Distributed observation method for underwater robot networking evaluation

Abstract

The invention provides a distributed observation method for underwater robot networking evaluation, and the method comprises the steps: arranging a plurality of underwater observation nodes in a target observation water area, and collecting underwater environment data, including water depth, water temperature, water velocity and water quality, through sensors disposed at the observation nodes; preprocessing the underwater environment data, and unifying data of different sources and different scales to the same preset scale to obtain preprocessed underwater environment data; based on the underwater environment data, the information transmission rate and the information loss rate of robot networking under different water depth, water temperature, water flow velocity and water quality conditions are tested, and the robot networking performance under different conditions is evaluated according to test results; through distributed observation of underwater environment data, evaluation results of the networking performance of the robot under different underwater environment data conditions are obtained, and the networking performance of the underwater robot is evaluated through the evaluation results.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and particularly to a distributed observation method for underwater robot networking evaluation. Background Art

[0002] In the distributed observation of underwater robot networking evaluation, there are many technical problems. First, the underwater environment is complex and changeable, and the water depth, water temperature, water flow velocity, and water quality data collected by each observation node vary greatly. How to preprocess and standardize the data to ensure the consistency and comparability of the data is an urgent problem to be solved. Second, underwater communication is affected by factors such as water absorption and noise interference, with limited transmission bandwidth and high latency. How to design an efficient and reliable networking protocol to ensure real-time data transmission and synchronization is another technical challenge. Third, underwater robots are affected by water pressure, water temperature, water flow, and water quality factors, resulting in a decline in networking performance. It is necessary to test the information transmission rate and information loss rate of robot networking performance under different conditions. Finally, the test results under different conditions are directly related to the performance evaluation of the networking system. How to design a reasonable evaluation and dynamically adjust the networking strategy and observation plan based on the evaluation results is also an urgent problem to be deeply studied. Summary of the Invention

[0003] The present invention provides a distributed observation method for underwater robot networking evaluation, mainly including:

[0004] Arrange a number of underwater observation nodes in the target observation water area, and collect underwater environment data through sensors set on the observation nodes, including water depth, water temperature, water flow velocity, and water quality;

[0005] Preprocess the underwater environment data, and unify data from different sources and different scales to the same preset scale to obtain preprocessed underwater environment data;

[0006] Adopt a preset underwater networking protocol to transmit the preprocessed underwater environment data from each distributed observation node to the central node, and perform an analysis of the impact of the preprocessed underwater environment data on the networking performance, and respectively analyze the influence laws of water depth, water temperature, water flow velocity, and water quality on the underwater robot networking performance;

[0007] Based on the underwater environment data, test the information transmission rate and information loss rate of robot networking under different water depth, water temperature, water flow velocity, and water quality conditions, and evaluate the robot networking performance under different conditions according to the test results;

[0008] Through the distributed observation of the underwater environment data, obtain the evaluation results of the robot networking performance under different underwater environment data conditions, and evaluate the networking performance of the underwater robot through the evaluation results;

[0009] Based on the evaluation results of the networking performance of robots under different water depth conditions, establish communication protocols for different water depths, adjust the transmission power and receiving sensitivity of the robots according to the attenuation characteristics of signals propagating in water, and layer the robots according to the water depth, and design different network topologies respectively;

[0010] Based on the evaluation results of the networking performance of robots under different water temperature conditions, establish the mapping relationship between temperature and networking performance, and adjust the working parameters of the robots, including battery management and sensor sensitivity, to adapt to different temperature conditions, and construct a water temperature distribution map to form a networking layout strategy for the robots;

[0011] Based on the evaluation results of the networking performance of robots under different flow velocity conditions, establish a relationship model between flow velocity, communication delay, and data packet loss rate, optimize the network topology and data transmission strategy of robot networking, and at the same time divide the priority of transmission tasks to achieve flow velocity-driven task scheduling;

[0012] Based on the evaluation results of the impact of different water quality conditions on the networking performance of robots, construct a water quality-driven task scheduling strategy according to the changes in water quality, divide the water quality into regions, dynamically adjust the priority of task processing, and give priority to processing data collection in water quality-sensitive areas.

[0013] The technical solution provided by the embodiment of the present invention may include the following beneficial effects:

[0014] The present invention discloses a distributed observation method for underwater robot networking evaluation. By arranging distributed observation nodes in the target water area to collect environmental data such as water depth, water temperature, water flow velocity, and water quality, the original data is preprocessed and then transmitted to the central node. The present invention analyzes the influence laws of different environmental factors on the networking performance of robots, and tests and evaluates the information transmission rate and loss rate under various conditions. For the evaluation results, the present invention designs a series of optimization strategies, including adjusting the communication protocol and network topology according to the water depth; establishing the mapping relationship between water temperature and performance, and adjusting the working parameters; constructing the model of flow velocity and communication delay, and optimizing the network topology and transmission strategy; dynamically adjusting the task priority according to the changes in water quality. Through the distributed observation and analysis of underwater environmental data, the present invention realizes the comprehensive evaluation and targeted optimization of the networking performance of robots, provides a decision-making basis for the observation and networking performance of underwater robots, and effectively improves the adaptability and reliability of robot networking in complex underwater environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a flowchart of a distributed observation method for underwater robot networking evaluation according to the present invention.

[0016] Figure 2 It is a schematic diagram of a distributed observation method for underwater robot networking evaluation according to the present invention.

[0017] Figure 3 This is another schematic diagram of a distributed observation method for underwater robot network evaluation according to the present invention. Specific implementation manners

[0018] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this specification. Obviously, the described embodiments are only a part of the embodiments of this specification, rather than all the embodiments. Based on the embodiments in this specification, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of this specification.

[0019] Such as Figures 1-3 , a distributed observation method for underwater robot network evaluation in this embodiment may specifically include:

[0020] Step S101, arranging a plurality of underwater observation nodes in the target observation water area, and collecting underwater environment data through sensors arranged on the observation nodes, including water depth, water temperature, water flow velocity, and water quality.

[0021] Using distributed observation nodes to obtain water temperature data at the surface layer, middle layer, and deep layer. When the temperature difference between two adjacent data collections exceeds a preset threshold, the observation node adjusts from the first sampling interval to the second sampling interval; for the water flow data collected by the observation node, obtaining three-dimensional vector components through an acoustic Doppler current profiler, and obtaining the actual flow velocity value according to the included angle between the vector component and the observation node, and calibrating the actual flow velocity value according to a preset flow velocity interval; receiving dissolved oxygen data, turbidity data, and conductivity data collected by the observation node, and storing the data in a hierarchical manner according to a preset parameter interval. If the dissolved oxygen is lower than the first threshold and the conductivity is higher than the second threshold, then trigger the observation node to adjust to the second sampling interval; according to the water depth data obtained by the observation node, obtaining an underwater topographic map with an isobath interval through triangular grid depth fitting, and supplementing the depth of the observation blind area in the underwater topographic map; for the water temperature data, the actual flow velocity value, the dissolved oxygen data, the turbidity data, and the conductivity data, constructing a multi-layer hydrological parameter distribution map corresponding to the sampling depth and the underwater topographic map.

[0022] Specifically, distributed observation nodes are arranged at 500-meter intervals to collect water temperature data in three water layers: the surface layer of 0-10 meters, the middle layer of 10-50 meters, and the deep layer below 50 meters. The collection frequency is set to once every 120 seconds. When the water temperature change value between two adjacent times exceeds 0.5 degrees Celsius, the sampling interval is automatically adjusted to once every 30 seconds and transmitted to the data storage node in real time via underwater acoustic communication. For the water flow velocity data collected by the observation nodes, an acoustic Doppler current profiler DS150 is used to decompose the velocity vector in three dimensions. The actual velocity value is calculated based on the angle between the velocity vector direction and the observation node. The velocity data is calibrated in segments at intervals of 0.5 meters per second, and a temperature-velocity correlation table is established in combination with the water temperature data. According to the dissolved oxygen, turbidity, and conductivity parameters collected by the water quality monitoring sensors, numerical segmentation is carried out in intervals of 2 mg / L for dissolved oxygen, 5 standard turbidity units for turbidity, and 0.5 mS / cm for conductivity to construct a three-parameter comparison grading table. When the dissolved oxygen concentration is lower than 4 mg / L and the conductivity exceeds 4.0 mS / cm, the sampling frequency is triggered to increase from 120 seconds to 30 seconds. Based on the water depth data collected by multiple observation nodes, a three-dimensional underwater topographic map with a 10-meter isobath interval is constructed by linear fitting of triangular grids to supplement the depth of the observation blind area. The water temperature, flow velocity, and water quality data are superimposed corresponding to the sampling depth and the topographic map to form a multi-layer hydrological parameter three-dimensional distribution map, which is stored in a hierarchical structure and a time-depth retrieval index table is established. Through 25 observation nodes arranged at 500-meter intervals in a rectangular observation area, multi-layer data is collected. A temperature sensor TW100 is set in the surface layer of 0-10 meters with a collection frequency of 120 seconds, a temperature sensor TW200 is set in the middle layer of 10-50 meters with a collection frequency of 180 seconds, and a temperature sensor TW300 is set in the deep layer below 50 meters with a collection frequency of 240 seconds. When the temperature sensor detects that the temperature difference between adjacent water layers exceeds 0.5 degrees Celsius, the sampling interval of this layer is automatically adjusted to 30 seconds, and data packets are transmitted to the storage node at a rate of 38.4 kbps through the underwater acoustic communication module HM500. The acoustic Doppler current profiler DS150 obtains velocity data in three dimensions. The velocity range of the X-axis is -2.5 to 2.5 meters per second, the velocity range of the Y-axis is -2.5 to 2.5 meters per second, and the velocity range of the Z-axis is -1.0 to 1.0 meters per second. The velocity levels are divided at intervals of 0.5 meters per second, and a temperature-velocity two-dimensional table is established by pairing with the temperature data. In water quality monitoring, the range of the dissolved oxygen sensor DO200 is 0-20 mg / L, and each 2 mg / L is divided into a grade interval; the range of the turbidity sensor TU100 is 0-100 standard turbidity units, and each 5 units is divided into a grade interval; the range of the conductivity sensor EC150 is 0-10 mS / cm, and each 0.5 mS / cm is divided into a grade interval. The water quality type is determined by the three-parameter numerical values through the comparison table.Observation nodes collect water depth data in a 6,000-square-meter observation area, calculate the water depth of the area through least-squares triangular grid fitting, draw isobaths at 10-meter intervals to form an underwater terrain surface, combine the measured temperature, flow velocity, and water quality data to form a hydrological parameter distribution map at corresponding depth points, store the data using a depth-stratified structure, and establish a two-dimensional hash index table for sampling time and depth.

[0023] Step S102: Preprocess the underwater environment data, and unify data from different sources and different scales to the same preset scale to obtain preprocessed underwater environment data.

[0024] Obtain the temperature gradient value at adjacent sampling points according to the water temperature sensor, perform outlier processing on data outside the preset temperature threshold range using wavelet basis functions, and obtain corrected temperature data through the three-times standard deviation principle; establish a sampling time reference for the corrected temperature data, process the water flow sensor data using a Butterworth low-pass filter, and obtain the water flow direction angle value according to the northward positive reference coordinate system; use the water quality sensor to obtain dissolved oxygen, turbidity, and conductivity data, perform anomaly judgment through preset water quality parameter thresholds, and perform spline interpolation on the water quality data according to the sampling time of the corrected temperature data; for the water depth sensor, use adaptive window median filtering to obtain depth sampling data; establish a unified depth reference based on the depth sampling data, and perform depth alignment on the corrected temperature data, water flow direction angle value, and water quality data using a linear interpolation algorithm to obtain multi-parameter associated data at unified depth sampling points.

[0025] Specifically, based on the data collected by the water temperature sensor at intervals of 10 meters, the temperature gradient values of adjacent sampling points are calculated. When the temperature change rate exceeds 0.2 degrees Celsius per meter or the temperature value exceeds the range of 0-35 degrees Celsius, the db4 wavelet basis function is used to remove outliers, and then the deviating data is eliminated by the three-time standard deviation principle, and the processed temperature data is resampled at a time interval of 120 seconds to obtain the corrected temperature value. For the water flow sensor data, a 4th-order Butterworth low-pass filter with a cutoff frequency of 2 Hz is set to filter out high-frequency noise, and the flow velocity data is resampled with a quantization step of 0.5 meters per second. The angle value of the water flow direction is converted according to the north positive reference coordinate system, and the temperature-flow velocity correlation data table is established in combination with the corrected temperature value. The dissolved oxygen, turbidity, and conductivity data collected by the water quality sensor within a horizontal radius of 20 meters are used to set the thresholds of 4 mg / L for dissolved oxygen, 20 turbidity units for turbidity, and 4.0 mS / cm for conductivity for abnormal judgment. The missing data is supplemented by natural boundary condition cubic spline interpolation, and the interpolated water quality data is aligned to the temperature sampling timestamp. According to the data collected by the water depth sensor, the median filter based on the sampling frequency adaptive window is used to remove outliers, the depth data is resampled at a quantized interval of 1 meter, and the temperature, flow rate, and water quality data are aligned to the unified depth sampling point using a linear interpolation algorithm, and the interpolation accuracy is ensured by data cross-validation. In the 100m×100m observation area, the water temperature sensor is arranged at a measuring point every 10 meters, with a total of 121 measuring points. Each measuring point collects temperature data every 5 meters within the water depth range of 0-50 meters, with a sampling frequency of 10 seconds. The temperature gradient of adjacent measuring points is calculated to be within the range of 0.05-0.15 degrees Celsius per meter. When data exceeding 0.2 degrees Celsius per meter or temperature values exceeding the range of 0-35 degrees Celsius appear, the db4 wavelet basis function 4-layer decomposition is used to remove outliers, and the calculated standard deviation is 0.12 degrees Celsius. Data points exceeding three times the standard deviation are eliminated, and the remaining temperature data points are resampled using 120-second interval linear interpolation. In water flow observation, a 4th-order Butterworth low-pass filter is used to set a 2 Hz cutoff frequency. After filtering out high-frequency noise, the velocity data range is 0.1-2.5 meters per second. The sampled data is quantized into integer multiples with a step size of 0.5 meters per second, and a polar coordinate reference system is established based on the true north direction to calculate the flow angle value of 0-360 degrees. In water quality monitoring, the dissolved oxygen values are distributed in the range of 2.5-8.5 mg / L, the turbidity is distributed in the range of 5-35 turbidity units, and the conductivity is distributed in the range of 1.2-6.8 millisieverts per centimeter. The sampled data within a radius of 20 meters are interpolated by natural boundary cubic spline, and the interpolation nodes are spaced 2 meters apart. In water depth observation, the original sampling frequency is 1 Hz, and an 11-point median filter window is set. The depth data is quantized at 1 meter intervals. The temperature, velocity, and water quality data are unified to the standard depth point using piecewise linear interpolation. The cross-validation root mean square error is less than 5% of the measurement accuracy.

[0026] In step S103, using a preset underwater networking protocol, the preprocessed underwater environment data is transmitted from each distributed observation node to the central node, and an analysis of the impact on the networking performance of the preprocessed underwater environment data is carried out. The impact laws of water depth, water temperature, water flow velocity, and water quality on the networking performance of the underwater robot are analyzed respectively.

[0027] Obtain the water temperature data, water depth data, and water quality data collected by the underwater acoustic communication link; divide the communication network levels according to the quantization interval of the depth sensor for the water depth data, obtain the data forwarding path by setting relay nodes at the network levels, and establish the mapping relationship between the water depth gradient and the communication delay according to the data forwarding path; calculate the sound speed values of each water layer using the water temperature data and salinity data, construct a decision tree model according to the sound speed values, and obtain the communication link quality evaluation result through the decision tree model; obtain the minimum-hop communication path by processing the flow velocity data through the greedy routing algorithm, and if it is detected that communication errors are caused by water flow disturbances, use forward error correction coding to compensate for the communication errors; obtain the throughput data, packet loss rate data, and end-to-end delay data according to the forward error correction coding compensation, and use the analytic hierarchy process to process the throughput data, packet loss rate data, end-to-end delay data, and water depth data, water temperature data, and water quality data to obtain the weight values of the performance evaluation indicators.

[0028] Specifically, the processed water temperature, water depth, and water quality data are transmitted through an underwater acoustic communication link. Based on the signal attenuation value collected within a 10-second measurement period for an underwater acoustic signal over a 50-meter transmission distance, the data is grouped into packets of 1024 bytes and transmitted at a transmission rate of 38.4 kilobits per second. The priority order is set according to three indicators: signal-to-noise ratio (SNR), bit error rate (BER), and delay of the data packet, and packets with an SNR higher than 15 dB are preferentially transmitted. For the collected water depth data, the underwater communication network levels are divided according to the 50-meter quantization interval of the depth sensor. Relay nodes with a communication radius of 100 meters are set at each level for data forwarding. A mapping table between the water depth gradient and communication delay is established based on path loss calculation. Forward error correction coding is used to correct the bit errors in the depth communication of each level, and the signal strength, path loss, and Doppler frequency shift values are recorded. According to the temperature data and salinity data collected by the water temperature sensor, the sound speed value of each water layer is calculated by combining the sound speed measurement formula, and the decision tree algorithm is used to calculate the communication link quality within each depth horizontal plane. A decision tree is constructed according to three features: signal strength, path loss, and Doppler frequency shift, and an association table between the acoustic channel characteristics and the network topology structure is generated. Based on the flow velocity data collected by the water flow sensor in the measurement area, the communication path is calculated by the greedy routing algorithm according to the principle of the minimum number of hops, and forward error correction coding compensation is performed on the communication bit errors caused by water flow disturbance. The three compensation results of throughput, packet loss rate, and end-to-end delay, together with the water depth, water temperature, and water quality data, form a performance evaluation index set, and the weight values of each index are determined through the analytic hierarchy process. Within a water area of 1000 meters × 1000 meters, 25 observation nodes are deployed with a node spacing of 50 meters. Each node is equipped with an acoustic communication device with a transmission rate of 38.4 kilobits per second and collects the underwater acoustic signal attenuation value with a 10-second measurement period. The signal attenuation within a 50-meter transmission distance is measured to be 0.5 dB per meter. The collected environmental data is grouped and packed into 1024-byte packets. The SNR threshold is set to 15 dB. Packets with an SNR higher than the threshold are set with a priority level of 1, and those lower than the threshold are set with a priority level of 2. Five network levels are divided at intervals of 50 meters within the water depth range of 0 - 200 meters, and five relay nodes with a communication radius of 100 meters are deployed at each level. The communication delay of the 0 - 50-meter level is measured to be 20 milliseconds, the 50 - 100-meter level is 40 milliseconds, the 100 - 150-meter level is 60 milliseconds, and the 150 - 200-meter level is 80 milliseconds. Forward error correction coding is used to reduce the bit error rate from 10% to 1%. Temperature data of 15 - 25 °C and salinity data of 29 - 35 per thousand are obtained at each level, and the calculated sound speed value is 1460 - 1540 m / s. Decision tree processing is performed on data of signal strength from -85 to -95 dB, path loss of 0.3 - 0.8 dB per meter, and Doppler frequency shift of -2 to 2 Hz to generate a network topology map of 20 nodes in 5 levels.Calculate the communication path based on the greedy routing method. Under the water flow disturbance of 0.5 - 2.5 m / s, use the forward error correction method to reduce the communication bit error rate to less than 0.1%. The measured throughput is 2.5 - 3.5 kbps, the packet loss rate is 0.5% - 1.5%, and the end-to-end delay is 150 - 250 ms. Through the analytic hierarchy process, the throughput weight is 0.5, the packet loss rate weight is 0.3, and the delay weight is 0.2.

[0029] Step S104: Based on the underwater environment data, test the information transmission rate and information loss rate of the robot network under different water depths, water temperatures, water flow velocities, and water quality conditions, and evaluate the performance of the robot network under different conditions according to the test results.

[0030] Collect the acoustic signals of the sensor array, perform spectral processing on the acoustic signals to obtain the acoustic wave propagation path loss data, and establish a water depth and signal attenuation curve based on the path loss data; obtain the sound velocity profile according to the water depth and signal attenuation curve, calculate the sound velocity values of each layer for the temperature gradient using the sound velocity profile, and obtain the temperature and sound velocity mapping relationship through the linear regression method; collect the water flow velocity data of the sensor array, calculate the Doppler frequency shift amount at the carrier frequency according to the water flow velocity data, and correct the Doppler frequency shift amount using a sliding mean window to obtain the water flow disturbance signal; obtain the water quality turbidity data of the sensor array, perform noise reduction processing on the water quality turbidity data through a Gaussian filter, and establish a water quality turbidity and bandwidth utilization rate index set based on the data after the noise reduction processing; perform normalization processing on the signal attenuation curve, the sound velocity mapping relationship, the water flow disturbance signal, and the bandwidth utilization rate index set to obtain the comprehensive evaluation data of the network performance.

[0031] Specifically, according to the sensor array deployed at 20-meter intervals within the water depth range of 0 to 200 meters, water depth data is collected and the acoustic signals within the range of 0 - 50 kHz are spectrally processed. The acoustic wave propagation path loss is calculated through the cylindrical expansion model, and the curve of water depth and signal attenuation is established. Then, the sound velocity profile is drawn at 10-meter depth intervals. For the water temperature change range of 15 to 25 degrees Celsius, the sound velocity value of each layer is calculated using the temperature gradient, and the mapping relationship between temperature and sound velocity is established using the linear regression algorithm. The change value of the transmission delay is recorded at a sound velocity change unit of 1 m / s, and the sound velocity profile data is fused with the water depth acoustic signal attenuation curve. By collecting the water flow velocity data from 0.5 m / s to 2.5 m / s, the Doppler frequency shift amount at a carrier frequency of 15 kHz is calculated. The 11-point sliding mean window is used to correct the water flow disturbance signal, and the adaptive filter is combined to compensate for the signal distortion caused by the water flow. The associated data of the transmission error rate and signal distortion is statistically analyzed. According to the data of 5 to 35 turbidity units measured by the water quality sensor, the Gaussian filter is used for noise reduction processing, and the change value of the signal-to-noise ratio before and after is compared. The water quality turbidity and bandwidth utilization index set is established with 10 dB as the reference unit, and the water depth, water temperature, water flow, and water quality are normalized to form a comprehensive evaluation database of the networking performance. In the 200 m × 200 m experimental water area, a water depth sensor array is deployed. One sensor node is set every 20 meters in the vertical direction, with a total of 11 layers, and one node is deployed every 20 meters in the horizontal direction, with a total of 100 nodes. The spectral data of the acoustic signals from 0 to 50 kHz is collected. Through the cylindrical expansion calculation, the acoustic wave propagation loss is 15 dB per octave distance. The sound velocity value is recorded every 10 meters within the range of 0 - 200 meters. The sound velocity of the surface layer from 0 to 50 meters is 1530 m / s, the layer from 50 to 100 meters is 1520 m / s, the layer from 100 to 150 meters is 1510 m / s, and the layer from 150 to 200 meters is 1500 m / s. The temperature sensor collects data points every 0.5 degrees Celsius within the range of 15 - 25 degrees Celsius. The temperature-sound velocity relationship curve is fitted using linear regression, with a slope of 4 m / s per degree Celsius. The sound velocity at 25 degrees Celsius is measured to be 1540 m / s, and at 15 degrees Celsius is 1500 m / s. The transmission delay increases by 0.1 ms for every 1 m / s decrease in the sound velocity. In the water flow velocity range of 0.5 - 2.5 m / s, the Doppler frequency shift amount is recorded at an interval of 0.1 m / s. The maximum frequency shift measured at a carrier frequency of 15 kHz is 20 Hz. After processing with the 11-point sliding mean window, the signal distortion degree is reduced by 50%. The adaptive filter compensation reduces the error rate from 5% to 1%. For the water quality sensor within the range of 5 - 35 turbidity units, after Gaussian filtering with a standard deviation of 1.5, the signal-to-noise ratio is increased by 8 dB, and the bandwidth utilization rate is increased from 45% to 75%. After normalization, the water depth index within the range of 0 - 1 is 0.8, the temperature index is 0.6, the water flow index is 0.5, and the water quality index is 0.7.

[0032] Step S105: Through the distributed observation of underwater environmental data, the evaluation results of the networking performance of the robot under different underwater environmental data conditions are obtained, and the networking performance of the underwater robot is evaluated through the evaluation results.

[0033] Obtain the networking communication transmission rate data according to the environmental data collection points, and use the weighted average method to calculate the transmission rate data in groups to obtain the networking communication rate score table; for the transmission rate levels divided in the networking communication rate score table, obtain the communication data packets collected within the preset water temperature range, and determine the information packet loss rate value by counting the communication data packets; according to the link stability levels divided by the information packet loss rate value, record the transmission delay of the data packets within the preset depth range, and perform grouped statistics on the transmission delay according to the node spacing to obtain the communication success rate; obtain the turbidity data measured by the water quality sensor from within the preset depth range, record the channel bandwidth occupancy rate for the turbidity data, and statistically obtain the network node distribution data according to the preset signal strength levels; perform normalization processing on the network node distribution data, and use the preset weight ratio to calculate the weighted values of the transmission rate, link quality, and the network node distribution data to obtain the comprehensive networking performance score.

[0034] Specifically, according to the distributed layout of environmental data collection points, the networking communication transmission rate is recorded at a sampling frequency of 1 Hz for each node. The data is grouped according to a flow velocity interval of 0.5 m / s, and the transmission rate scores of each group are calculated through weighted average. It is divided into five levels: below 20 kbps, 20 - 40 kbps, 40 - 60 kbps, 60 - 80 kbps, and above 80 kbps. A networking communication rate scoring table is established and the link stability index is calculated. For the communication data collected within the water temperature range of 15 to 25 degrees Celsius, 100 data packets are statistically analyzed within a 60 - second evaluation period, the information packet loss rate value is calculated and divided into four levels: stable, relatively stable, less stable, and unstable. The channel quality thresholds are set at four levels: 95%, 85%, 75%, and 65%. By comparing the link stability indexes under different temperature conditions, the mapping relationship data between water temperature and communication performance is generated. By recording the data transmission delay within the water depth range of 0 to 200 meters, the communication distance points are grouped into three levels: short - range 50 meters, medium - range 100 meters, and long - range 200 meters according to a node spacing of 20 meters. The communication success rate and channel occupancy rate of each group of nodes within 100 data packet cycles are statistically analyzed, and the networking performance quantization data for each depth interval is generated in combination with the link stability index. According to the data of 5 to 35 turbidity units measured by the water quality sensor, the percentage of channel bandwidth occupancy is recorded. The node distribution within the network coverage is statistically analyzed according to six levels of signal strength: - 95 dB, - 90 dB, - 85 dB, - 80 dB, - 75 dB, - 70 dB. The underwater environmental parameters and networking indexes are normalized within a score range of 0 - 100, and a comprehensive score is generated according to the weight ratio of transmission rate 0.4, link quality 0.3, and coverage range 0.3. 25 communication nodes are arranged in a 500 m × 500 m observation area, and networking data is collected at a frequency of 1 Hz. When the measured water flow velocity is 0.5 m / s, the transmission rate is 85 kbps; when it is 1.0 m / s, it drops to 65 kbps; when it is 1.5 m / s, it drops to 45 kbps; when it is 2.0 m / s, it drops to 25 kbps; when it is 2.5 m / s, it drops to 15 kbps. Through weighted average calculation, the scores of each rate level are 95 points, 85 points, 75 points, 65 points, and 55 points respectively. Within the water temperature range of 15 to 25 degrees Celsius, the transmission situation of 100 data packets is statistically analyzed every 60 seconds. When the water temperature is 15 degrees Celsius, the link stability is 96%; when it is 20 degrees Celsius, it is 89%; when it is 25 degrees Celsius, it is 72%, corresponding to three levels: stable, relatively stable, and less stable. Within the water depth range of 0 - 200 meters, the short - range 50 - meter communication delay is 20 milliseconds and the success rate is 98%; the medium - range 100 - meter delay is 45 milliseconds and the success rate is 92%; the long - range 200 - meter delay is 85 milliseconds and the success rate is 85%. The channel occupancy rates are 45%, 65%, and 85% respectively.In the turbidity range of 5 to 35 units, the node quantity distribution of 6 signal intensity levels from -95 dB to -70 dB was recorded. After normalization, the transmission rate index was 85 points, the link quality index was 78 points, and the coverage range index was 82 points. The comprehensive score was 82 points calculated according to the weights of 0.4, 0.3, and 0.3.

[0035] Step S106: Based on the evaluation results of the robot networking performance under different water depth conditions, establish communication protocols for different water depths, adjust the transmission power and receiving sensitivity of the robot according to the attenuation characteristics of the signal propagation in water, and layer the robots according to the water depth, and design different network topologies respectively.

[0036] Divide the communication space into four depth layers according to the measurement data of the water depth sensor. The depth layers successively adopt increasing transmission power values from shallow to deep to obtain a depth layer power configuration scheme. For the depth layer power configuration scheme, measure the signal transmission path loss value and background noise of each layer, and set the corresponding receiving threshold according to the signal-to-noise ratio to obtain a receiving sensitivity adjustment strategy. Use the receiving sensitivity adjustment strategy to lay out the communication units in layers. The communication units are deployed in each depth layer according to the preset node density, and form a point-to-point within-layer, star-shaped between-layers, and tree-shaped cross-layer network topology structure by dividing three communication radii of near, medium, and far. Establish a primary and backup dual-path mechanism for the network topology structure. If the primary path communication is detected to be interrupted, switch to the backup path to transmit data. Monitor the link quality and node energy status according to the primary and backup dual-path mechanism. If the link bandwidth utilization rate exceeds the preset threshold, switch among the three communication protocols of time division multiplexing, carrier sense, and fixed time slots, and ensure data transmission by adjusting the channel bandwidth.

[0037] Specifically, according to the acoustic signal attenuation data measured by the water depth sensor within the range of 0 to 200 meters, the underwater communication levels are divided at 50-meter intervals. Exponentially increasing transmission power levels are set for each layer. The transmission power for the 0 - 50-meter layer is 180 dB, for the 50 - 100-meter layer is 186 dB, for the 100 - 150-meter layer is 194 dB, and for the 150 - 200-meter layer is 204 dB. The minimum reception threshold is set according to an 85% node communication success rate. For the measured results of the signal transmission path loss value and background noise for each layer, the reception sensitivity is dynamically adjusted. The reception threshold is set to -80 dB at a signal-to-noise ratio of 20 dB, -85 dB at 15 dB, -90 dB at 10 dB, -95 dB at 5 dB, and -100 dB at 0 dB. The power compensation value of the adjacent layer signal is calculated using distance-based hierarchical routing. Through the network data after water depth stratification, communication units are deployed at a node density of 25 per square kilometer for each layer. Three communication radii of 50 meters for short distance, 100 meters for medium distance, and 200 meters for long distance are defined. The network is formed according to three topological structures: point-to-point within the layer, star between layers, and tree across layers. A primary and backup dual-path is set to ensure data forwarding between layers. According to the communication link quality and node energy status at different water depths, three communication protocols of time-division multiplexing, carrier sense, and fixed time slots are generated. The channel resource priority is divided according to the percentage of the remaining energy of the node. When the link bandwidth utilization rate exceeds 90%, the protocol is switched, and the channel bandwidth is adaptively adjusted within the range of 2 - 10 kbps. In a 1000 m × 1000 m experimental sea area, a four-layer network is divided at 50-meter water depth intervals. Communication units are deployed at a node density of 25 per square kilometer for each layer. The amplitude of the acoustic signal is recorded at a sampling rate of 1 Hz. It is measured that the attenuation in the 0 - 50-meter layer is 0.6 dB per meter and the transmission power is set to 180 dB, in the 50 - 100-meter layer the attenuation is 0.8 dB per meter and the transmission power is set to 186 dB, in the 100 - 150-meter layer the attenuation is 1.1 dB per meter and the transmission power is set to 194 dB, and in the 150 - 200-meter layer the attenuation is 1.5 dB per meter and the transmission power is set to 204 dB. The measured node communication success rate reaches 88%. The measured value of the background noise is -110 dB. The reception threshold is divided according to the signal-to-noise ratio. The threshold is -80 dB and the success rate is 98% at a signal-to-noise ratio of 20 dB, -85 dB and the success rate is 95% at 15 dB, -90 dB and the success rate is 92% at 10 dB, -95 dB and the success rate is 87% at 5 dB, and -100 dB and the success rate is 82% at 0 dB. In the communication topology configuration, the point-to-point protocol with a bandwidth of 10 kbps is used within the short distance of 50 meters, the star topology with a bandwidth of 6 kbps is used within the medium distance of 100 meters, and the tree structure with a bandwidth of 2 kbps is used within the long distance of 200 meters.The node with more than 80% remaining energy has the highest priority, followed by 50%-80%, and the lowest is 20%-50%. When the bandwidth utilization rate exceeds 90%, the communication frequency is automatically reduced, and the measured average network delay is 120 milliseconds.

[0038] Step S107: Based on the evaluation results of the robot networking performance under different water temperature conditions, establish the mapping relationship between temperature and networking performance, adjust the robot working parameters, including battery management and sensor sensitivity, to adapt to different temperature conditions, and construct a water temperature distribution map to form a robot networking layout strategy.

[0039] Obtain the temperature data collected by the water temperature sensor, and obtain the temperature distribution heat map through the distance weighted average method; determine the signal transmission power compensation value according to the temperature distribution heat map, and the signal transmission power compensation value is obtained by multiplying the transmission power at the reference temperature by the temperature change amount; set the battery discharge voltage threshold for the temperature data, and the battery discharge voltage threshold corresponds to the remaining battery power to set three levels of power consumption control states; adjust the sensor sampling parameters according to the temperature data, and the sensor sampling parameters include the sampling frequency value and the gain coefficient, and calibrate the temperature data through the sampling error; divide the network coverage area based on the temperature distribution heat map, and the network coverage area triggers the topology structure update according to the communication error rate and the remaining power between nodes, and updates the routing table according to the preset time.

[0040] Specifically, according to the temperature data collected by the water temperature sensor in the range of 15 to 25 degrees Celsius, a temperature distribution heat map is constructed by the distance weighted average method. The temperature is graded at 2-degree intervals, and the signal transmission intensity is adaptively compensated. The transmission power is 180 dB at the reference temperature of 15 degrees Celsius, and the compensation value increases by 1 dB for every 1-degree Celsius increase in temperature. The signal error rate value is recorded in real time. Regarding the influence of temperature change on the battery discharge characteristics, the battery discharge voltage is set to 3.6 volts at 15 degrees Celsius, 3.3 volts at 20 degrees Celsius, and 3.0 volts at 25 degrees Celsius. Three-level power consumption control is carried out according to the remaining battery power. When the remaining power is above 80%, it works normally; when it is between 50% and 80%, the transmission power is reduced; when it is below 30%, it enters the sleep charging protection state. By recording the data of the influence of temperature on the sensor sensitivity, the sensor sampling frequency and gain parameters are adjusted. The sampling frequency is 1.0 Hz and the gain is 1.1 times at 15 degrees Celsius, the sampling frequency is 1.5 Hz and the gain is 1.0 times at 20 degrees Celsius, and the sampling frequency is 2.0 Hz and the gain is 0.9 times at 25 degrees Celsius. And the temperature value is calibrated based on the sampling error. According to the signal transmission quality and the remaining energy of the node corresponding to the water temperature distribution map, the hierarchical routing method is used to divide the network coverage area. The node spacing is set to 40 meters in the temperature range of 15 to 20 degrees Celsius and 30 meters in the range of 20 to 25 degrees Celsius. When the node communication error rate exceeds 5% or the remaining power is lower than 50%, the topology structure is triggered to be updated, and the routing table is updated every 300 seconds. 100 water temperature sensor nodes are deployed in a 500 m × 500 m water area. The distance weighted average method is used to interpolate the temperature data in the range of 15 - 25 degrees Celsius. The weight coefficient decays by 20% for every 10-meter increase in distance, and a distribution heat map of 5 temperature intervals is constructed. The transmission power is 180 dB and the error rate is 2.1% in the 15 - 17 degrees Celsius area, the transmission power is 182 dB and the error rate is 2.5% in the 17 - 19 degrees Celsius area, the transmission power is 184 dB and the error rate is 2.8% in the 19 - 21 degrees Celsius area, the transmission power is 186 dB and the error rate is 3.2% in the 21 - 23 degrees Celsius area, and the transmission power is 188 dB and the error rate is 3.6% in the 23 - 25 degrees Celsius area. The battery management adopts a three-level control strategy. It is measured that the remaining power is 90% with a voltage of 3.6 volts and a current of 200 mA at 15 degrees Celsius, the remaining power is 65% with a voltage of 3.3 volts and a current of 180 mA at 20 degrees Celsius, and the remaining power is 40% with a voltage of 3.0 volts and a current of 150 mA at 25 degrees Celsius. The sensor performance is adjusted with temperature change. The sampling frequency is 1.0 Hz, the signal gain is 1.1 times, and the measurement error is 0.2 degrees Celsius in the 15 degrees Celsius area, the sampling frequency is 1.5 Hz, the signal gain is 1.0 times, and the measurement error is 0.3 degrees Celsius in the 20 degrees Celsius area, and the sampling frequency is 2.0 Hz, the signal gain is 0.9 times, and the measurement error is 0.4 degrees Celsius in the 25 degrees Celsius area.The network topology adopts a hierarchical routing structure. 45 nodes are arranged at a spacing of 40 meters in the area of 15 - 20 degrees Celsius, and 55 nodes are arranged at a spacing of 30 meters in the area of 20 - 25 degrees Celsius. The routing table is updated every 300 seconds. The measured average network delay is 85 milliseconds, and the communication success rate is 94%.

[0041] Step S108, based on the evaluation results of the robot networking performance under different flow velocity conditions, establish a relationship model between the flow velocity, communication delay, and data packet loss rate, optimize the network topology and data transmission strategy of the robot networking, and at the same time divide the transmission task priorities to achieve flow velocity-driven task scheduling.

[0042] Obtain the flow velocity data collected by the water flow velocity sensor, perform K-means clustering analysis on three performance indicators of communication throughput, delay, and jitter according to the flow velocity data to obtain the mapping relationship between network performance and flow velocity; calculate the number of nodes and the flow velocity change rate according to the mapping relationship between network performance and flow velocity, and use the number of nodes and the flow velocity change rate to determine the network topology structure, and the network topology structure includes a full connection topology, a star topology, and a chain topology; allocate bandwidth resources for the network topology structure, set the transmission priority according to the bandwidth resources, and the transmission priority includes the highest priority, the second highest priority, the medium priority, the second lowest priority, and the lowest priority; perform priority queue sorting on the transmission tasks according to the transmission priority and the flow velocity change trend, and determine the transmission order through the priority queue sorting; if the flow velocity change rate exceeds the preset threshold, trigger the priority preemption mechanism, and use the priority preemption mechanism to store the preempted tasks in the cache queue, and perform data packet retransmission through the cache queue.

[0043] Specifically, according to the flow velocity data collected by the water flow velocity sensor in the range of 0.5 m / s to 2.5 m / s, the K-means clustering is used to analyze three performance indicators: communication throughput, latency, and jitter. The bit error rate and channel utilization rate are calibrated under five flow velocity gears at intervals of 0.5 m / s. The link state data is collected every 10 seconds to generate the mapping relationship between flow velocity and network performance. By recording the link bandwidth occupancy data under different flow velocity conditions, the network topology is optimized according to the number of nodes and the flow velocity change rate. When there are 10 nodes and the flow velocity change rate is less than 20%, the full connection topology is adopted; when there are 20 nodes and the change rate is 20%-50%, the star topology is adopted; when there are 30 nodes and the change rate exceeds 50%, the chain topology is adopted. The data packet length is limited within 1024 bytes. For the transmission delay change caused by water flow disturbance, the tasks are divided into five priorities. The highest priority occupies 50% of the bandwidth with a transmission cycle of 25 milliseconds, the second highest priority occupies 30% with a cycle of 50 milliseconds, the medium priority occupies 10% with a cycle of 100 milliseconds, the second lowest priority occupies 7% with a cycle of 150 milliseconds, and the lowest priority occupies 3% with a cycle of 200 milliseconds. The bandwidth allocation ratio floats within the range of plus or minus 10% according to the link quality. The transmission tasks are reordered according to the flow velocity change trend, and the priority queue is used to arrange the transmission order. When the flow velocity change rate exceeds 50%, the priority preemption mechanism is triggered, and the preempted tasks are incorporated into the cache queue with a size of 100 data packets. A retransmission timeout of 500 milliseconds is set, and the basic scheduling cycle is adjusted within the range of 25 milliseconds to 100 milliseconds according to the network scale. 30 communication nodes are deployed in the 500 m × 500 m observation area, and the K-means clustering is used to analyze the communication performance within the flow velocity range of 0.5 - 2.5 m / s. The measured throughput is 3.5 Mbps, latency is 25 milliseconds, and jitter is 5 milliseconds at 0.5 m / s; the throughput is 3.0 Mbps, latency is 35 milliseconds, and jitter is 8 milliseconds at 1.0 m / s; the throughput is 2.5 Mbps, latency is 45 milliseconds, and jitter is 12 milliseconds at 1.5 m / s; the throughput is 2.0 Mbps, latency is 60 milliseconds, and jitter is 15 milliseconds at 2.0 m / s; the throughput is 1.5 Mbps, latency is 80 milliseconds, and jitter is 20 milliseconds at 2.5 m / s. The network topology is dynamically adjusted according to the flow velocity change. The full connection topology is adopted at 0.5 m / s with a bit error rate of 0.5%; it switches to the star topology at 1.5 m / s with a bit error rate of 1.2%; it switches to the chain topology at 2.5 m / s with a bit error rate of 2.5%. The task priorities are divided into five levels. The real-time control data accounts for 50% of the bandwidth with a limit of 512 bytes, the sensing data accounts for 30% with a limit of 768 bytes, the status data accounts for 10% with a limit of 1024 bytes, the log data accounts for 7%, and the alarm data accounts for 3%. When the flow velocity suddenly changes from 0.5 m / s to 2.5 m / s, the priority preemption is triggered, and 85 data packets are accumulated in the cache queue. The retransmission is completed after 450 milliseconds. The measured average network scheduling cycle is 75 milliseconds, and the communication success rate is 92%.

[0044] Step S109: Based on the evaluation results of the robot networking performance under different water quality conditions, construct a water quality-driven task scheduling strategy according to the changes in water quality. Divide the water quality into regions, dynamically adjust the task processing priorities, and preferentially process the data collection in water quality sensitive regions.

[0045] Obtain the turbidity data collected by the sensor, and obtain a gridded water quality distribution map through linear interpolation operations. The water quality distribution map divides the monitoring area and sampling period according to the turbidity unit interval; determine the data transmission priority level according to the monitoring area division result and the signal strength value, and the priority level corresponds to the divided data packet size and bandwidth occupancy ratio; use a moving average window to verify the validity of the sampling data in the monitoring area, and adjust the sampling period according to the turbidity data change rate value; for the computing tasks in the monitoring area, use a hierarchical queue for scheduling and distribution, and the hierarchical queue allocates the computing resource occupancy ratio according to the turbidity interval; determine whether the number of tasks in the hierarchical queue exceeds a preset threshold. If the number of tasks exceeds the preset threshold, trigger a dynamic increase in the priority and adjust the computing resource allocation ratio.

[0046] Specifically, according to the turbidity data collected by the water quality sensor, establish a 20-meter

[0047] ×20 m grid water quality distribution map, the monitoring area is divided at intervals of 5 turbidity units. The sampling period is set to 30 seconds in the area with turbidity of 5 to 10 units, 20 seconds in the area with turbidity of 10 to 15 units, and 10 seconds in the area with turbidity above 15 units. The sensor range is dynamically calibrated within the range of 0 to 50 turbidity units. For the data collection tasks in different water quality areas, five priorities are divided according to the signal strength from -75 dB to -95 dB. The data packet limit in the area with turbidity above 15 units is 512 bytes, occupying 45% of the bandwidth. The data packet limit in the area with turbidity of 10 to 15 units is 768 bytes, occupying 35% of the bandwidth. The data packet limit in the area with turbidity of 5 to 10 units is 1024 bytes, occupying 20% of the bandwidth. The task timeout is set to 180 seconds. By recording the water quality change trend, the 60-second sliding average window is used to verify the validity of the sampling data. When the turbidity change rate exceeds 1 unit per minute, the sampling frequency of the nodes in the area is increased to a 5-second interval, and the data transmission delay is limited within 50 milliseconds, and the transmission bandwidth ratio is dynamically adjusted according to the data packet integrity. According to the distribution of water quality sensitive areas, a hierarchical queue with a depth of 200 tasks is used for scheduling and distribution. 45% of the computing resources are invested in the area with turbidity above 15 units, 35% in the area with turbidity of 10 to 15 units, and 20% in the area with turbidity of 5 to 10 units. The cyclic scheduling is carried out according to the 25-millisecond basic cycle. When the number of tasks in the queue exceeds 160, the priority dynamic promotion mechanism is triggered. In the 500 m × 500 m monitoring water area, 625 grid cells of 20 m × 20 m are divided, and a water quality distribution map is constructed by linear interpolation. The measured turbidity in Area 1 is 18 units, the sampling period is 10 seconds, the data packet is 512 bytes, and the signal strength is -75 dB. The turbidity in Area 2 is 12 units, the sampling period is 20 seconds, the data packet is 768 bytes, and the signal strength is -85 dB. The turbidity in Area 3 is 7 units, the sampling period is 30 seconds, the data packet is 1024 bytes, and the signal strength is -95 dB. There are 185 pending tasks stored in the task priority queue. Area 1 is allocated 45% of the bandwidth to process 85 tasks with a delay of 28 milliseconds. Area 2 is allocated 35% of the bandwidth to process 65 tasks with a delay of 42 milliseconds. Area 3 is allocated 20% of the bandwidth to process 35 tasks with a delay of 68 milliseconds. Through the 60-second sliding window, it is monitored that the turbidity in Area 1 rises from 18 units to 19.2 units, and the change rate reaches 1.2 units per minute. The sampling frequency is automatically increased to a 5-second interval, and the occupied bandwidth is dynamically adjusted to 52%. In the hierarchical task queue, 45% of the computing resources in Area 1 are invested to process high-priority tasks with an average processing time of 22 milliseconds. 35% of the resources in Area 2 are invested to process medium-priority tasks with an average processing time of 35 milliseconds. 20% of the resources in Area 3 are invested to process low-priority tasks with an average processing time of 58 milliseconds. The task completion rate reaches 96%, and the data validity verification passing rate is 98%.

[0048] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A distributed observation method for underwater robot network evaluation, characterized in that The method includes: Deploy a number of underwater observation nodes in the target observation water area, and collect underwater environment data through sensors set on the observation nodes, including water depth, water temperature, water flow velocity, and water quality; Preprocess the underwater environment data, and unify data from different sources and different scales to the same preset scale to obtain the preprocessed underwater environment data; Adopt a preset underwater networking protocol to transmit the preprocessed underwater environment data from each distributed observation node to the central node, conduct an analysis on the impact of the preprocessed underwater environment data on the networking performance, and respectively analyze the influence laws of water depth, water temperature, water flow velocity, and water quality on the networking performance of the underwater robot; Based on the underwater environment data, test the information transmission rate and information loss rate of the robot networking under different water depth, water temperature, water flow velocity, and water quality conditions, and evaluate the networking performance of the robot under different conditions according to the test results; Through the distributed observation of the underwater environment data, obtain the evaluation results of the robot networking performance under different underwater environment data conditions, and evaluate the networking performance of the underwater robot through the evaluation results; Based on the evaluation results of the robot networking performance under different water depth conditions, establish communication protocols for different water depths, adjust the transmission power and receiving sensitivity of the robot according to the attenuation characteristics of the signal propagation in water, and layer the robots according to the water depth, and respectively design different network topologies; Based on the evaluation results of the robot networking performance under different water temperature conditions, establish the mapping relationship between temperature and networking performance, adjust the working parameters of the robot, including battery management and sensor sensitivity, to adapt to different temperature conditions, and construct a water temperature distribution map to form a robot networking layout strategy; Based on the evaluation results of the robot networking performance under different flow velocity conditions, establish a relationship model between flow velocity, communication delay, and data packet loss rate, optimize the network topology and data transmission strategy of the robot networking, and at the same time divide the transmission task priorities to achieve flow velocity-driven task scheduling; Based on the evaluation results of the impact of different water quality conditions on the robot networking performance, construct a water quality-driven task scheduling strategy according to the change of water quality, divide the water quality into regions, dynamically adjust the task processing priorities, and give priority to processing data collection in water quality sensitive areas.

2. The method according to claim 1, wherein The step of deploying a number of underwater observation nodes in the target observation water area and collecting underwater environment data through sensors set on the observation nodes, including water depth, water temperature, water flow velocity, and water quality, includes: Use distributed observation nodes to obtain water temperature data at the surface layer, middle layer, and deep layer. When the temperature difference between two adjacent data collections exceeds the preset threshold, the observation node adjusts from the first sampling interval to the second sampling interval; For the water flow data collected by the observation node, obtain the three-dimensional vector components through an acoustic Doppler velocimeter, obtain the actual flow velocity value according to the included angle between the vector component and the observation node, and calibrate the actual flow velocity value according to the preset flow velocity interval; Receive the dissolved oxygen data, turbidity data, and conductivity data collected by the observation node, store the data in different levels according to the preset parameter interval. If the dissolved oxygen is lower than the first threshold and the conductivity is higher than the second threshold, trigger the observation node to adjust to the second sampling interval; Based on the water depth data obtained by the observation nodes, an underwater topographic map with isobath intervals is obtained through triangular grid depth fitting, and the depth in the observation blind area of the underwater topographic map is supplemented. For the water temperature data, the actual flow velocity value, the dissolved oxygen data, the turbidity data, and the conductivity data, a multi-layer hydrological parameter distribution map is constructed corresponding to the underwater topographic map according to the sampling depth.

3. The method according to claim 1, wherein The preprocessing of the underwater environment data and the unification of data from different sources and different scales to the same preset scale to obtain the preprocessed underwater environment data include: According to the temperature gradient value obtained by the water temperature sensor at adjacent sampling points, wavelet basis functions are used to process the data outside the preset temperature threshold range for outliers, and the corrected temperature data is obtained through the three - standard - deviation principle. A sampling time reference is established for the corrected temperature data, and a Butterworth low - pass filter is used to process the water flow sensor data, and the water flow direction angle value is obtained according to the north - facing positive reference coordinate system. Dissolved oxygen, turbidity, and conductivity data are obtained using a water quality sensor, anomaly judgment is performed through preset water quality parameter thresholds, and spline interpolation is performed on the water quality data according to the sampling time of the corrected temperature data. For the water depth sensor, adaptive window median filtering is used to obtain depth sampling data. A unified depth reference is established based on the depth sampling data, and a linear interpolation algorithm is used to align the corrected temperature data, the water flow direction angle value, and the water quality data in depth to obtain multi - parameter correlation data at unified depth sampling points.

4. The method according to claim 1, wherein The preset underwater networking protocol is used to transmit the preprocessed underwater environment data from each distributed observation node to the central node, and an analysis of the impact of the preprocessed underwater environment data on the networking performance is carried out. The influence rules of water depth, water temperature, water flow velocity, and water quality on the underwater robot networking performance are analyzed respectively, including: Obtain the water temperature data, water depth data, and water quality data collected by the underwater acoustic communication link. The communication network levels are divided according to the quantization interval of the depth sensor for the water depth data. Relay nodes are set in the network levels to obtain a data forwarding path, and a mapping relationship between the water depth gradient and the communication delay is established according to the data forwarding path. The sound velocity values of each water layer are calculated using the water temperature data and the salinity data, a decision tree model is constructed according to the sound velocity values, and the communication link quality evaluation result is obtained through the decision tree model. The flow velocity data is processed by the greedy routing algorithm to obtain the minimum - hop communication path. If communication error codes caused by water flow disturbances are detected, forward error correction coding is used to compensate for the communication error codes. Based on the compensation by the forward error correction coding, throughput data, packet loss rate data, and end - to - end delay data are obtained. The analytic hierarchy process is used to process the throughput data, packet loss rate data, end - to - end delay data, and water depth data, water temperature data, and water quality data to obtain the weight values of the performance evaluation indicators.

5. The method according to claim 1, characterized in that, Based on the underwater environment data, the information transmission rate and information loss rate of the robot networking under different water depths, water temperatures, water flow velocities, and water quality conditions are tested, and the robot networking performance under different conditions is evaluated according to the test results, including: Collect the acoustic signals of the sensor array, perform spectral processing on the acoustic signals to obtain the acoustic wave propagation path loss data, and establish a water depth and signal attenuation curve according to the path loss data; Obtain the sound velocity profile according to the water depth and signal attenuation curve, calculate the sound velocity values of each layer for the temperature gradient by using the sound velocity profile, and obtain the temperature and sound velocity mapping relationship through the linear regression method; Collect the water flow velocity data of the sensor array, calculate the Doppler frequency shift amount at the carrier frequency according to the water flow velocity data, and correct the Doppler frequency shift amount by using a sliding mean window to obtain the water flow disturbance signal; Obtain the water quality turbidity data of the sensor array, perform noise reduction processing on the water quality turbidity data through a Gaussian filter, and establish a water quality turbidity and bandwidth utilization rate index set according to the data after the noise reduction processing; Perform normalization processing on the signal attenuation curve, the sound velocity mapping relationship, the water flow disturbance signal, and the bandwidth utilization rate index set to obtain the comprehensive evaluation data of the networking performance.

6. The method according to claim 1, characterized in that, The evaluation result of the robot networking performance under different underwater environment data conditions is obtained through the distributed observation of the underwater environment data, and the networking performance of the underwater robot is evaluated through the evaluation result, including: Obtain the networking communication transmission rate data according to the environmental data collection points, and perform grouped calculation on the transmission rate data by using the weighted average method to obtain the networking communication rate score table; For the transmission rate levels divided by the networking communication rate score table, obtain the communication data packets collected within the preset water temperature range, and determine the information packet loss rate value by counting the communication data packets; According to the link stability level divided by the information packet loss rate value, record the transmission delay of the data packets within the preset depth range, and perform grouped statistics on the transmission delay according to the node spacing to obtain the communication success rate; Obtain the turbidity data measured by the water quality sensor from within the preset depth range, record the channel bandwidth occupancy rate for the turbidity data, and statistically obtain the network node distribution data according to the preset signal strength levels; Perform normalization processing on the network node distribution data, calculate the weighted values of the transmission rate, the link quality, and the network node distribution data by using the preset weight ratio to obtain the comprehensive networking performance score.

7. The method according to claim 1, characterized in that Based on the evaluation results of the robot networking performance under different water depth conditions, establish communication protocols for different water depths, adjust the transmission power and receiving sensitivity of the robot according to the attenuation characteristics of the signal propagation in water, and layer the robots according to the water depth, and design different network topologies respectively, including: Divide the communication space into four depth layers according to the water depth sensor measurement data, and sequentially adopt increasing transmission power values from shallow to deep in the depth layers to obtain the depth layer power configuration scheme; For the depth layer power configuration scheme, measure the signal transmission path loss value and the background noise of each layer, and set the corresponding receiving threshold according to the signal-to-noise ratio to obtain the receiving sensitivity adjustment strategy; The communication units are hierarchically arranged by adopting the receiving sensitivity adjustment strategy. The communication units are deployed in each depth layer according to a preset node density, and a point-to-point network topology within the layer, a star network topology between layers, and a cross-layer tree network topology are formed by dividing three communication radii of near, medium, and far. A primary and backup dual-path mechanism is established for the network topology. If a communication interruption is detected in the primary path, the data is switched to the backup path for transmission. Based on the primary and backup dual-path mechanism, the link quality and node energy status are monitored. If the link bandwidth utilization rate exceeds a preset threshold, a switch is made among three communication protocols of time-division multiplexing, carrier sense, and fixed time slots, and the data transmission is ensured by adjusting the channel bandwidth.

8. The method according to claim 1, characterized in that, Based on the evaluation results of the robot networking performance under different water temperature conditions, a mapping relationship between the temperature and the networking performance is established, and the working parameters of the robot are adjusted, including battery management and sensor sensitivity, to adapt to different temperature conditions, and a water temperature distribution map is constructed to form a robot networking layout strategy, including: Obtain the temperature data collected by the water temperature sensor, and the temperature distribution heat map is obtained by the distance weighted average method for the temperature data. Determine the signal transmission power compensation value according to the temperature distribution heat map, and the signal transmission power compensation value is obtained by multiplying the transmission power at the reference temperature by the temperature change amount. Set the battery discharge voltage threshold for the temperature data, and three-level power consumption control states are set corresponding to the remaining battery power for the battery discharge voltage threshold. Adjust the sensor sampling parameters according to the temperature data. The sensor sampling parameters include the sampling frequency value and the gain coefficient, and the temperature data is calibrated by the sampling error. Divide the network coverage area based on the temperature distribution heat map. The network coverage area triggers the topology update according to the communication error rate and the remaining power between nodes, and the routing table is updated according to a preset time.

9. The method according to claim 1, characterized in that, Based on the evaluation results of the robot networking performance under different flow velocity conditions, a relationship model between the flow velocity, the communication delay, and the data packet loss rate is established, the network topology and data transmission strategy of the robot networking are optimized, and the transmission task priorities are divided to achieve the flow velocity-driven task scheduling, including: Obtain the flow velocity data collected by the water flow velocity sensor, and perform K-means clustering analysis on three performance indicators of communication throughput, delay, and jitter according to the flow velocity data to obtain the mapping relationship between the network performance and the flow velocity. Calculate the number of nodes and the flow velocity change rate according to the mapping relationship between the network performance and the flow velocity, and determine the network topology structure by using the number of nodes and the flow velocity change rate. The network topology structure includes a fully connected topology, a star topology, and a chain topology. Allocate bandwidth resources for the network topology structure, and set the transmission priority according to the bandwidth resources. The transmission priority includes the highest priority, the second highest priority, the medium priority, the second lowest priority, and the lowest priority. Sort the transmission tasks in a priority queue according to the transmission priority and the flow velocity change trend, and determine the transmission order through the priority queue sorting. If the flow velocity change rate exceeds a preset threshold, a priority preemption mechanism is triggered, and the preempted task is stored in the cache queue by using the priority preemption mechanism, and the data packet is retransmitted through the cache queue.

10. The method according to claim 1, characterized in that, Based on the evaluation results of the robot networking performance under different water quality conditions, a water quality-driven task scheduling strategy is constructed according to the changes in water quality. The water quality is divided into regions, and the task processing priorities are dynamically adjusted to preferentially process data collection in water quality-sensitive areas, including: Obtain the turbidity data collected by the sensor, and obtain the grid water quality distribution map through linear interpolation operation. The water quality distribution map divides the monitoring area and the sampling period according to the turbidity unit interval; Determine the data transmission priority level according to the monitoring area division result and the signal strength value. The priority level corresponds to the division of the data packet size and the bandwidth occupancy ratio; Use a sliding average window to verify the validity of the sampling data in the monitoring area, and adjust the sampling period according to the turbidity data change rate value; For the computing tasks in the monitoring area, use a hierarchical queue for scheduling and distribution. The hierarchical queue allocates the computing resource occupancy ratio according to the turbidity interval; Judge whether the number of tasks in the hierarchical queue exceeds a preset threshold. If the number of tasks exceeds the preset threshold, trigger a dynamic increase in priority and adjust the computing resource allocation ratio.

Images