Search

CN-122019268-A - Sonar database multi-node hot standby and intelligent switching management method

CN122019268ACN 122019268 ACN122019268 ACN 122019268ACN-122019268-A

Abstract

The invention discloses a sonar database multi-node hot standby and intelligent switching management method which comprises the steps of dynamically distributing received sonar data streams to a main node or each standby node according to priority and data types, continuously exchanging health information by each ship node and performing cross verification, calculating the trusted service state of each node through a consensus mechanism based on the health information and the cross verification result, triggering a switching flow when the trusted service state of the main node fails, constructing a benefit evaluation model according to the data synchronous timeliness, the platform survival state and the task cruising ability of each available standby node, comprehensively grading, selecting the standby node with the highest comprehensive grading as a new main node, and switching all data service requests to the new main node. The invention realizes the high availability, high reliability and intelligent management of the offshore formation sonar data processing system through dynamic data distribution, cross validation, consensus calculation and income evaluation.

Inventors

  • Chen Shuaike
  • XU FENG
  • XU JIALIN
  • RUAN YUXING

Assignees

  • 中国舰船研究设计中心

Dates

Publication Date
20260512
Application Date
20260413

Claims (9)

  1. 1. A sonar database multinode hot standby and intelligent switching management method is applied to a sonar data processing system formed by a plurality of marine ship nodes, and is characterized by comprising the following steps: according to the sonar data processing load periodically reported by each ship node and the communication quality of the offshore link, dynamically distributing the received sonar data stream to a main node or each standby node according to the priority and the data type; Each ship node continuously exchanges health information and performs cross verification while distributing and processing sonar data; Based on the health information and the cross verification result, calculating the trusted service state of each node through a consensus mechanism, and triggering a switching flow when the trusted service state of the master node fails; In the switching process, a benefit evaluation model is constructed and comprehensive scoring is carried out according to the data synchronous timeliness, the platform survival state and the task cruising ability of each available standby node; And selecting the standby node with the highest comprehensive score as a new main node, and switching all data service requests to the new main node.
  2. 2. The sonar database multinode hot standby and intelligent switching management method of claim 1, wherein the sonar data processing load and the offshore link communication quality reported periodically by each ship node are calculated by the following modes: Each ship node obtains each first type of index data of the ship node, and carries out weighted summation after normalization to obtain a node sonar data processing load coefficient, wherein the first type of index data comprises CPU utilization rate, GPU utilization rate, memory occupancy rate, disk I/O queue length, sonar data packet input/output rate and floating point operation times consumed by a sonar signal processing algorithm in unit time; Each ship node acquires each second type of index data of the ship node, and substitutes the second type of index data into a preset link quality function to calculate to obtain a marine link communication quality score, wherein the second type of index data comprises a satellite/radio link Round Trip Time (RTT) and a packet loss rate Received signal strength RSSI, bit error rate BER, and available link bandwidth ; The link quality function is expressed as: Wherein the method comprises the steps of Representing the quality score of the communication of the offshore link, Representing the parameters of the intermediate calculation, Indicating the maximum allowed round trip delay upper bound, Which is indicative of the strength of the reference signal, The bandwidth factor is represented as a function of the bandwidth factor, Represents the historical average bandwidth of the fleet node, 、 、 、 、 Respectively representing the calculated weights for the respective second type index data.
  3. 3. The method for multi-node hot standby and intelligent switching management of sonar database according to claim 2, wherein the dynamic distributing step adopts a weighted polling scheduling algorithm based on priority, and the process comprises: Dividing sonar data streams to be distributed into different priorities according to task urgency, and giving differentiated scheduling weights according to active sonar, passive sonar, underwater acoustic communication and environmental noise data types under the same priority; Calculating real-time service bearer margins for each candidate node Wherein The load factor is processed for the node sonar data, Scoring link communication quality; Guiding the data packet to the available node with the largest C value, and if the C values are the same, distributing the data packet according to a polling mode; guiding the data packet to the available node with the maximum current service bearing allowance C; When the service bearing allowance C values of the plurality of available nodes are the same, the system distributes data packets in a weighted polling mode, wherein the polling weights take differentiated scheduling weights of data types; and for the high-throughput original sonar data flow, parallel distribution is carried out according to the service bearing allowance proportion of each available node.
  4. 4. The sonar database multinode hot standby and intelligent switching management method of claim 1, wherein each ship node continuously exchanges health information process comprises: each ship node periodically collects platform running state data and sonar task key indexes, and generates health information statement, wherein the health information statement comprises: platform running state data, namely platform attitude information, self-checking state codes of key equipment, energy reserve percentage and environmental cabin data; the key indexes of the sonar task include the working state of each sonar sensor, the success rate of the current data processing pipeline, and the counting and packet loss rate of sonar data packets processed in the last period; after the statement of health information is generated, the private key of the node is used for digital signature, and periodic broadcasting is carried out through the inter-ship data link.
  5. 5. The sonar database multi-node hot standby and intelligent switch management method of claim 4, wherein performing cross-validation when a node receives a statement of health information from another node comprises: Sending a lightweight probe activity request data packet to a health information statement initiating node, and recording the response time and the correctness of response content; Extracting one or more data samples from sonar data streams recently distributed by the statement initiating node according to a preset strategy, and checking format compliance, data validity and time stamp rationality of the data samples; and integrating the response result of the probe activity request with the data sampling verification result to generate a cross verification report aiming at the target node, and sending the cross verification report to a consensus management module in the system.
  6. 6. The sonar database multi-node hot standby and intelligent switching management method of claim 5, wherein the consensus management module performs operations comprising: collecting statement of health information from a plurality of nodes to the same target node and corresponding cross verification reports; the Bayesian fault-tolerant consensus algorithm is adopted to carry out consistency verification on the collected information, if the cross verification report exceeding a preset threshold value confirms that the statement of the target node is valid and the detection result is normal, the health information of the node is judged to be credible; Integrating the direct health information statement and the indirect consensus verification result, and calculating and outputting a trusted service state value of the node; And determining a trusted service state value, judging the trusted service state of the master node, and triggering a switching flow when the trusted service state of the master node fails.
  7. 7. The method for multi-node hot standby and intelligent switching management of sonar database according to claim 1, wherein constructing a benefit evaluation model and comprehensively scoring according to the data synchronization timeliness, the platform survival state and the task cruising ability of each available standby node in the switching flow comprises: Calculating data synchronization lag time between each standby node and fault main node ; Converting lag time to data synchronization timeliness scoring Wherein Presetting an allowable maximum synchronization hysteresis threshold; According to the real-time platform state data of each standby node, carrying out normalization and weighted summation on each state data, and calculating to obtain a platform survival state score in the range of 0 to 1, wherein the state data comprises a damage management system alarm level, an energy system residual percentage and a navigation/communication core equipment availability state; The method comprises the steps of obtaining residual task resource data of a standby node, wherein the residual task resource data comprises residual fuel percentage, comprehensive replenishment allowance, key sonar consumable stock and residual planning duration of a current task of the node obtained from a fleet task management system; Based on the residual task resource data and the corresponding consumption rate model, estimating the sustainable operation duration of the node under the condition of maintaining the current sonar data processing capacity: Will be Minimum necessary guarantee duration preset with system Comparing, calculating to obtain the endurance score through a preset scoring function : And normalizing the data synchronization timeliness score, the platform survival state score and the endurance score, substituting the normalized data into a preset linear weighted gain evaluation model for comprehensive scoring, and obtaining a scoring result.
  8. 8. The sonar database multinode hot standby and intelligent switching management method of claim 1, further comprising the step of, after the switching is completed: counting the caching, forwarding and processing behaviors of each standby node on sonar data during faults, and recording the data quantity and priority related to each behavior; calculating the cooperative contribution degree of each node by adopting a shape algorithm based on the recorded behavior data, wherein the value contribution of the node to the alliance is determined by the total value of the preserved sonar data; And according to the cooperative contribution ratio of each node, correspondingly improving the priority of each node in a subsequent request scheduling queue, and increasing the data receiving bandwidth quota for each node.
  9. 9. The sonar database multi-node hot standby and intelligent switching management method of claim 8, wherein based on cooperative behavior data, an improved shape algorithm is adopted to calculate the cooperative contribution degree of each node, specifically: For a collaboration subset composed of partial backup nodes Define its contribution value The value of single data is determined by the data priority, the data quantity and the time attenuation coefficient of the single data; for any backup node i, enumerating that it joins each possible subset of collaboration And calculates the value increment brought by the situation as marginal contribution The calculation formula is as follows: the degree of collaboration contribution of node i is determined as the average of its marginal contributions in all possible collaboration subset permutations.

Description

Sonar database multi-node hot standby and intelligent switching management method Technical Field The invention relates to the technical field of database management, in particular to a sonar database multi-node hot standby and intelligent switching management method. Background Offshore formation collaboration has become a core model for improving the overall performance of fleets. The sonar system is used as a main means of underwater situation awareness, and the data processing capacity of the sonar system directly determines the anti-submergence, underwater defense and target locking capacity of formation. Under the formation operation scene, a plurality of ships are interconnected and intercommunicated through a data link between ships to form a distributed sonar data processing system, so that the sharing and fusion of sonar original data, signal processing results and target track information are realized, a unified and continuous underwater operation scene situation map is formed, and the decision of a commander is supported. The sonar data has the characteristics of large data volume, strong real-time performance and uneven value density, and efficient distribution and cooperative processing among multiple nodes are needed. Specifically, the requirements on transmission bandwidth, processing delay and reliability are different from each other for different types of data such as detection echoes of active sonar, broadband noise data of passive sonar, instruction streams of underwater acoustic communication and the like. Based on the above, the system generally adopts a main-standby architecture design, namely, a ship is elected as a main node to bear core data storage and processing tasks, and other nodes are used as backup nodes to provide redundant calculation and storage resources so as to ensure that sonar data service is not interrupted when the main node fails. However, the traditional sonar database master-slave management method mostly adopts a fixed master node or a simple heartbeat detection mechanism, and has a plurality of limitations. Firstly, the data distribution strategy is statically stiff, and cannot be dynamically adjusted according to the real-time processing capacity (such as CPU/GPU load and memory occupation) of each ship node and the communication quality (such as time delay, packet loss rate and signal strength of satellite/radio link) of the marine link, so that the high-load node is overloaded and the light-load node is idle, or the data is still forcedly transmitted when the link quality is deteriorated, thereby causing data backlog and processing delay. And secondly, the health state monitoring means is single, only depends on 'dead' heartbeats periodically reported by the nodes, lacks cross verification on the real processing capacity of the nodes and the survival state of the platform, is difficult to identify complex fault modes such as network partition, node dead, malicious deception and the like, and has higher fault misjudgment rate. And thirdly, the main-standby switching decision lacks intelligent evaluation, a certain standby node is directly designated to take over after the main node is determined to be out of connection, key factors such as the data synchronization timeliness (such as log lag time of a fault main node), the survival state of a platform (such as ship body damage management and energy storage), the task endurance (such as residual fuel and replenishment stock) and the like of the standby node are not comprehensively considered, and the switching to the nodes with unsynchronized data, poor self state or insufficient endurance can be caused, so that secondary faults or task interruption can be caused. In addition, the traditional method lacks quantitative evaluation and excitation mechanisms for cooperative behaviors (such as caching, forwarding and processing sonar data) of each standby node in the switching process, so that the enthusiasm of the nodes to participate in cooperation cannot be fully mobilized, and the whole fault-tolerant capacity and the data security rate of the system are affected. Based on the above, the invention aims to solve the technical problems, and provides a sonar database multi-node hot standby and intelligent switching management method which realizes high availability, high reliability and intelligent management of an offshore formation sonar data processing system through dynamic data distribution, cross verification, consensus calculation and income evaluation. Disclosure of Invention In view of the above, the invention provides a sonar database multi-node hot standby and intelligent switching management method, which realizes high availability, high reliability and intelligent management of an offshore formation sonar data processing system through dynamic data distribution, cross validation, consensus calculation and income evaluation. In order to achieve the above purpose, the present invention provides the fo