them into the broadcast frame structure unit in this order. After completing the frame encapsulation, the scheduler writes the encapsulation order and message ID mapping relationship into the multicast scheduling control list, and generates a multicast status release scheduling list for the lower-level broadcast module to execute according to the timing.

[0035] See also Figure 5 , the synchronous boundary correction module includes: The time difference value calculation submodule calls the multicast state release scheduling list, collects the receiving time stamp values of the remaining nodes in the same multicast frame according to the receiving time stamp corresponding to each message, takes the receiving time stamp of the first node as a reference, performs time difference value calculation and generates a node time offset absolute value sequence, and obtains a time offset difference value set between nodes; The receiving time stamp corresponding to each message in the multicast state release scheduling list is called, the receiving time stamp values of the remaining nodes in the same multicast frame are collected, the first node is extracted as a time reference node, the receiving time stamp of the first node is taken as a reference, the receiving time stamps of the other nodes in the same frame are sequentially subjected to difference value calculation, the difference value is in the form of an absolute value after positive and negative processing, to avoid directional offset misjudgment due to receiving sequence, and a synchronization time offset difference value set between multiple nodes is constructed, the offset of each node is marked with a node number, for example, the receiving time of node N1 is 124.058 seconds, the receiving time of node N2 is 124.091 seconds, the offset is 0.033 seconds, the receiving time of node N3 is 124.045 seconds, and the offset is 0.013 seconds, each offset value is recorded in the offset difference value matrix, and then written into the time offset difference value set for subsequent module reference, and a time offset difference value set between nodes is generated; The offset target identification submodule obtains a set clock synchronization tolerance threshold value from the time offset difference value set between nodes, compares each offset value with the threshold value, and constructs a node evaluation index sequence in combination with the receiving time stability of each node and the historical calibration times, using the formula: ; The node drift evaluation coefficient is obtained by operation, the evaluation coefficient is compared with the synchronization control reference value, the time offset identification set is written for the out-of-limit node, and a node drift evaluation result is generated; wherein, the node drift evaluation coefficient is represented by, the current node receiving time stamp value is represented by, the reference node time stamp value is represented by, the set clock synchronization tolerance threshold value is represented by, the cumulative number of state message arrivals of the node in the current period is represented by, the cumulative number of calibrations of the node in the historical period is represented by; The set clock synchronization tolerance threshold value is obtained from the time offset difference value set between nodes, which is set to 0.030 seconds. The specific basis is that the maximum offset between multicast frames is allowed to be not more than 30 milliseconds in the urban rail transit scene, and if it exceeds, there is a risk of tearing of the multi-channel synchronization of NVR recording, therefore, the value is taken as a unified upper and lower limit, in combination with the time difference value calculation result, the current receiving time stamp of each node, the reference time stamp, The cumulative number of times the node has been calibrated in the historical period , the drift risk is assessed using the following formula: ; In this formula structure, the numerator is the time series amplification term of the offset difference, which represents the weight adjustment amplitude of the current node drift under the influence of the arrival density. Used to expose nodes where messages arrive in a concentrated manner as high drift risk points; the denominator is based on the synchronization tolerance threshold The square root function of the number of historical node calibrations is multiplied as a proportional benchmark to amplify the sensitivity of nodes with long-term drift trends. This logic design can enhance the identification of frequently offset nodes and is suitable for predicting node drift with asynchronous mutations in urban rail operation. Finally, the node offset degree index can be quantified through the F value.

[0036] Taking node N4 as an example, the current timestamp is 124.089 seconds, the base timestamp is 124.058 seconds, A is 9 times, and Z is 3 times. Substitute the formula into the calculation: ; Table 4 Node drift evaluation input and results

[0037] As shown in Table 4, the drift assessment coefficient for node N1 exceeds the baseline value of 1.5, indicating a clear risk status. Therefore, during the drift identification operation, this node is assigned an over-limit drift flag set as a target for subsequent correction. This result indicates that an assessment coefficient greater than 1.5 indicates that the node's drift level and frequency in the current cycle have reached a level requiring calibration. This indicator also serves as a logical benchmark for subsequent synchronization window reconstruction.

[0038] The unified timing correction submodule obtains the node number list and the network time protocol master time source timestamp corresponding to the node drift assessment results, and corrects the difference between the current timestamp and the time source timestamp of each node that exceeds the assessment interval, replaces the time index record, and generates the cross-node synchronization window alignment parameters. According to the node drift assessment results, obtain the set of node numbers that are recorded as exceeding the assessment interval. Assuming that the current node is node N1, after its mark drifts, perform difference correction processing on its current receiving timestamp 124.089 seconds and the network main time source value 124.063 seconds. Use the difference value reverse correction strategy to replace the original value with 124.063 seconds. At the same time, refresh its time index record table, locate the data before and after correction with the sequence number in the cross-frame data structure, synchronize the timestamp window used by its downstream control scheduling module after updating, write it into the "synchronization window timing set" structure, and then generate the cross-node synchronization window alignment parameters.

[0039] See also Figure 6 , the collaborative control configuration module includes: The synchronization time point generation submodule extracts the timestamp correction offset of each NVR in each carriage according to the cross-node synchronization window alignment parameter, calculates the average of all offset values, aligns with the network time protocol time source, and selects the smallest available execution time point that meets the synchronization tolerance reference on the corrected time axis to generate unified instruction execution time point value information; According to the cross-node synchronization window alignment parameter, the timestamp correction offset data corresponding to all NVR nodes in each carriage is extracted, the statistical sequence range is set to a maximum of 16 nodes per carriage based on the number of nodes, and after obtaining the offset, a node number weighted average form is used for aggregation calculation, wherein the weight is the state message reception success rate of each node. Due to the periodic jitter of the state message and the frequent clock drift of part of the nodes, the system sets the state message reception success rate weight setting value as: , wherein represents the node the number of successful receptions in the last five periods, is the stable reception threshold, is the reception stability growth coefficient, which is set to ensure that only stable nodes have a greater impact on the average offset value, and to avoid errors introduced by drift nodes. The numerical range varies from 0 to 1, and increases monotonously with ; Let the node offset be , then the offset correction reference value is: , the correction reference value is aligned with the network time protocol master time source timestamp to obtain the reference correction time, and the offset correction time axis is searched backward at a resolution of 1ms to find the smallest executable time point that meets the condition that all node offset difference values do not exceed the synchronization tolerance reference value . The reference value is fixed at 5ms. Since the minimum network delay between carriages is 2ms and the maximum offset standard deviation is about 1.4ms, 5ms is set as the boundary value to ensure that all nodes meet the alignment condition at this time point. After multiple sampling verification, the fluctuation is controlled within ±0.2ms, and the stability is above 97%.

[0040] The channel instruction packaging submodule calls the unified instruction execution time point value information, combines the channel ID set recorded in the active state node index set, groups the channel IDs by node belonging carriage, matches the video channel control instruction template for each group of channel IDs and binds the unified execution time point, builds a complete control instruction structure within each group, and generates a time window packaging structure parameter. The channel ID set data in the unified instruction execution time point value information and the active state node index set is called, and grouped according to the car classification to which the channel belongs. For example, there are channel IDs 3, 6, and 9 in car A, and channel IDs 1, 2, and 4 in car B. The grouped channel sets are extracted respectively, and the recording control instruction template is read for each channel ID in each group. The template records three fields: control type, instruction code, and execution mode. While binding the unified execution time point, the identification field is appended to build a structure. The structured template example is shown in the following table: Table 5 Control instruction structure encapsulation example

[0041] As shown in Table 5, each structure contains a complete set of control parameter fields. All channel control data are aggregated to form a time window encapsulation structure parameter, which is used to align the operation execution order and execution logic consistency within the synchronous execution window.

[0042] The multicast command broadcast submodule extracts the control content, target channel mapping sequence, and execution time field of each command structure based on the time window encapsulation structure parameters. It then writes the frame header, address index, and payload information in accordance with the multicast control protocol frame format specification. It then writes all command groups into the broadcast queue in the encapsulation order to generate a synchronous trigger recording control command group. According to the time window encapsulation structure parameters, the control content, target channel mapping sequence and execution time field are extracted in sequence. Combined with the multicast protocol frame structure, the control frame header identification field (16 bits), destination channel ID index field (8 bits) and execution control field (32 bits) are written. The order of all command frames in the broadcast queue is organized according to the preset broadcast priority table sorting rules. During the encapsulation write operation, the system generates a broadcast frame sequence based on the channel-to-car order to ensure that the physical transmission order is not disordered, and records the frame transmission sequence number at the same time. After all control frames in the multicast queue are written, a synchronous trigger recording control command group is output. This command group is parsed and executed by all active nodes at the command trigger point, forming a complete synchronous trigger recording control command flow.

[0043] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. Multi-carriage NVR status reporting and collaborative control system based on multicast communication, characterized by: The system comprises: The state perception aggregation module obtains the video channel activation flag, temperature sensor reading, and hard disk write rate in the carriage NVR, calculates the load index score, combines the channel activation frequency, identifies the high-frequency state source group, and generates an active state node index set; The communication load evaluation module calls the total number of message data bytes of the corresponding node in the previous multicast cycle and the maximum transmission unit of the multicast channel based on the active state node index set, determines whether it has entered the warning interval, and generates a status reporting conflict risk mark; The cache scheduling and release module calls the status message residence time, cache priority flag and hard disk health status value of the active state node in the cache queue according to the status reporting conflict risk mark, filters the message data group whose residence time exceeds the set window, sorts them in descending order according to the hard disk health status value, and generates a multicast status release scheduling list; The synchronization boundary correction module calls the multicast state release scheduling list, compares the timestamp difference of each node in the same group with the set clock synchronization tolerance threshold, corrects it based on the timestamp of the network time protocol master time source, and generates cross-node synchronization window alignment parameters.

2. The multi-carriage NVR status reporting and collaborative control system based on multicast communication according to claim 1 is characterized in that: The active state node index set includes a channel index sequence number, an activation frequency label, a load score label and a node active identifier; the status reporting conflict risk mark is specifically a communication warning label, a node load level, a cycle occupancy ratio and a conflict judgment result; the multicast state release scheduling list includes a message priority index, a release sorting sequence, a cache residency flag and a health status label; the cross-node synchronization window alignment parameters include a node time offset, a synchronization adjustment amount, a calibration reference value and a tolerance judgment label.

3. The multi-carriage NVR status reporting and collaborative control system based on multicast communication according to claim 2 is characterized in that: The state awareness aggregation module includes: The channel activation identification submodule obtains the video channel activation flag in the carriage NVR, classifies and counts each activation flag according to the channel sequence, records the number of times the same channel number is activated in a unit period, and constructs an activation frequency sequence to obtain channel activation frequency distribution information; The load index conversion submodule calls the temperature sensor readings and hard disk write rates in the carriage NVR, performs minimum-maximum normalization conversion on the temperature sensor readings and hard disk write rates, calculates and obtains the node load index score value, and indexes and maps the score value with the channel number to obtain the corresponding load score sequence; The status node screening submodule calls the corresponding value difference of the two sets of data under the same channel number dimension according to the channel activation frequency distribution information and the corresponding sequence of the load score, sorts the difference value sequence by absolute value and screens the channel numbers of the sorting target number, combines the channel index identifier and the node position label to construct a screening result set, and generates an active status node index set.

4. The multi-carriage NVR status reporting and collaborative control system based on multicast communication according to claim 3 is characterized in that: The communication load evaluation module includes: The message traffic collection submodule calls the total number of bytes of status message data of the corresponding node in the previous multicast cycle based on the active state node index set, records the corresponding node identifier and time index, counts the cumulative transmission data size of all nodes, and generates node message byte statistics; The channel load judgment submodule calls the node message byte statistical information, and calculates the channel load risk value corresponding to the node with the normalized value of the maximum transmission unit, the normalized value of the broadcast channel rate benchmark, and the normalized upper limit of the uplink occupancy ratio in the carriage multicast channel configuration parameters, and compares the load risk value with the communication load threshold value. The node index outside the threshold range is identified and the communication load judgment result is generated; The early warning status screening submodule calls the node index based on the communication load judgment result, performs cluster identification encoding on the index sequence and channel mapping according to the index set marked as exceeding the threshold node, writes the encoded data into the monitoring output stack, and generates a status reporting conflict risk mark.

5. The multi-carriage NVR status reporting and collaborative control system based on multicast communication according to claim 4 is characterized in that: The cache scheduling release module includes: The residence time screening submodule obtains the residence time of the state message corresponding to the active state node in the cache queue and the message residence time threshold of the current cycle according to the state reporting conflict risk mark, judges the size of the residence time of each message and the threshold, filters the message data group whose residence time exceeds the threshold, and obtains the timeout message screening result; The health value sorting submodule, based on the timeout message screening result, calls the hard disk health status value and cache priority flag of each screened message corresponding to the NVR device, sorts the health status values ​​from large to small, and adds the priority flag as an auxiliary tag to the sorted message data to obtain the health priority sorting result; The release list generation submodule extracts the sorted status message identifiers and corresponding sending orders according to the health priority sorting result, performs index rearrangement and multicast encapsulation order writing operations on the identifiers, and generates a multicast status release scheduling list.

6. The multi-carriage NVR status reporting and collaborative control system based on multicast communication according to claim 5 is characterized in that: The synchronization boundary correction module includes: The time difference calculation submodule calls the multicast state release scheduling list, collects the reception timestamp values ​​of the remaining nodes in the same multicast frame according to the reception timestamp corresponding to each message, and performs time difference calculation based on the reception timestamp of the first node as the benchmark to generate a sequence of node time offset absolute values, thereby obtaining a set of time offset differences between nodes; The offset target identification submodule obtains the set clock synchronization tolerance threshold based on the time offset difference set between the nodes, compares each offset value with the threshold, and constructs a node evaluation index sequence based on the receiving time stability and historical calibration times of each node. It calculates and obtains the node drift evaluation coefficient, compares the evaluation coefficient with the synchronization control reference value, writes the time offset identification set to the node that exceeds the limit, and generates the node drift evaluation result; The unified timing correction submodule obtains the node number list and the network time protocol main time source timestamp corresponding to the node drift evaluation results that exceed the evaluation interval, corrects the difference between the current timestamp and the time source timestamp of each exceeding node, replaces the time index record, and generates the cross-node synchronization window alignment parameters.

7. The multi-carriage NVR status reporting and collaborative control system based on multicast communication according to claim 6 is characterized in that: The system further comprises: The collaborative control configuration module generates a control instruction frame structure encapsulated by a time window based on the cross-node synchronization window alignment parameters and the channel ID involved in the active state node index set, performs multicast broadcast in the set frame structure encapsulation order, and generates a synchronous trigger recording control command group; The synchronous trigger video recording control command group specifically refers to a control frame structure identifier, a channel control instruction index, a trigger execution time point, and a node group target address.

8. The multi-carriage NVR status reporting and collaborative control system based on multicast communication according to claim 7 is characterized in that: The collaborative control configuration module includes: The synchronization time point generation submodule extracts the timestamp correction offset of the NVR in each carriage according to the cross-node synchronization window alignment parameters, averages all the offset values ​​and aligns them with the network time protocol time source, selects the minimum available execution time point that meets the synchronization tolerance benchmark on the corrected time axis, and generates unified instruction execution time point value information; The channel instruction encapsulation submodule calls the unified instruction execution time value information, combines the channel ID set recorded in the active state node index set, groups the channel IDs according to the car to which the nodes belong, matches the video channel control instruction template to each group of channel IDs and binds them to the unified execution time, builds a complete control instruction structure within each group, and generates time window encapsulation structure parameters; The multicast command broadcast submodule extracts the control content, target channel mapping sequence and execution time field of each instruction structure according to the time window encapsulation structure parameters, writes the frame header, address index and payload information in combination with the multicast control protocol frame format specification, writes all instruction groups into the broadcast queue in the encapsulation order, and generates a synchronous trigger recording control command group.

Images