Adaptive optimization system and method for a data model-based conductivity-temperature-depth instrument

Abstract

This invention discloses an adaptive optimization system and method for a temperature, salinity, and depth (TDM) measuring instrument based on a data model, relating to the field of measuring instrument optimization technology. The status feedback information includes S associated front-end measuring node identifiers and the first anomaly prediction duration corresponding to each front-end measuring node. Based on the S front-end measuring node identifiers contained in each of the M1 status feedback information sets, N measuring node identifiers are selected through deduplication and intersection analysis. Based on the selected N measuring node identifiers and their corresponding first anomaly prediction durations, the second optimization adjustment duration corresponding to each measuring node is calculated, and a global collaborative optimization command is sent to the control center of the second region. This optimization system, through adaptive optimization of the TDM measuring instrument using a data model, achieves dynamic perception and global collaborative optimization of the data acquisition and transmission links in the ocean observation network.

Description

Technical Field

[0001] This invention relates to the field of instrument optimization technology, specifically to an adaptive optimization system and method for temperature, salinity, and depth measuring instruments based on data models. Background Technology

[0002] In the field of marine environmental monitoring, temperature, salinity, and depth (TDM) meters are widely used to acquire key parameters such as seawater temperature, salinity, and depth. The quality of their data directly depends on the stability and real-time performance of the transmission link between the front-end measurement nodes and the data acquisition system. However, actual sea conditions are complex and variable, often resulting in abnormal congestion phenomena such as data transmission delays exceeding the baseline, increased sampling packet loss rates, or accelerated signal attenuation due to environmental interference. These phenomena severely affect the validity and continuity of the data.

[0003] Traditional monitoring methods lack the ability to aggregate and correlate abnormal information from multiple observation points, and it is difficult to predict the duration of anomalies and take optimization measures in advance, resulting in lagging regulation, blind resource allocation, and inability to achieve efficient and coordinated response in a wide sea area.

[0004] Therefore, there is an urgent need for an adaptive system that can sense the link status in real time, intelligently identify key abnormal measurement nodes, and combine predictive information to perform global optimization and adjustment, so as to improve the reliability and data acquisition efficiency of temperature, salinity and depth measurement networks in complex environments. Summary of the Invention

[0005] The purpose of this invention is to provide an adaptive optimization system and method for temperature, salinity, and depth measurement instruments based on data models, in order to address the shortcomings in the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an adaptive optimization method for a temperature, salinity, and depth measurement instrument based on a data model, comprising:

[0007] Feedback module: Real-time status feedback information is transmitted back to the main control analysis module of the first area through M1 local sensing units. The status feedback information of any local sensing unit is used to indicate that there is abnormal congestion in the temperature, salinity and depth data acquisition and transmission link of its associated S front-end measurement nodes. The status feedback information includes the identifiers of the associated S front-end measurement nodes and the first abnormal prediction duration corresponding to each front-end measurement node.

[0008] Main control analysis module: Based on the S front-end measurement node identifiers contained in each of the M1 status feedback information, the module filters out the identifiers of N measurement nodes through deduplication and intersection analysis. Based on the identifiers of the N measurement nodes and their corresponding first anomaly prediction durations, the module calculates the second optimized adjustment duration for each measurement node and sends a global collaborative optimization command to the control center of the second region.

[0009] Preferably, the main control analysis module calculates the second optimized adjustment time for each measurement node based on the identifiers of the selected N measurement nodes and their corresponding first anomaly prediction times:

[0010] The prediction duration of the first anomaly is adjusted by weighting coefficients based on the anomaly type and combining the importance score of the measurement node in the region.

[0011] The correction process analyzes the remaining safety margin of the abnormal duration and the time required to initiate and complete control measures to derive the execution time of the optimized action, i.e., the second optimized adjustment time.

[0012] Preferably, the main control analysis module, based on the S front-end measurement node identifiers contained in each of the M1 status feedback messages, performs deduplication and intersection analysis:

[0013] For the same measurement node identifier that appears repeatedly in different feedback information, only a unique entry is retained, and the first anomaly prediction duration corresponding to the measurement node is archived separately according to the source.

[0014] Preferably, the main control analysis module filters out the identifiers of N measurement nodes:

[0015] Establish a mapping table where the key is the unique identifier of the measurement node and the value is an accumulator of the occurrence count;

[0016] Read the list of measurement nodes in the feedback information one by one, increment the counter of each measurement node in the mapping table, and identify the measurement node that is jointly reported as abnormal by at least two local sensing units;

[0017] The average regional level is obtained by summing the first anomaly prediction duration of all measurement nodes and then dividing by the total number of measurement nodes.

[0018] For each measurement node, its first anomaly prediction duration is compared with the regional average. If the first anomaly prediction duration is more than 1.5 times the regional average, the measurement node is determined to be more abnormal than the regional average and is included in the category of key measurement nodes.

[0019] Preferably, the main control analysis module sends a global collaborative optimization command to the control center of the second region:

[0020] The key measurement node identifier is paired and encapsulated with the second optimization adjustment duration to form a global collaborative optimization instruction. The instruction structure includes measurement node location information, expected anomaly duration, suggested adjustment duration, and optimization strategy category.

[0021] The instruction is sent to the control center of the second region via the communication link. After receiving the instruction, the control center of the second region mobilizes the available resources in the region to carry out cross-regional collaborative link optimization, parameter adjustment or redundancy backup switching operations on the key measurement nodes involved.

[0022] Preferably, the main control analysis module will classify measurement nodes that meet either the condition of being marked as abnormal by at least two local sensing units or the first abnormality prediction time exceeding 1.5 times the regional average into a key measurement node set, forming a key measurement node identifier list of number N.

[0023] Preferably, the feedback module transmits real-time status feedback information back to the main control analysis module of the first region through M1 local sensing units. The status feedback information of any local sensing unit is used to indicate that there is abnormal congestion in the temperature, salinity, and depth data acquisition and transmission link of its associated S front-end measurement nodes.

[0024] A stable data acquisition and transmission link is established between M1 local sensing units and S front-end measurement nodes within their respective areas. The local sensing units have a built-in multi-dimensional status monitoring mechanism to continuously collect three types of indicators for each link:

[0025] The transmission delay of data packets from the front-end measurement node to the local sensing unit is compared with the preset baseline value in real time.

[0026] The packet loss rate of the sampled data during transmission is obtained by statistically analyzing the difference between the number of data packets sent and received within a fixed time window;

[0027] The degree of signal strength attenuation caused by environmental factors such as wind, waves, ocean currents, and electromagnetic noise is quantitatively assessed using received signal strength indicators or signal-to-noise ratio change curves.

[0028] When any indicator deviates from the normal range and reaches the congestion determination threshold, it is marked as an abnormal congestion state of the link.

[0029] Preferably, the status feedback information includes the identifiers of the S associated front-end measurement nodes and the first anomaly prediction duration corresponding to each front-end measurement node:

[0030] Perform mean and variance statistics on the delay data within the sliding observation window;

[0031] A time series fitting method is introduced, using the sliding observation window statistics and the latest delay value as the sequence input, and a trend model is constructed in chronological order to extrapolate the delay change trend over a future period.

[0032] By combining the rate of change of packet loss rate and signal attenuation, the delayed recovery time point output by the trend model is corrected to obtain an estimated duration reflecting the duration of the anomaly, which is the first anomaly prediction duration for the measurement node.

[0033] Preferably, the local sensing unit organizes the analysis results into a unified format of state feedback information, including:

[0034] The unique identifier set of the S front-end measurement nodes associated with the local sensing unit;

[0035] The first anomaly prediction duration value corresponding to each measurement node is kept in a one-to-one correspondence with the measurement node identifier.

[0036] This application also provides an adaptive optimization method for temperature, salinity, and depth measurement instruments based on a data model. The optimization method includes the following steps:

[0037] M1 local sensing units respectively transmit real-time status feedback information back to the main control analysis module of the first area. The status feedback information of any local sensing unit is used to indicate that there is abnormal congestion in the temperature, salinity and depth data acquisition and transmission link of its associated S front-end measurement nodes. The status feedback information includes the identifiers of the associated S front-end measurement nodes and the first abnormal prediction duration corresponding to each front-end measurement node.

[0038] Based on the S front-end measurement node identifiers contained in each of the M1 state feedback information, N measurement node identifiers are selected through deduplication and intersection analysis. Based on the selected N measurement node identifiers and their corresponding first anomaly prediction durations, the second optimized adjustment duration corresponding to each measurement node is calculated, and a global collaborative optimization instruction is sent to the control center of the second region.

[0039] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0040] This invention deploys multiple sets of local sensing units at different observation points in the target sea area, which can collect and transmit the operating status of each front-end measurement node in real time. It can accurately capture abnormal phenomena such as excessive data transmission delay, increased packet loss rate, and aggravated signal attenuation, so that the originally scattered and isolated abnormal information can be gathered and formed into a structured status feedback, which greatly improves the panoramic perception capability of the health of the temperature, salinity and depth measurement network and provides a reliable data foundation for subsequent optimization.

[0041] This invention utilizes deduplication and intersection analysis to intelligently filter out key measurement nodes that are jointly marked as abnormal or have a prominent degree of abnormality from massive feedback, rather than simply listing all abnormal points. This strategy of focusing on high-impact measurement nodes avoids the scattered consumption of resources on secondary issues, ensuring that optimization measures directly address the core bottlenecks that restrict the effectiveness of data, and significantly improves the efficiency and pertinence of anomaly localization.

[0042] This invention not only identifies abnormal measurement nodes, but also combines historical data and time series models to generate a first anomaly prediction duration, and calculates a second optimization adjustment duration accordingly. This gives the optimization decision a forward-looking time dimension, and allows for the advance allocation of control resources based on the expected duration of the anomaly, transforming passive response into proactive intervention and effectively reducing the impact of anomalies on the continuity and accuracy of long-term data collection.

[0043] This invention achieves global collaborative optimization across regions and levels by sending key measurement node information and optimization instructions to the control center of the second region. It breaks down the information silos of a single observation point or local area, and can coordinate and allocate a wider range of monitoring and processing resources to form a linkage optimization mechanism covering the target sea area and even adjacent areas. This improves the environmental adaptability and data effectiveness of the temperature, salinity and depth measurement network as a whole, and ensures the stable and efficient execution of marine observation missions in complex and ever-changing environments. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0045] Figure 1 This is a mind map of the optimized system for this invention. Detailed Implementation

[0046] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] Example: This example provides an adaptive optimization system for a temperature, salinity, and depth measurement instrument based on a data model. Please refer to [link / reference]. Figure 1 As shown, it includes:

[0048] Feedback Module: M1 local sensing units deployed at different observation points in the target sea area transmit real-time status feedback information back to the main control and analysis module of the first region. The status feedback information of any local sensing unit indicates that the temperature, salinity, and depth data acquisition and transmission link of its associated S front-end measurement nodes is experiencing abnormal congestion. This is specifically manifested as data transmission delays exceeding a preset baseline, increased packet loss rate of sampled data, or increased signal attenuation due to environmental interference, all of which affect data validity. This status feedback information contains two core components: first, the unique identifiers of the associated S front-end measurement nodes (used to distinguish different sampling locations); and second, the first anomaly prediction duration for each front-end measurement node (i.e., the preliminary estimated duration of the abnormal state based on the current and recent historical data characteristics of the measurement node, for example, obtained by using a sliding window to statistically analyze the mean and variance of the recent 5 transmission delays, combined with a time series model fitting). The identifiers of the S front-end measurement nodes contained in each of the M1 status feedback messages are sent to the main control and analysis module.

[0049] The main control analysis module, based on the S front-end measurement node identifiers contained in each of the M1 status feedback messages, filters out N key measurement node identifiers (N>0) through deduplication and intersection analysis. The filtering logic does not simply list all measurement nodes that report anomalies, but focuses on measurement nodes that report anomalies across multiple local sensing units or whose single measurement node anomaly level is significantly higher than the regional average. For example, if a measurement node is marked as anomaly by at least two local sensing units, or its first anomaly prediction duration exceeds 1.5 times the regional average prediction duration, it is determined to be a key measurement node requiring global collaborative optimization. Based on the identifiers of the N key measurement nodes and their corresponding first anomaly prediction durations, the second optimization adjustment duration for each measurement node is further calculated, and a global collaborative optimization command is sent to the control center of the second region (which may be an adjacent observation area of ​​the target sea area or the area where the data processing center is located).

[0050] This embodiment provides an adaptive optimization method for a temperature, salinity, and depth measurement instrument based on a data model. The optimization method includes the following steps:

[0051] S1: M1 local sensing units deployed at different observation points in the target sea area transmit real-time status feedback information back to the main control and analysis module of the first region. The status feedback information of any local sensing unit indicates that there is abnormal congestion in the temperature, salinity, and depth data acquisition and transmission link of its associated S front-end measurement nodes. Specifically, this manifests as data transmission delays exceeding the preset baseline, increased packet loss rate of sampled data, or increased signal attenuation due to environmental interference, all of which affect the validity of the data. This status feedback information contains two core components: first, the unique identifiers of the associated S front-end measurement nodes (used to distinguish different sampling locations); and second, the first anomaly prediction duration for each front-end measurement node (i.e., the preliminary estimated duration of the abnormal state based on the current and recent historical data characteristics of the measurement node, for example, obtained by using a sliding window to statistically analyze the mean and variance of the recent 5 transmission delays and fitting it with a time series model).

[0052] S2: The main control analysis module, based on the S front-end measurement node identifiers contained in each of the M1 status feedback messages, filters out the identifiers of N key measurement nodes (N>0) through deduplication and intersection analysis. The filtering logic is not simply listing all measurement nodes that have reported anomalies, but focusing on measurement nodes that report anomalies across multiple local sensing units or whose single measurement node anomaly level is significantly higher than the regional average. For example, if a measurement node is marked as anomaly by at least 2 local sensing units, or if its first anomaly prediction time exceeds 1.5 times the regional average prediction time, it is determined to be a key measurement node that requires global collaborative optimization.

[0053] S3: Based on the identifiers of the selected N key measurement nodes and their corresponding first anomaly prediction durations, the main control analysis module further calculates the second optimized adjustment duration for each measurement node and sends a global collaborative optimization instruction to the control center of the second region (which may be the observation area adjacent to the target sea area or the area where the data processing center is located).

[0054] This implementation also provides a detailed introduction to the functions of each module, as follows:

[0055] Preferably, M1 local sensing units are deployed at different observation points in the target sea area, and each unit establishes a stable data acquisition and transmission link connection with S front-end measurement nodes within its respective area. The local sensing units have a built-in multi-dimensional status monitoring mechanism, continuously collecting three types of key indicators for each link:

[0056] First, the transmission delay of data packets from the front-end measurement node to the local sensing unit is compared with a preset baseline value in real time. Second, the packet loss rate of sampled data during transmission is obtained by statistically analyzing the difference in the number of data packets sent and received within a fixed time window. Third, the degree of signal strength attenuation caused by environmental factors such as wind, waves, ocean currents, and electromagnetic noise is quantitatively assessed using the Received Signal Strength Indication (RSSI) or Signal-to-Noise Ratio (SNR) change curves. When any indicator deviates from the normal range and reaches the congestion judgment threshold, it is marked as an abnormal congestion state in the link.

[0057] Preferably, for each front-end measurement node in an abnormal congestion state, the local sensing unit calls its built-in predictive analysis logic to estimate the expected duration of the abnormal state of that measurement node, i.e., the first abnormal prediction duration:

[0058] A sliding observation window is set, for example, the five most recent transmission delay samples. The mean and variance of the delay data within this window are statistically analyzed to characterize the central tendency and fluctuation range of recent delay levels. Then, a time series fitting method (such as a process based on the autoregressive moving average) is introduced. The window statistics and the latest delay value are used as sequence inputs to construct a trend model in chronological order, extrapolating the delay change trend over a future period. The delay recovery time point output by the trend model is corrected by combining the rate of change of packet loss rate and signal attenuation, ultimately obtaining an estimated value reflecting the possible duration of the anomaly, which is the predicted duration of the first anomaly at this measurement node. For example, if the five most recent delays are 120ms, 135ms, 140ms, 150ms, and 160ms, and the mean and variance show an increasing trend, and the time series fitting suggests that this trend will slow down to below the baseline after about 8 minutes, while the packet loss rate is rapidly decreasing, then the predicted duration of the first anomaly can be estimated to be about 8 minutes.

[0059] Set a sliding observation window of length 5, and take the five most recent transmission delay samples, for example, 120ms, 135ms, 140ms, 150ms, and 160ms respectively. First, calculate the mean and variance of the delay data within this window. The mean is calculated by adding the five values ​​and averaging them, i.e., (120+135+140+150+160)÷5=705÷5=141ms. The variance is calculated by averaging the sum of the squares of the differences between each value and the mean, i.e., [(120-141)²+(135-141)²+(140-141)²+(150-141)²+(160-141)²]÷5=[441 [+36+1+81+361]÷5=920÷5=184 (ms²). The mean shows an increasing trend and the variance is large, indicating that the recent latency has not only increased horizontally but also fluctuated significantly. Subsequently, based on the time series fitting of the autoregressive moving average idea, the window statistics (mean 141ms, variance 184ms²) and the latest latency value of 160ms are arranged in chronological order to construct a trend model. Extrapolating future latency changes, the model analysis suggests that the latency will fall back to below the preset baseline (assuming the baseline is 100ms) in about 8 minutes, that is, the trend recovery point corresponds to the current time plus 8 minutes. On this basis, the changes in packet loss rate and signal attenuation rate are combined for correction.

[0060] For example, if the current packet loss rate is rapidly decreasing and the signal attenuation is stabilizing, it indicates that the cause of congestion is weakening and will not cause additional delays in recovery. Therefore, the 8-minute prediction value should not be extended; instead, it can be slightly shortened to close to 8 minutes. In summary, the calculation of the first anomaly prediction duration can be simplified as follows: first, calculate the moving window mean and variance to determine the trend direction; then, use time series fitting to obtain the time required for delay recovery, T_trend; finally, perform linear or nonlinear correction ΔT based on the rate of change of packet loss rate and the rate of change of signal attenuation. The formula is: first anomaly prediction duration ≈ T_trend + ΔT. In this example, T_trend = 8 minutes and ΔT ≈ 0 minutes, so the first anomaly prediction duration ≈ 8 minutes, thus achieving a quantitative prediction of the duration of the anomaly.

[0061] Preferably, the local sensing unit organizes the above analysis results into a unified format of state feedback information, which consists of two core parts:

[0062] The first part is a unique set of identifiers for the S front-end measurement nodes associated with this unit, typically using geographic coordinate encoding or device serial numbers, to ensure that different sampling locations can be clearly distinguished on the main control analysis module side. The second part is the first anomaly prediction duration value corresponding to each measurement node, maintaining a one-to-one correspondence with the measurement node identifier. This encapsulation process ensures that the information is semantically intact during transmission and can be automatically parsed, avoiding subsequent analysis failures due to identifier confusion or missing durations.

[0063] Let S be the S front-end measurement nodes associated with a certain local sensing unit, with their unique identifier set denoted as ID_set={ID1,ID2,…,ID_S}, and the corresponding first anomaly prediction duration set denoted as T_set={T1,T2,…,T_S}, where Ti represents the estimated anomaly duration of node IDi. The encapsulation process follows the principle of mapping integrity: for each i∈[1,S], it must be ensured that IDi and Ti appear in pairs without omission or misalignment. This can be abstracted into a construction rule: state feedback information = {(ID1,T1),(ID2,T2),…,(ID_S,T_S)}.

[0064] Suppose a local sensing unit is associated with S=3 front-end measurement nodes, whose unique identifiers are: ID1="Node_A_120.35E_25.78N (geographic coordinate code), ID2="Node_B_120.36E_25.79N", and ID3="Node_C_120.37E_25.77N". The predicted durations of the first anomaly, obtained through predictive analysis, are T1=8min, T2=5min, and T3=12min, respectively. According to the above mapping rules, the state feedback information of this unit... This can be represented as: {(Node_A_120.35E_25.78N, 8min),(Node_B_120.36E_25.79N, 5min),(Node_C_120.37E_25.77N, 12min)}. In this structure, any node identifier can be directly indexed to its corresponding prediction duration on the main control analysis module side, avoiding errors in subsequent deduplication and intersection analysis due to identifier confusion or missing duration, thereby achieving semantic integrity and reliable automatic parsing of information during transmission.

[0065] Preferably, each local sensing unit sends its encapsulated status feedback information to the main control analysis module of the first area according to a preset communication protocol and scheduling cycle. The feedback process employs a reliable transmission mechanism (such as introducing acknowledgment retransmission or redundancy verification in unreliable wireless environments) to ensure that critical information arrives intact even if some links are disrupted. All S×M1 front-end measurement node identifiers contained in the feedback information sent by the M1 local sensing units are centrally aggregated on the main control analysis module side, laying the data foundation for subsequent deduplication and intersection analysis.

[0066] Preferably, the main control analysis module receives M1 status feedback messages from M1 local sensing units. Each message contains unique identifiers of S front-end measurement nodes associated with that unit and the corresponding first anomaly prediction duration. The module extracts the measurement node identifiers from all feedback messages, merges them into a unified measurement node list, and performs deduplication.

[0067] For the same measurement node identifier that appears repeatedly in different feedback messages, only a unique entry is retained. Simultaneously, the first anomaly prediction duration corresponding to that measurement node is archived separately according to its source, so as to facilitate subsequent determination of the frequency and anomaly distribution of the measurement node being marked by different local sensing units. Deduplication ensures that each measurement node is counted as an independent object only once in subsequent analysis, avoiding statistical bias and redundant calculations.

[0068] Preferably, in the deduplicated set of measurement nodes, the main control analysis module traverses each measurement node and counts the number of times it is marked as abnormal in M1 feedback messages, that is, how many different status feedback messages the measurement node appears in:

[0069] A mapping table is established, with the key being the unique identifier of the measurement node and the value being an accumulator of occurrence counts. The list of measurement nodes in the feedback information is read one by one, and the counter for each measurement node in the mapping table is incremented. This statistical analysis identifies measurement nodes that are jointly reported as abnormal by at least two local sensing units. These measurement nodes are often located in weak links within the shared sensing range of multiple observation points or are significantly affected by regional environmental interference, thus possessing higher global optimization priority. For example, if measurement node X is listed in all three feedback reports from observation points A, B, and C, its cross-unit anomaly frequency is 3, clearly meeting the criteria for a key measurement node marked as abnormal by at least two local sensing units.

[0070] In the deduplicated node set, the main control analysis module uses a frequency accumulation mapping algorithm to count the number of times each node is marked as an anomaly in M1 feedback messages. The calculation logic can be described as follows: Let the node set be U={u1,u2,…,un}, and the M1 feedback messages be F1, F2,…,F_M1, where each F_k message contains several node identifiers. A mapping table CountMap is established, with the key being the unique node identifier u, initially set to 0. Then, iterates from k=1 to M1, and for each node u in F_k, executes CountMap[u]←CountMap[u]+1. After the iteration, the value of CountMap[u] is the number of times node u is reported as an anomaly by different local sensing units, denoted as Occur(u). The judgment condition is: if Occur(u)≥2, then the node is marked as an anomaly by at least two local sensing units and can be considered a candidate for a key node.

[0071] Suppose M1 = 4 feedback messages come from observation points A, B, C, and D respectively, where:

[0072] F1 (observation point A) contains nodes X and Y;

[0073] F2 (observation point B) includes nodes X and Z;

[0074] F3 (observation point C) contains nodes Y and W;

[0075] F4 (observation point D) contains nodes X and W.

[0076] Initialize CountMap[X]=0, CountMap[Y]=0, CountMap[Z]=0, CountMap[W]=0.

[0077] Traverse F1: X → CountMap[X] = 1, Y → CountMap[Y] = 1;

[0078] Traverse F2: X → CountMap[X] = 2, Z → CountMap[Z] = 1;

[0079] Traverse F3: Y → CountMap[Y] = 2, W → CountMap[W] = 1;

[0080] Traverse F4: X → CountMap[X] = 3, W → CountMap[W] = 2.

[0081] We obtain Occur(X)=3, Occur(Y)=2, Occur(Z)=1, and Occur(W)=2. Based on the formula for key node determination: Occur(u)≥2, nodes X, Y, and W satisfy the condition. Node X appears in all three feedback signals, meeting the condition of being marked as an anomaly by at least two local sensing units, and can be prioritized as a common weak link across regions for global optimization.

[0082] Preferably, the main control analysis module collects the first anomaly prediction time for each of all deduplicated measurement nodes (if the same measurement node has multiple source values, the maximum or average value can be taken as the representative value of that measurement node to ensure the robustness of the judgment), and obtains the regional average first anomaly prediction time by traversing, accumulating and counting:

[0083] The first anomaly prediction duration of all measurement nodes is summed and divided by the total number of measurement nodes to obtain the regional average. Then, for each measurement node, its first anomaly prediction duration is compared to the regional average. If this value is greater than 1.5 times the regional average, the measurement node is determined to have a significantly higher anomaly degree than the regional average, and is included in the category of critical measurement nodes even if it is not jointly marked by multiple local sensing units. This step can capture measurement nodes that are locally isolated but have extremely strong anomaly persistence, preventing serious single-point anomalies from being overlooked. For example, if the regional average prediction duration is 10 minutes, and a measurement node's prediction duration is 16 minutes, it is identified as a critical measurement node because it exceeds 1.5 times the threshold (15 minutes).

[0084] Let the deduplicated node set be U={u1,u2,…,un}. For node ui, if it has multiple first anomaly prediction duration values ​​in the feedback from different local sensing units, first take its representative value Rep(ui), which can be either the maximum value or the mean value, to ensure a robust estimate of the anomaly persistence. Then, traverse all nodes, accumulate the sum of these representative values, and count the total number of nodes n. Calculate the region's average first anomaly prediction duration using the formula AvgRegion=(∑_{i=1}^{n}Rep(ui))÷n. Then, for each node ui, compare Rep(ui) with 1.5×AvgRegion: if Rep(ui)>1.5×AvgRegion, then determine that the node's anomaly degree is significantly higher than the region mean, and include it in the key node set even if it is only marked by a single local sensing unit.

[0085] Let the deduplicated node set U contain 5 nodes with representative values: Rep(u1) = 8 min, Rep(u2) = 12 min, Rep(u3) = 9 min, Rep(u4) = 10 min, and Rep(u5) = 11 min. First, calculate the sum ∑Rep = 8 + 12 + 9 + 10 + 11 = 50 min. Since the number of nodes n = 5, the average region AvgRegion = 50 ÷ 5 = 10 min. The threshold Thr = 1.5 × AvgRegion = 1.5 × 10 = 15 min. Then, compare each node: Rep(u1) = 8 < 15, so it's not selected; Rep(u2) = 12 < 15, so it's not selected; Rep(u3) = 9 < 15, so it's not selected; Rep(u4) = 10 < 15, so it's not selected; Rep(u5) = 11 < 15, so it's not selected. In this example, no node exceeds the threshold. If Rep(u2) is changed to 16min, then 16>15, and node u2 will be included in the critical node. Even if it only appears in the feedback of a single local sensing unit, it can be captured due to the outstanding persistence of the anomaly.

[0086] Preferably, based on the comprehensive judgment results, the main control analysis module will classify measurement nodes that meet either the condition of being marked as abnormal by at least two local sensing units or the first abnormality prediction time exceeding 1.5 times the regional average into a key measurement node set, forming a key measurement node identification list of number N (N>0). This set eliminates ordinary abnormal measurement nodes that are only slightly marked by a single unit, focusing on locations that pose a significant threat to the effectiveness of global data acquisition, ensuring highly targeted subsequent optimization of resource investment.

[0087] For each measurement node in the set of key measurement nodes, the main control analysis module calculates the second optimization adjustment time based on its first anomaly prediction duration and anomaly type characteristics:

[0088] First, weighting coefficients are assigned based on the anomaly type (e.g., higher weighting for data transmission delay anomalies, medium weighting for packet loss rate anomalies, and lower weighting for signal attenuation anomalies). Then, the weighted correction of the first anomaly prediction duration is applied, taking into account the importance score of the measurement node in the region (which can be determined by historical data contribution or the representativeness of the sea area). The correction process can consider the remaining safety margin of the anomaly duration and the time required to initiate and complete control measures. Thus, while ensuring the forward-looking nature of the control, the recommended execution duration of the optimized action is derived, i.e., the second optimized adjustment duration. This duration reflects both the expected length of the anomaly itself and the realistic time window for implementing control.

[0089] Let the set of key nodes be K={k1,k2,…,k_N}. For node ki, let its first anomaly prediction duration be T1i, the weight coefficient corresponding to the anomaly type be wi (the value range depends on normalization, for example, data transmission delay type w=0.9, packet loss rate type w=0.6, signal attenuation type w=0.3), and the importance score of the node in the region be si (which can be normalized to the range of 0~1 based on historical data contribution or the representativeness of the sea area). A remaining safety margin factor mi (value 0~1, reflecting the proportion of time available for regulation before the anomaly ends) and a regulation implementation time factor t_impli (normalized representation of the proportion of time required to start and complete regulation) are introduced. The formula for calculating the second optimized adjustment duration T2i can be expressed as:

[0090] T2i = T1i × wi × si × mi - t_impli × T1i. First, the anomaly prediction duration is amplified and corrected according to the importance of the anomaly type and the region. Then, the effective duration that can be actually used for regulation is calculated based on the remaining safety margin. Finally, the time required for regulation itself is deducted to obtain the duration for which the recommended optimization action should be maintained or executed.

[0091] Suppose that for a critical node k1, T11 = 10 min, the anomaly type is data transmission delay (w1 = 0.9), the regional importance score is s1 = 0.8, the remaining safety margin is m1 = 0.7 (meaning that control can be implemented for 70% of the entire anomaly period), and the control implementation time ratio is t_impl1 = 0.2 (meaning that 20% of the predicted time is required to complete the control initiation and execution). Substituting into the formula: T21 = 10 × 0.9 × 0.8 × 0.7 - 0.2 × 10 = 10 × 0.504 - 2 = 5.04 - 2 = 3.04 min. That is, the suggested second optimized adjustment time for this node is approximately 3.04 minutes, meaning that the control action should continue or be completed within this time window to match the expected progress of the anomaly and leave sufficient forward margin.

[0092] Preferably, the main control analysis module pairs and encapsulates key measurement node identifiers with a second optimized adjustment duration to form a global collaborative optimization command. The command structure clearly includes measurement node location information, expected anomaly duration, suggested adjustment duration, and optimization strategy category (such as switching transmission paths, increasing signal power, or temporarily increasing sampling frequency). Subsequently, this command is sent via a reliable communication link to the control center in the second region (which may be an adjacent observation area or the area where the data processing center is located). Upon receiving the command, the control center in the second region can mobilize available resources within its region to perform cross-regional collaborative link optimization, parameter adjustment, or redundant backup switching operations on the key measurement nodes involved, thereby achieving closed-loop control from state awareness to global response.

[0093] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0094] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. An adaptive optimization system for a temperature, salinity, and depth measurement instrument based on a data model, characterized in that: include: Feedback module: Real-time status feedback information is transmitted back to the main control analysis module of the first area through M1 local sensing units. The status feedback information of any local sensing unit is used to indicate that there is abnormal congestion in the temperature, salinity and depth data acquisition and transmission link of its associated S front-end measurement nodes. The status feedback information includes the identifiers of the associated S front-end measurement nodes and the first abnormal prediction duration corresponding to each front-end measurement node. Main control analysis module: Based on the S front-end measurement node identifiers contained in each of the M1 status feedback information, the module filters out the identifiers of N measurement nodes through deduplication and intersection analysis. Based on the identifiers of the N measurement nodes and their corresponding first anomaly prediction durations, the module calculates the second optimized adjustment duration for each measurement node and sends a global collaborative optimization command to the control center of the second region.

2. The adaptive optimization system for a temperature, salinity, and depth measurement instrument based on a data model according to claim 1, characterized in that: The main control analysis module calculates the second optimized adjustment time for each measurement node based on the identifiers of the selected N measurement nodes and their corresponding first anomaly prediction times. The prediction duration of the first anomaly is adjusted by weighting coefficients based on the anomaly type and combining the importance score of the measurement node in the region. The correction process analyzes the remaining safety margin of the abnormal duration and the time required to initiate and complete control measures to derive the execution time of the optimized action, i.e., the second optimized adjustment time.

3. The adaptive optimization system for a data model-based temperature, salinity, and depth measuring instrument according to claim 2, characterized in that: The main control analysis module, based on the S front-end measurement node identifiers contained in each of the M1 status feedback messages, performs deduplication and intersection analysis: For the same measurement node identifier that appears repeatedly in different feedback information, only a unique entry is retained, and the first anomaly prediction duration corresponding to the measurement node is archived separately according to the source.

4. The adaptive optimization system for a data model-based temperature, salinity, and depth measuring instrument according to claim 3, characterized in that: The main control analysis module filters out the identifiers of N measurement nodes: Establish a mapping table where the key is the unique identifier of the measurement node and the value is an accumulator of the occurrence count; Read the list of measurement nodes in the feedback information one by one, increment the counter of each measurement node in the mapping table, and identify the measurement node that is jointly reported as abnormal by at least two local sensing units; The average regional level is obtained by summing the first anomaly prediction duration of all measurement nodes and then dividing by the total number of measurement nodes. For each measurement node, its first anomaly prediction duration is compared with the regional average. If the first anomaly prediction duration is more than 1.5 times the regional average, the measurement node is determined to be more abnormal than the regional average and is included in the category of key measurement nodes.

5. The adaptive optimization system for a data model-based temperature, salinity, and depth measurement instrument according to claim 4, characterized in that: The main control analysis module sends a global collaborative optimization command to the control center of the second region: The key measurement node identifier is paired and encapsulated with the second optimization adjustment duration to form a global collaborative optimization instruction. The instruction structure includes measurement node location information, expected anomaly duration, suggested adjustment duration, and optimization strategy category. The instruction is sent to the control center of the second region via the communication link. After receiving the instruction, the control center of the second region mobilizes the available resources in the region to carry out cross-regional collaborative link optimization, parameter adjustment or redundancy backup switching operations on the key measurement nodes involved.

6. The adaptive optimization system for a data model-based temperature, salinity, and depth measurement instrument according to claim 2, characterized in that: The main control analysis module will classify measurement nodes that meet either the condition of being marked as abnormal by at least two local sensing units or the prediction time of the first abnormality exceeding 1.5 times the regional average into a key measurement node set, forming a key measurement node identifier list of number N.

7. The adaptive optimization system for a data model-based temperature, salinity, and depth measurement instrument according to claim 1, characterized in that: The feedback module transmits real-time status feedback information back to the main control analysis module of the first region through M1 local sensing units. The status feedback information of any local sensing unit is used to indicate that there is abnormal congestion in the temperature, salinity, and depth data acquisition and transmission link of its associated S front-end measurement nodes. A stable data acquisition and transmission link is established between M1 local sensing units and S front-end measurement nodes within their respective areas. The local sensing units have a built-in multi-dimensional status monitoring mechanism to continuously collect three types of indicators for each link: The transmission delay of data packets from the front-end measurement node to the local sensing unit is compared with the preset baseline value in real time. The packet loss rate of the sampled data during transmission is obtained by statistically analyzing the difference between the number of data packets sent and received within a fixed time window; The degree of signal strength attenuation caused by environmental factors such as wind, waves, ocean currents, and electromagnetic noise is quantitatively assessed using received signal strength indicators or signal-to-noise ratio change curves. When any indicator deviates from the normal range and reaches the congestion determination threshold, it is marked as an abnormal congestion state of the link.

8. The adaptive optimization system for a data model-based temperature, salinity, and depth measuring instrument according to claim 7, characterized in that: The status feedback information includes the identifiers of the S associated front-end measurement nodes and the first anomaly prediction duration corresponding to each front-end measurement node: Perform mean and variance statistics on the delay data within the sliding observation window; A time series fitting method is introduced, using the sliding observation window statistics and the latest delay value as the sequence input, and a trend model is constructed in chronological order to extrapolate the delay change trend over a future period. By combining the rate of change of packet loss rate and signal attenuation, the delayed recovery time point output by the trend model is corrected to obtain an estimated duration reflecting the duration of the anomaly, which is the first anomaly prediction duration for the measurement node.

9. The adaptive optimization system for a data model-based temperature, salinity, and depth measurement instrument according to claim 8, characterized in that: The local sensing unit organizes the analysis results into a unified format of state feedback information, including: The unique identifier set of the S front-end measurement nodes associated with the local sensing unit; The first anomaly prediction duration value corresponding to each measurement node is kept in a one-to-one correspondence with the measurement node identifier.

10. An adaptive optimization method for a temperature, salinity, and depth measuring instrument based on a data model, implemented by the optimization system described in any one of claims 1-9, characterized in that: The optimization method includes the following steps: M1 local sensing units respectively transmit real-time status feedback information back to the main control analysis module of the first area. The status feedback information of any local sensing unit is used to indicate that there is abnormal congestion in the temperature, salinity and depth data acquisition and transmission link of its associated S front-end measurement nodes. The status feedback information includes the identifiers of the associated S front-end measurement nodes and the first abnormal prediction duration corresponding to each front-end measurement node. Based on the S front-end measurement node identifiers contained in each of the M1 state feedback information, N measurement node identifiers are selected through deduplication and intersection analysis. Based on the selected N measurement node identifiers and their corresponding first anomaly prediction durations, the second optimized adjustment duration corresponding to each measurement node is calculated, and a global collaborative optimization instruction is sent to the control center of the second region.

Images