Method and system for in-situ detection of ocean water quality based on underwater robot

Abstract

The present application relates to the technical field of water quality detection, and particularly relates to a marine water quality in-situ detection method and system based on an underwater robot. The present application discards the wrong assumption that the water quality boundary is regarded as a static reference in the traditional method, and accurately deduces the coordinate of the water body drifting with the ocean current. Meanwhile, with the aid of the water quality characteristic data in the process of cutting into the water quality front twice, the observation visual angle difference is accurately restored. Furthermore, based on the physical objective law of the dispersion of the second integral of the internal accelerometer of the inertial navigation system, the navigation cumulative error vector is determined, and the coordinate of the underwater robot is corrected, so as to improve the navigation positioning accuracy of the underwater robot, and then accurately deduce the water quality distribution of the target sea area.

Description

Technical Field

[0001] This invention relates to the field of water quality testing technology, specifically to a method and system for in-situ testing of marine water quality based on an underwater robot. Background Technology

[0002] Marine nearshore aquaculture areas and estuarine mixed zones are ecologically complex and economically valuable. Accurately understanding their water quality distribution is crucial for marine ecological protection, resource development, and the sustainable development of the aquaculture industry. Currently, underwater robots equipped with multi-parameter water quality probes are used for underway measurements to obtain high-resolution data on salinity, turbidity, and other water quality distributions. However, these areas are generally characterized by high turbidity and high current velocities, forcing underwater robots to rely solely on their internal inertial navigation systems and Doppler velocimeters for autonomous dead reckoning.

[0003] During the calibration process of underwater navigation, it is assumed that the water feature boundaries are stationary and calibration is performed by comparing the coordinates of cross-course routes. However, the actual water quality gradient boundaries will objectively move with local ocean currents, and the difference in observation angles when entering from different directions will cause the waveform data collected by the sensors to be equivalently stretched or compressed in spatial distance. At the same time, the accelerometer inside the inertial navigation system is affected by white noise and constant bias interference. As the underwater navigation time increases, the calculated position coordinates will show a cumulative error of quadratic divergence. These factors will affect the accuracy of the underwater robot's navigation and positioning, and thus affect the accuracy of ocean water quality detection. Summary of the Invention

[0004] To address the low accuracy of existing in-situ marine water quality detection methods, this invention aims to provide a method and system for in-situ marine water quality detection based on an underwater robot. The specific technical solution adopted is as follows: A method for in-situ detection of ocean water quality based on an underwater robot, the method comprising: The underwater robot's course, sea area coordinates, bottom tracking speed, and the absolute velocity of seawater relative to the ground and water quality parameters were collected during its navigation in the target sea area. Based on the temporal variation characteristics of water quality parameters and bottom tracking speed, the first cut time of the underwater robot's first entry into the water quality front is determined, the first cut trajectory sequence of the underwater robot during the first entry into the water quality front is fitted, and the first cut water quality feature matrix is ​​fitted in combination with water quality parameters; the second cut time and second cut trajectory sequence of the underwater robot returning and re-entering the water quality front are determined, and the second cut water quality feature matrix is ​​fitted in combination with the difference in heading between the first cut time and the second cut time. Based on the initial and subsequent water quality feature matrices, the acquisition deviation of the underwater robot is obtained, and the corrected coordinates of the sea area at the subsequent cutting time are determined by combining the heading at the subsequent cutting time. At each time after the initial cutting time, the drift projection coordinates of the underwater robot are determined based on the sea area coordinates and the absolute velocity of seawater relative to the ground. Based on the difference between the drift-induced coordinates and the corrected coordinates at the re-cutting moment, the navigation cumulative error vector is determined; between the first cutting moment and the re-cutting moment, the drift-induced coordinates at each moment are corrected according to the preset time allocation weight and the navigation cumulative error vector at each moment, and the water quality distribution of the target sea area is inferred based on the corrected drift-induced coordinates and water quality parameter interpolation.

[0005] Furthermore, the method for the first cut time includes: When the underwater robot is navigating in the target sea area, the water quality derivative at each moment is determined based on the water quality change rate and the bottom tracking speed; the first moment when the water quality derivative is greater than a preset derivative threshold is taken as the first cut-off moment.

[0006] Furthermore, the method for obtaining the first cut track sequence and the first cut water quality feature matrix includes: A preset time window is determined with the first cut time as the center. The bottom tracking speed is integrated in the time domain to determine the cumulative distance of each moment relative to the first cut time within the preset time window. The cumulative distances are sorted to construct the first cut track sequence. The first cut trajectory sequence is resampled to obtain the first cut standard trajectory sequence; the water quality parameters within the preset time window of the first cut time are standardized and resampled to obtain the first cut standard water quality sequence; the first cut standard trajectory sequence and the first cut standard water quality sequence are respectively used as column vectors to construct the first cut water quality feature matrix.

[0007] Furthermore, the method for obtaining the recutting time includes: After the initial cut-off time, a preset time window is constructed with each time point as the center, the trajectory sequence is determined, and a water quality feature matrix is ​​constructed in combination with water quality parameters. Based on the similarity between the water quality feature matrix and the initial cut-off water quality feature matrix, and the deviation of the sea area coordinates at each time point from the sea area coordinates at the initial cut-off time, it is determined whether the sea area at each time point is the sea area that first cuts into the water quality front. The water quality derivative at each time point is calculated, and the first time point in which the sea area is the sea area that first cuts into the water quality front and the water quality derivative is greater than a preset derivative threshold is taken as the recut-off time point.

[0008] Furthermore, the method for obtaining the recut water quality feature matrix includes: The re-cut track sequence is resampled to obtain the re-cut standard track sequence; the water quality parameters within the preset time window of the re-cut time are standardized and resampled to obtain the re-cut standard water quality sequence; the heading angle between the heading at the first cut time and the heading at the re-cut time is determined. If the heading angle is less than or equal to a preset angle, the recut standard track sequence and the recut standard water quality sequence are used as column vectors to construct a recut water quality feature matrix; if the heading angle is greater than the preset angle, the mirrored recut standard track sequence and the recut standard water quality sequence are used as column vectors to construct a recut water quality feature matrix.

[0009] Furthermore, the method for obtaining the acquisition deviation includes: Variable scaling factors and translation parameters are used to transform the recut standard track sequence. Based on the column mapping relationship between the standard track sequence and the standard water quality sequence, the standard water quality corresponding to each standard track after each transformation in the recut standard track sequence is determined, and the standard water quality corresponding to the first cut water quality feature matrix and the recut water quality feature matrix are determined. The water quality difference between the two standard water qualities is calculated. Based on the water quality differences corresponding to all standard tracks after each transformation, a registration objective function is constructed. The registration objective function is solved iteratively to determine the target scale scaling factor and target translation parameter that minimize the registration objective function. The target translation parameter is then weighted using the target scale scaling factor to obtain the acquisition bias.

[0010] Furthermore, the sea area coordinates at the recutting moment are shifted in the opposite direction along the course by the acquisition deviation, and used as the correction coordinates at the recutting moment.

[0011] Furthermore, the method for obtaining the drift-induced coordinates includes: At each time point after the initial cut, the seawater drift vector is determined based on the time-domain integral of the absolute velocity of the seawater relative to the ground. The sea area coordinates at each time point are superimposed with the seawater drift vector to obtain the drift projection coordinates of the underwater robot.

[0012] Furthermore, the method for correcting the sea area coordinates at each moment includes: At each time point between the first cutoff time and the second cutoff time, the square of the time interval between each time point and the first cutoff time is normalized, and the normalized value is used as a preset time allocation weight. The navigation cumulative error vector is weighted using the preset time allocation weight to obtain the navigation cumulative error amortization vector. The navigation cumulative error amortization vector is subtracted from the drift projection coordinates at each time point to obtain the corrected drift projection coordinates.

[0013] The system for in-situ detection of marine water quality based on an underwater robot includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the method for in-situ detection of marine water quality based on an underwater robot.

[0014] The present invention has the following beneficial effects: This invention acquires the heading, sea area coordinates, bottom-tracking speed, and collected absolute seawater velocity and water quality parameters of an underwater robot navigating a target sea area, providing a data foundation for subsequent analysis. It analyzes abrupt changes in the physicochemical properties of water quality based on the temporal variation characteristics of water quality parameters and eliminates speed interference by utilizing bottom-tracking speed, thereby accurately determining the initial cut-off time of the underwater robot's first entry into the water quality front. It then fits an initial cut-off water quality feature matrix characterizing the water quality characteristics during the initial entry into the water quality front. Furthermore, it determines the re-cut-off time of the underwater robot's return and re-entry into the same water quality front, and eliminates the influence of cut-off angle differences by combining the difference in heading between the initial and re-cut-off times, fitting a re-cut-off water quality feature matrix characterizing the water quality characteristics during the re-entry into the water quality front. Finally, based on the initial and re-cut-off water quality feature matrices, it analyzes the effects of two entries into the same water quality front. The heading angle causes track distortion, making it impossible to directly calculate the water quality characteristic deviation between the two cuts through conventional point-to-point comparison, thus failing to obtain the underwater robot's acquisition deviation. Further, by combining the heading at the re-cutting moment, the corrected coordinates of the sea area at the re-cutting moment are determined. At each moment after the initial cut, the influence of ocean currents is analyzed, and the drift projection coordinates of the underwater robot are determined based on the sea area coordinates and the absolute velocity of seawater relative to the ground. Then, based on the difference between the drift projection coordinates and the corrected coordinates at the re-cutting moment, the navigation cumulative error vector is determined. Between the initial and re-cutting moments, the navigation cumulative error is allocated according to the preset time allocation weights and the navigation cumulative error vector at each moment, correcting the drift projection coordinates at each moment, and determining the accurate coordinates of the underwater robot in the target sea area at each moment. Finally, based on the corrected drift projection coordinates and water quality parameter interpolation, the water quality distribution of the target sea area is projected. This invention abandons the erroneous assumption of treating water quality boundaries as static references and accurately extrapolates the coordinates of water bodies drifting with ocean currents. At the same time, it accurately reconstructs the differences in observation perspectives by using water quality characteristic data during two water quality front processes. Furthermore, based on the physical objective law of the divergence of the second integral of the accelerometer inside the inertial navigation system, it corrects the coordinates of the underwater robot, thereby improving the accuracy of the underwater robot's navigation and positioning, and thus accurately extrapolating the water quality distribution of the target sea area. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart of an in-situ marine water quality detection method based on an underwater robot, provided as an embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a marine water quality in-situ detection method and system based on an underwater robot proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] The following description, in conjunction with the accompanying drawings, details the specific scheme of the in-situ marine water quality detection method and system based on an underwater robot provided by this invention.

[0020] This invention abandons the erroneous assumption of treating water quality boundaries as static references and accurately extrapolates the coordinates of water bodies drifting with ocean currents. At the same time, it accurately reconstructs the differences in observation perspectives by using water quality characteristic data during two water quality front processes. Furthermore, based on the physical objective law of the divergence of the second integral of the accelerometer inside the inertial navigation system, it corrects the coordinates of the underwater robot, thereby improving the accuracy of the underwater robot's navigation and positioning, and thus accurately extrapolating the water quality distribution of the target sea area.

[0021] Please see Figure 1 The diagram illustrates a flowchart of an in-situ ocean water quality detection method based on an underwater robot, according to an embodiment of the present invention, specifically including: Step S1: Obtain the underwater robot's heading, sea area coordinates, bottom tracking speed, and the collected absolute flow velocity and water quality parameters of the seawater as it navigates the target sea area.

[0022] It should be noted that the implementation scenario targeted by the embodiments of the present invention is as follows: the underwater robot executes the autonomous navigation instructions set in the task file. These instructions are used to drive the underwater robot to navigate in the target sea area in a back-and-forth or grid-like preset route. The preset route must include at least one set of intersecting segments or opposite parallel segments to ensure that at least one secondary revisit to the water quality front is generated. During the navigation, the yaw angle is corrected in real time by the PID controller to ensure that the deviation between the robot's trajectory and the preset task route in geographic space converges within a preset threshold (such as ±2 meters).

[0023] In one embodiment of the present invention, the underwater robot is equipped with a high-precision clock to provide a unified reference time for the main control computer; externally, it is equipped with an inertial navigation system, a Doppler velocimeter, and a multi-parameter water quality probe to collect data on the underwater robot's heading, sea area coordinates, bottom tracking speed, absolute seawater velocity relative to the ground, and water quality parameters during its navigation process; the above data acquisition processes are all known technical means, which are briefly described here: The inertial navigation system outputs attitude data, including roll, pitch, and yaw angles, as well as calculated coordinates (including the longitude, latitude, and depth of the underwater robot) at a set sampling frequency, such as 100Hz. The yaw angle is smoothed to obtain the underwater robot's heading. Conventional map projection algorithms, such as Mercator or Gauss-Kruger projection, are used to project the longitude and latitude in the calculated coordinates into position coordinates in the Northeast Geodetic absolute coordinate system (with the origin of the navigation start point). This, combined with the depth, determines the underwater robot's (3D) sea area coordinates. The main control computer timestamps the heading, attitude data, and sea area coordinates. It should be noted that all subsequent coordinate calculations are performed in the same Northeast Geodetic absolute coordinate system to ensure consistent vector dimensions. The Doppler velocimeter simultaneously activates bottom tracking and water tracking modes, outputting the acquisition results according to a set sampling frequency, such as 10Hz. Specifically, it uses the reflection of seabed sound waves to obtain the bottom tracking velocity (i.e., the absolute velocity of the underwater robot relative to the seabed) in the carrier coordinate system (with the underwater robot's center of gravity as the origin), and uses the scattering of sound waves from suspended objects in the water to obtain the underwater robot's velocity relative to the water in the carrier coordinate system. The main control computer timestamps the bottom tracking velocity and the velocity relative to the water, extracts the attitude data corresponding to the timestamps to construct a three-dimensional rotation matrix, and then projects and transforms the bottom tracking velocity and the velocity relative to the water into the Northeast Earth absolute coordinate system. The transformed bottom tracking velocity is subtracted from the velocity relative to the water to obtain the absolute velocity of the seawater relative to the land, which is then projected into the Northeast Earth absolute coordinate system and timestamped. A multi-parameter water quality probe collects water quality parameters at the location of the underwater robot at a set frequency, such as 1 Hz. The water quality parameters include at least salinity and turbidity. The implementer can also adjust the types and quantities of water quality parameters according to the actual application, such as increasing chloroplast concentration. The main control computer adds a system timestamp to the above water quality parameters. Using the system timestamp as the alignment index, and taking the 1Hz timestamp as the benchmark, the high-frequency (10Hz / 100Hz) data is downsampled or aligned by moving average. The heading, sea area coordinates, bottom tracking speed, seawater absolute velocity to the ground, and water quality parameters at the same time are packaged and encapsulated into a navigation data frame to provide a data foundation for subsequent analysis.

[0024] Step S2: Based on the temporal variation characteristics of water quality parameters and bottom tracking speed, determine the first cut time of the underwater robot's first entry into the water quality front, fit the first cut trajectory sequence during the underwater robot's first entry into the water quality front, and fit the first cut water quality feature matrix in combination with water quality parameters; determine the second cut time and second cut trajectory sequence of the underwater robot returning to enter the water quality front again, and fit the second cut water quality feature matrix in combination with the difference in heading between the first cut time and the second cut time.

[0025] In marine environments, water quality fronts (such as estuarine mixing boundaries and salinity step layers) are narrow bands of regions where physical and chemical properties change abruptly. Since the variation of water quality parameters in the frontal region is much higher than that in the background stable water body, relying solely on the temporal variation characteristics of water quality parameters will be severely interfered with by the changes in the underwater robot's speed (bottom tracking speed), which is very easy to produce false extreme value misjudgments. Therefore, based on the temporal variation characteristics of water quality parameters and bottom tracking speed, it can be determined whether the underwater robot has entered the water quality front, thereby determining the first entry time of the underwater robot into the water quality front and the second entry time into the same water quality front.

[0026] Preferably, in one embodiment of the present invention, considering that in complex ocean currents, if an underwater robot travels at high speed downstream through a gentle water quality gradient zone, it may collect water quality data that changes drastically in a short period of time, with an extremely high rate of water quality change; if the underwater robot travels slowly upstream through a steep water quality gradient zone, the water quality data collected over a long period of time changes slowly, with an extremely low rate of water quality change; the rate of water quality change cannot objectively reflect the true physical spatial structure of the water body, but by combining the underwater robot's bottom tracking speed, the interference of speed can be eliminated, thereby determining the water quality derivative that truly represents the objective gradient of seawater distribution, and then the moment when the underwater robot cuts into the true water quality front can be determined based on the abrupt change in the water quality derivative; therefore, the method for determining the initial cutting moment includes: When the underwater robot is navigating in the target sea area, the water quality derivative is determined at each moment based on the water quality change rate and bottom tracking speed; the first moment when the water quality derivative is greater than the preset derivative threshold is taken as the first cut-off moment.

[0027] It should be noted that in actual marine phenomena, gradients such as salinity and turbidity often occur in a coupled manner. To ensure the reliability of the analysis, a water quality parameter (such as salinity) is usually selected as the anchor point for the cutting front. The embodiments of this invention take the cutting salinity front as an example for analysis and description.

[0028] As an example, the rate of change of salinity at each moment is calculated (the salinity value at this moment minus the salinity value at the previous moment, and then divided by the time interval between adjacent moments; this is a well-known technique and will not be elaborated further); the rate of change of salinity is divided by the bottom tracking velocity (value) at that moment to obtain the salinity derivative (water quality derivative) at that moment; the salinity derivative is used to characterize the rate of change of salinity caused by displacement per meter.

[0029] It should be noted that when the bottom tracking speed is less than the preset speed measurement threshold, such as 0.05 m / s, the salinity derivative (water quality derivative) at that moment is set to 0 to avoid the denominator being too small, which would lead to numerical explosion or meaninglessness.

[0030] It should be noted that the method for obtaining the preset derivative threshold includes: having the underwater robot navigate in a sea area with known single water quality, and calculating the salinity derivative at each moment during the navigation process; using three times the variance of the absolute value of the salinity derivative at all moments as the preset derivative threshold, which is used to characterize the degree of natural fluctuation background noise of the water body when it is not significantly disturbed by external sources. Implementers can also determine the preset derivative threshold based on a steady-state self-calibration method: after the underwater robot enters the water, it uses a sliding window of a preset length (e.g., 3 minutes) to calculate the rolling variance of the salinity derivative in real time; when the fluctuation of the rolling variance within a consecutive number (e.g., 120) sliding windows is less than a preset proportion (e.g., 5%) of the mean of the rolling variance, it is determined that the current environment is a background noise environment, and the mean of the rolling variance within the consecutive sliding windows is used as the preset derivative threshold.

[0031] Ultimately, the first moment when the salinity derivative exceeds the preset derivative threshold is taken as the first cut-off moment; in other examples, the implementer may also take the first moment within a continuous period when the salinity derivative is greater than the preset derivative threshold as the first cut-off moment.

[0032] Further fitting of the first-cut trajectory sequence during the underwater robot's initial entry into the water quality front will prepare for subsequent fitting of the first-cut water quality feature matrix in conjunction with water quality parameters.

[0033] In a preferred embodiment of the present invention, the method for obtaining the first cut track sequence and the first cut water quality feature matrix includes: A preset time window is determined with the first cut time as the center. The bottom tracking speed is integrated in the time domain to determine the cumulative distance of each moment relative to the first cut time within the preset time window. The cumulative distances are sorted to construct the first cut track sequence. The first cut track sequence is resampled to obtain the first cut standard track sequence; the water quality parameters within the preset time window of the first cut time are standardized and resampled to obtain the first cut standard water quality sequence; the first cut standard track sequence and the first cut standard water quality sequence are used as column vectors to construct the first cut water quality feature matrix.

[0034] As an example, a preset time window is constructed by taking the first cut-off moment as the center and obtaining time periods of preset duration, such as 30 seconds, from both ends of the time sequence; implementers can also adjust the preset time window according to the actual situation.

[0035] It should be noted that, in order to avoid the underwater robot making a sharp turn when cutting into the water front, which would cause the subsequent calculated first cut trajectory sequence to lose its geometric representativeness, the validity of the straight course needs to be determined after the first cut moment: if the underwater robot's preset course includes a turning course (i.e., the difference in heading is greater than the preset straight course tolerance threshold, such as 15°) within the preset time window of the first cut moment, the first cut moment, i.e. the preset time window, is automatically discarded, and a new first cut moment when the underwater robot is in straight course is re-determined, and the preset time window is reconstructed.

[0036] Then, taking the first cut moment as the zero point of time axis, that is, setting the value corresponding to the track at the first cut moment to 0; performing time-domain integration on the tracking speed of each ground, and determining the cumulative distance of each moment relative to the first cut moment by summing them up; where, the sign of the cumulative distance before the first cut moment is negative, and the sign of the cumulative distance after the first cut moment is positive, thereby constructing the first cut track sequence. Further determine the track interval corresponding to the first cut track sequence, with the endpoints of the track interval being the cumulative distances at the beginning and end of the sequence; divide the track interval into equal intervals with a fixed step size (e.g., 0.1 meters) for resampling to generate the first cut standard track sequence; the first cut standard track sequence eliminates the uneven point density caused by speed fluctuations in the original first cut track sequence; Similarly, the water quality parameter sequences in the time domain aligned with the first cut trajectory sequence within the preset time window at the first cut time can be determined, namely the salinity sequence and the turbidity sequence. The salinity sequence and the turbidity sequence are standardized independently to eliminate the influence of dimensions. Specifically, Z-score standardization can be used, which is a well-known technique and will not be elaborated further. If the salinity sequence and the turbidity sequence are constant sequences with a standard deviation of 0, then the elements of the standardized sequence are directly set to 0. Then, the standardized salinity and turbidity sequences are resampled to obtain the initial standard water quality sequences, namely the initial standard salinity sequence and the initial standard turbidity sequence. Specifically, resampling can be performed using linear interpolation. Taking any data point Y(k) in the initial standard salinity sequence Y as an example, using well-known techniques, the following is a brief description: For each data point M(k) in the initial standard trajectory sequence M, the two data points m(j) and m(j+1) closest to M(k) are determined in the initial trajectory sequence m, where m(j) < M(k) < m(j+1). Similarly, the j-th and j+1-th data points y(j) and y(j+1) in the salinity sequence y are determined. The linear scaling factor is calculated based on the two-point formula. ,but ; Finally, the first cut standard track sequence, the first cut standard salinity sequence, and the first cut standard turbidity sequence are used as column vectors to construct the first cut water quality feature matrix. The three sequences are of equal length and correspond one-to-one. Each row vector in the first cut water quality feature matrix can be represented as the salinity and turbidity of the sea area at a track position.

[0037] Further determine the re-cutting time and re-cutting trajectory sequence of the underwater robot when it returns and re-enters the water quality front.

[0038] Preferably, in one embodiment of the present invention, considering that the method based on determining the initial cut time can re-determine the time when the underwater robot cuts into the water quality front, however, in complex ocean currents, the type of water quality front may not be the same; to ensure the accuracy of subsequent coordinate correction of the underwater robot, it is necessary to ensure that the water quality front into which it cuts again upon return is consistent with the initial cut, so it is also necessary to introduce spatial proximity constraints and feature similarity of the water quality front for verification. The sea area location helps to initially and roughly verify whether it is the same body of water, while the water quality feature matrix uses the water quality change characteristics during the cut to perform a more precise verification, thereby accurately determining the re-cut time when the underwater robot cuts into the water quality front upon return; therefore, the method for obtaining the re-cut time includes: After the initial cut, a preset time window is constructed with each time point as the center, the trajectory sequence is determined, and a water quality feature matrix is ​​constructed in combination with water quality parameters. Based on the similarity between the water quality feature matrix and the initial cut water quality feature matrix, and the deviation of the sea area coordinates at each time point from the sea area coordinates at the initial cut, it is determined whether the sea area at each time point is the sea area that first cuts into the water quality front. The water quality derivative at each time point is calculated, and the first time point in which the sea area is the sea area that first cuts into the water quality front and the water quality derivative is greater than the preset derivative threshold is taken as the recut time point.

[0039] As an example, firstly, after the initial cut, take any time as an example; construct a preset time window centered on this time, determine the trajectory sequence, and construct a water quality feature matrix in combination with water quality parameters; the method of constructing the water quality feature matrix is ​​the same as the method of constructing the initial cut water quality feature matrix, and will not be repeated here; The similarity between the water quality feature matrix at a given moment and the initial water quality feature matrix is ​​measured by the mean absolute value of the cosine similarity between column vectors with the same index in the matrix. The Euclidean distance between the sea area coordinates at that moment and the sea area coordinates at the initial cutting moment is calculated. If the similarity is greater than a preset threshold such as 0.85 and the Euclidean distance is less than a preset search radius such as 200 meters, the sea area at that moment is determined to be the sea area where the water quality front first cuts in; otherwise, it is not. The salinity derivative is calculated synchronously at each moment, which will not be elaborated further; the first moment when the sea area is the first to cut into the water quality front and the water quality derivative is greater than the preset derivative threshold is taken as the recutting moment; the method for obtaining the preset derivative threshold will not be elaborated further.

[0040] It should be noted that the validity of the straight route also needs to be determined at the re-cut time, and the determination method is the same as that for the first cut time, so it will not be repeated here.

[0041] After determining the recut time, the recut track sequence is further determined; the construction method of the recut track sequence is the same as that of the first cut track sequence, and will not be described again.

[0042] Because the course of the underwater robot when it first enters the water quality front is affected by the preset route, the course of the two entries may be opposite or intersect at a large angle. That is, the underwater robot may enter water quality B from water quality A when it first enters, and may enter water quality A from water quality B when it enters again. The water quality parameters collected in different entry processes may not correspond sequentially in time, making it difficult to assess the observation parallax at the water quality boundary when entering for the first and second times. Therefore, it is necessary to combine the difference in course between the first and second entry times to remove parallax interference and fit the water quality feature matrix of the second entry to prepare for obtaining the acquisition bias of the underwater robot in the future.

[0043] Preferably, in one embodiment of the present invention, considering that the method for determining the water quality feature matrix based on the first cut can determine the water quality feature matrix for the second cut, however, if the heading at the time of the first cut is opposite to or intersects the heading at the time of the second cut at a large angle, the spatial water quality features (such as a sudden change in salinity from low to high) will be mirrored in the time series (recorded as a sudden drop in salinity from high to low). In order to ensure that the water quality features collected at the time of the second cut maintain the same spatial variation trend as those at the time of the first cut, the method for obtaining the water quality feature matrix for the second cut includes: The re-cut track sequence is resampled to obtain the re-cut standard track sequence; the water quality parameters within the preset time window of the re-cut time are standardized and resampled to obtain the re-cut standard water quality sequence; the heading angle between the heading at the first cut time and the heading at the re-cut time is determined. If the heading angle is less than or equal to the preset angle, the recut standard track sequence and the recut standard water quality sequence are used as column vectors to construct the recut water quality feature matrix; if the heading angle is greater than the preset angle, the mirrored recut standard track sequence and the recut standard water quality sequence are used as column vectors to construct the recut water quality feature matrix.

[0044] As an example, the re-cut track sequence is first resampled to obtain the re-cut standard track sequence; the method for obtaining the re-cut standard track sequence is the same as that for obtaining the first cut standard track sequence, and will not be repeated here; further, the re-cut standard water quality sequence, namely the re-cut standard salinity sequence and the re-cut standard turbidity sequence, is determined, and the method for obtaining the first cut standard salinity sequence and the first cut standard turbidity sequence is the same as that for obtaining the first cut standard salinity sequence and the first cut standard turbidity sequence, and will not be repeated here. Furthermore, the minimum angle between the heading at the initial cut and the heading at the re-cut is taken as the heading angle. The preset angle is set to 90°. When the heading angle is ≤90°, it is considered that the underwater robot's cutting direction when it cuts into the water quality front again is similar to or the same as when it cuts into the water quality front. The re-cut standard track sequence, re-cut standard salinity sequence, and re-cut standard turbidity sequence can be directly used as column vectors to construct the re-cut water quality feature matrix. When the heading angle is >90°, it is considered that the underwater robot cuts into the water quality front head-on from the opposite direction. The sequence needs to be mirrored (a well-known technical means, which will not be elaborated here). The mirrored re-cut standard track sequence, re-cut standard salinity sequence, and re-cut standard turbidity sequence are used as column vectors to construct the re-cut water quality feature matrix.

[0045] Step S3: Based on the first cut water quality feature matrix and the second cut water quality feature matrix, obtain the acquisition deviation of the underwater robot, and combine it with the heading at the second cut time to determine the corrected coordinates of the sea area at the second cut time; at each time after the first cut time, determine the drift projection coordinates of the underwater robot based on the sea area coordinates and the absolute velocity of seawater relative to the ground.

[0046] After determining the initial water quality feature matrix and the subsequent water quality feature matrix, the differences in water quality characteristics when entering the same water quality front twice can be further compared, thereby assessing the acquisition bias of the underwater robot. The acquisition bias is the spatial deviation caused by the equivalent stretching or compression of water quality change characteristics caused by parallax distortion and turbulence disturbance when the underwater robot enters the same water quality front at different directions on the cross-course.

[0047] Preferably, in one embodiment of the present invention, due to the distortion of the trajectory caused by the heading angles of two cuts into the same water quality front, it is impossible to directly obtain the water quality characteristic deviation at the two cuts through conventional point-by-point comparison; in order to compensate for the trajectory stretching or compression distortion caused by the cutting heading angle, a dimensionally constrained registration algorithm is used to perform scaling and translation transformations on the recut standard trajectory sequence, and the column mapping relationship between the standard trajectory sequence and the standard water quality sequence is used to evaluate the water quality characteristic deviation of two cuts into the same water quality front, so as to construct a registration objective function; then, the scaling and translation parameters are continuously adjusted to find the scaling and translation parameters that make the registration objective function converge; the scaling and translation parameters at convergence respectively characterize the water body spatial distortion factor and translation amount caused by two different cutting heading angles, which can help measure the spatial deviation caused by parallax distortion, i.e., the acquisition deviation; the method for obtaining the acquisition deviation includes: Variable scaling factors and translation parameters are used to transform the recut standard track sequence. Based on the column mapping relationship between the standard track sequence and the standard water quality sequence, the standard water quality corresponding to each standard track after each transformation in the recut standard track sequence is determined, and the standard water quality corresponding to the first cut water quality feature matrix and the recut water quality feature matrix are determined. The water quality difference between the two standard water qualities is calculated. Based on the water quality differences corresponding to all standard tracks after each transformation, a registration objective function is constructed. The registration objective function is solved iteratively to determine the target scale scaling factor and target translation parameter that minimize the registration objective function. The target translation parameter is then weighted using the target scale scaling factor to obtain the acquisition bias.

[0048] As an example, a variable scaling factor b and a translation parameter c are set to perform a linear transformation on each standard track N(k) in the column vector corresponding to the recut standard track sequence in the recut water quality feature matrix. The transformed standard track is: The initial value of the scaling factor b is set to 1, and the search interval is [0.5, 1.5]. The initial value of the translation parameter c is set to 0, and the search interval is [-50, 50]. Taking salinity as an example, based on the column mapping relationship between the standard track sequence and the standard salinity sequence, the standard salinity mapped to each transformed standard track in the first cut water quality feature matrix (first cut standard salinity sequence) can be determined. Similarly, the standard salinity mapped to each transformed standard track in the second cut water quality feature matrix (second cut standard salinity sequence) can be determined, and the sum of squares of the differences between the two standard salinities can be calculated. Similarly, for turbidity, the sum of squares of the differences between the two standard turbidities corresponding to each transformed standard track can be determined. The sum of all sums of squares of differences corresponding to all transformed standard tracks is used as the registration objective function. By changing the scaling factor b and the translation parameter c, different registration objective function values ​​can be obtained. The scaling factor b and translation parameter c corresponding to the minimum registration objective function are used as the target scaling factor and target translation parameter. Since the target translation parameter is a relative value obtained in a distorted, compressed, or stretched scaling space, directly using this value will produce scaling bias. Therefore, the target scaling factor is multiplied by the target translation parameter to perform inverse scaling and restore the actual deviation caused by local turbulent sloshing of the water body along the current route direction at the real geographical absolute scale. The product is used as the acquisition bias.

[0049] Once the acquisition bias is determined, the sea area coordinates at the re-cutting moment can be further corrected by combining the heading at the re-cutting moment, and the corrected coordinates of the sea area at the re-cutting moment can be determined.

[0050] Preferably, in one embodiment of the present invention, considering that the coordinates calculated by the inertial navigation system of the underwater robot are not equivalent to the actual coordinates of the water quality or water mass due to the influence of the water front intrusion angle and local turbulence, the actual calculated sea area coordinates at the re-cutting moment can be compensated by reverse translation, thereby generating corrected coordinates that eliminate observation parallax and local turbulence interference; then the sea area coordinates at the re-cutting moment are moved in the reverse direction along the course to collect the deviation, and the corrected coordinates at the re-cutting moment are obtained; the corrected coordinates are used to characterize the position that the underwater robot should be in without considering ocean current movement.

[0051] Specifically, the scalar acquisition deviation is decomposed into a three-dimensional vector in the absolute coordinate system of the northeast according to the heading angle at the recutting time. This three-dimensional vector is then superimposed onto the sea area coordinates at the recutting time to obtain the corrected coordinates.

[0052] Since the water front is not a static seabed structure, it will undergo overall spatial translation with the movement of regional ocean currents; and the absolute velocity of seawater relative to the ground can help assess the movement of ocean currents. Therefore, the drift simulation coordinates of the underwater robot can be determined at each moment after the initial cut, based on the sea area coordinates and the absolute velocity of seawater relative to the ground.

[0053] Preferably, in one embodiment of the present invention, the method for obtaining drift-induced coordinates includes: At each time point after the initial cut, the seawater drift vector is determined based on the time-domain integral of the absolute seawater velocity relative to the ground. The sea area coordinates at each time point are superimposed with the seawater drift vector to obtain the drift projection coordinates of the underwater robot.

[0054] Specifically, the absolute velocity of the seabed relative to the ground is integrated in the time domain, and the integration result is used as the seawater drift vector at the corresponding time. This is a well-known technical method and will not be elaborated further. Furthermore, the sea area coordinates at each time are added to the seawater drift vector to obtain the drift projection coordinates of the underwater robot.

[0055] Step S4: Determine the navigation cumulative error vector based on the difference between the drift projection coordinates and the correction coordinates at the re-cutting time; between the first cutting time and the re-cutting time, correct the drift projection coordinates at each time according to the preset time allocation weight and the navigation cumulative error vector at each time, and extrapolate the water quality distribution of the target sea area based on the corrected drift projection coordinates and water quality parameter interpolation.

[0056] Considering that the drift projection coordinates at the re-cutting moment are calculated based on high-precision velocity integrals and are closer to the actual displacement of the underwater robot; while the correction coordinates at the re-cutting moment, although eliminating observation parallax and local turbulence interference, still contain divergent cumulative errors caused by the inertial navigation system; therefore, this embodiment of the invention will determine the navigation cumulative error vector based on the difference between the drift projection coordinates at the re-cutting moment and the correction coordinates.

[0057] Specifically, the navigation cumulative error vector is obtained by subtracting the correction coordinates at the re-cutting time from the drift-induced coordinates at the re-cutting time.

[0058] The navigation cumulative error vector isolates the observation parallax scaling of macroscopic drift and heading difference in hydrodynamics, and characterizes the sum of invalid cumulative divergence generated by the underwater robot's inertial navigation system between the first cut and the second cut due to the noise of internal hardware (accelerometer and gyroscope) after mathematical double integration.

[0059] Furthermore, between the initial cut and the recut, the drift simulation coordinates at each moment are corrected based on the preset time allocation weights and navigation cumulative error vectors, thereby determining the accurate coordinates of the underwater robot in the target sea area at each moment, in order to prepare for subsequent accurate simulation of water quality distribution.

[0060] The navigation cumulative error vector is the cumulative error between the first cut and the second cut. The navigation cumulative error can be allocated by preset time allocation weights, thereby helping to correct the drift simulation coordinates at each time.

[0061] Preferably, in one embodiment of the present invention, considering that due to the underlying hardware characteristics of the inertial navigation system, the constant bias error of its internal accelerometer, after two consecutive integrations over time, results in a cumulative divergence in the final position domain that is proportional to the square of time; if a conventional linear ratio (i.e., directly dividing the time increment by the total duration) is used to distribute this total error, the corrected track, although forcibly aligned at the start and end points, will exhibit severe bow-shaped geometric distortion in the middle of the track, which does not conform to actual physical laws; using a pre-defined time allocation weight with a power of two can conform to the physical laws of the hardware's underlying divergence, ensuring that the error distributed to each historical trajectory point is truly accurate; therefore, the method for correcting the sea area coordinates at each moment includes: At each time point between the first cut and the second cut, the square of the time interval between each time point and the first cut is normalized. The normalized value is used as a preset time allocation weight. The navigation cumulative error vector is weighted using the preset time allocation weight to obtain the navigation cumulative error distribution vector. The drift projection coordinates at each time point are subtracted from the navigation cumulative error distribution vector to obtain the corrected drift projection coordinates.

[0062] As an example, the normalization method is to divide by the square of the time interval between the first cut and the second cut; further, the preset time allocation weight at each time step is multiplied by the navigation cumulative error vector to obtain the navigation cumulative error amortization vector at that time step; by subtracting the navigation cumulative error amortization vector from the drift projection coordinates at that time step, the divergence cumulative error caused by the inertial navigation system can be eliminated to obtain the corrected drift projection coordinates; the corrected drift projection coordinates are the true and accurate coordinates of the underwater robot in the target sea area.

[0063] Finally, the water quality distribution of the target sea area is inferred based on the corrected drift coordinates and water quality parameters at each time point.

[0064] Specifically, during the navigation process, the water quality parameters (including salinity and turbidity) at each moment are extracted using timestamps. The water quality parameters at each moment are then packaged with the corrected drift projection coordinates to obtain a composite coordinate point containing the coordinate position and water quality parameters. Since the wake of an underwater robot appears as discrete points or single lines with specific spacing in the ocean, these discrete lines cannot directly cover and present the full picture of the water quality gradient of the entire target sea area. Therefore, the inverse distance weighting (IDW) spatial interpolation algorithm, which is commonly used in geostatistics, is further applied to expand and extrapolate the array of composite coordinate points into a seamless, continuous two-dimensional water quality distribution field that covers the entire target sea area. In the interpolation process, an anisotropic search ellipse is used, with the major axis along the tangent of the front to fit the fluid distribution characteristics. After generating the two-dimensional water quality distribution field, further color rendering operations can be performed to output the final in-situ two-dimensional water quality thermal map, which eliminates coordinate breakpoints and eliminates interference from fluid offset and observation parallax, and then visualizes it.

[0065] It should be noted that the IDW algorithm and color rendering operations mentioned above are well-known technical methods and will not be described in detail here.

[0066] Based on the same inventive concept, this invention also proposes an in-situ marine water quality detection system based on an underwater robot. The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the in-situ marine water quality detection method based on an underwater robot described in steps S1-S4.

[0067] In summary, this invention acquires the underwater robot's heading, sea area coordinates, bottom tracking speed, and collected seawater absolute current velocity and water quality parameters during its navigation in the target sea area. Based on the temporal variation characteristics of the water quality parameters and the bottom tracking speed, it determines the initial cut-in time of the underwater robot's first entry into the water quality front, fits the initial cut-in trajectory sequence during the initial entry into the water quality front, and fits the initial cut-in water quality feature matrix in combination with the water quality parameters. It then determines the re-cut-in time and re-cut-in trajectory sequence of the underwater robot's return to re-enter the water quality front, and fits the re-cut-in water quality feature matrix in combination with the difference in heading between the initial cut-in time and the re-cut-in time. Based on the initial cut-in water quality feature matrix... This invention uses a matrix of water quality characteristics and a re-cutting water quality feature matrix to obtain the underwater robot's acquisition deviation. Combined with the heading at the re-cutting moment, the corrected coordinates of the sea area at that moment are determined. At each moment after the initial cut, the underwater robot's drift-based coordinates are determined based on the sea area coordinates and the absolute velocity of the seawater relative to the ground. The navigation cumulative error vector is determined based on the difference between the drift-based coordinates and the corrected coordinates at the re-cutting moment. Between the initial and re-cutting moments, the drift-based coordinates at each moment are corrected according to the preset time allocation weights and the navigation cumulative error vector. Based on the corrected drift-based coordinates and water quality parameter interpolation, the water quality distribution of the target sea area is inferred. This invention abandons the erroneous assumption of treating the water quality boundary as a static reference, accurately inferring the water body's drift coordinates with ocean currents. Simultaneously, it accurately reconstructs the difference in observation perspective by utilizing water quality feature data from two water quality front processes. Furthermore, based on the physical objective law of the second integral divergence of the accelerometer within the inertial navigation system, the underwater robot's coordinates are corrected, thereby improving the underwater robot's navigation and positioning accuracy, and ultimately accurately inferring the water quality distribution of the target sea area.

[0068] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0069] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for in-situ detection of ocean water quality based on an underwater robot, characterized in that, The method includes: Acquire the underwater robot's heading, sea area coordinates, bottom tracking speed, and the collected absolute flow velocity of seawater to the ground and water quality parameters when it navigates in the target sea area; Based on the temporal variation characteristics of water quality parameters and bottom tracking speed, the first cut time of the underwater robot's first entry into the water quality front is determined, the first cut trajectory sequence of the underwater robot during the first entry into the water quality front is fitted, and the first cut water quality feature matrix is ​​fitted in combination with water quality parameters; the second cut time and second cut trajectory sequence of the underwater robot returning and re-entering the water quality front are determined, and the second cut water quality feature matrix is ​​fitted in combination with the difference in heading between the first cut time and the second cut time. Based on the initial and subsequent water quality feature matrices, the acquisition deviation of the underwater robot is obtained, and the corrected coordinates of the sea area at the subsequent cutting time are determined by combining the heading at the subsequent cutting time. At each time after the initial cutting time, the drift projection coordinates of the underwater robot are determined based on the sea area coordinates and the absolute velocity of seawater relative to the ground. Based on the difference between the drift-induced coordinates and the corrected coordinates at the re-cutting moment, the navigation cumulative error vector is determined; between the first cutting moment and the re-cutting moment, the drift-induced coordinates at each moment are corrected according to the preset time allocation weight and the navigation cumulative error vector at each moment, and the water quality distribution of the target sea area is inferred based on the corrected drift-induced coordinates and water quality parameter interpolation. The method for obtaining the acquisition deviation includes: Variable scaling factors and translation parameters are used to transform the recut standard track sequence. Based on the column mapping relationship between the standard track sequence and the standard water quality sequence, the standard water quality corresponding to each standard track after each transformation in the recut standard track sequence is determined, and the standard water quality corresponding to the first cut water quality feature matrix and the recut water quality feature matrix are determined. The water quality difference between the two standard water qualities is calculated. Based on the water quality differences corresponding to all standard tracks after each transformation, a registration objective function is constructed. The registration objective function is solved iteratively to determine the target scale scaling factor and target translation parameter that minimize the registration objective function. The target translation parameter is then weighted using the target scale scaling factor to obtain the acquisition bias. The method for correcting the sea area coordinates at each moment includes: At each time point between the first cutoff time and the second cutoff time, the square of the time interval between each time point and the first cutoff time is normalized, and the normalized value is used as a preset time allocation weight. The navigation cumulative error vector is weighted using the preset time allocation weight to obtain the navigation cumulative error amortization vector. The navigation cumulative error amortization vector is subtracted from the drift projection coordinates at each time point to obtain the corrected drift projection coordinates.

2. The method for in-situ detection of marine water quality based on an underwater robot according to claim 1, characterized in that, The method for the initial cut-off time includes: When the underwater robot is navigating in the target sea area, the water quality derivative at each moment is determined based on the water quality change rate and the bottom tracking speed; the first moment when the water quality derivative is greater than a preset derivative threshold is taken as the first cut-off moment.

3. The in-situ marine water quality detection method based on an underwater robot according to claim 1, characterized in that, The methods for obtaining the first cut track sequence and the first cut water quality feature matrix include: A preset time window is determined with the first cut time as the center. The bottom tracking speed is integrated in the time domain to determine the cumulative distance of each moment relative to the first cut time within the preset time window. The cumulative distances are sorted to construct the first cut track sequence. The first cut trajectory sequence is resampled to obtain the first cut standard trajectory sequence; the water quality parameters within the preset time window of the first cut time are standardized and resampled to obtain the first cut standard water quality sequence; the first cut standard trajectory sequence and the first cut standard water quality sequence are respectively used as column vectors to construct the first cut water quality feature matrix.

4. The in-situ marine water quality detection method based on an underwater robot according to claim 2, characterized in that, The method for obtaining the recutting time includes: After the initial cut-off time, a preset time window is constructed with each time point as the center, the trajectory sequence is determined, and a water quality feature matrix is ​​constructed in combination with water quality parameters. Based on the similarity between the water quality feature matrix and the initial cut-off water quality feature matrix, and the deviation of the sea area coordinates at each time point from the sea area coordinates at the initial cut-off time, it is determined whether the sea area at each time point is the sea area that first cuts into the water quality front. The water quality derivative at each time point is calculated, and the first time point in which the sea area is the sea area that first cuts into the water quality front and the water quality derivative is greater than a preset derivative threshold is taken as the recut-off time point.

5. The in-situ marine water quality detection method based on an underwater robot according to claim 3, characterized in that, The method for obtaining the recut water quality feature matrix includes: The re-cut track sequence is resampled to obtain the re-cut standard track sequence; the water quality parameters within the preset time window of the re-cut time are standardized and resampled to obtain the re-cut standard water quality sequence; the heading angle between the heading at the first cut time and the heading at the re-cut time is determined. If the heading angle is less than or equal to a preset angle, the recut standard track sequence and the recut standard water quality sequence are used as column vectors to construct a recut water quality feature matrix; if the heading angle is greater than the preset angle, the mirrored recut standard track sequence and the recut standard water quality sequence are used as column vectors to construct a recut water quality feature matrix.

6. The in-situ marine water quality detection method based on an underwater robot according to claim 1, characterized in that, The sea area coordinates at the recutting time are shifted in the opposite direction along the course by the acquisition deviation to obtain the corrected coordinates at the recutting time.

7. The in-situ marine water quality detection method based on an underwater robot according to claim 1, characterized in that, The method for obtaining the drift-induced coordinates includes: At each time point after the initial cut, the seawater drift vector is determined based on the time-domain integral of the absolute velocity of the seawater relative to the ground. The sea area coordinates at each time point are superimposed with the seawater drift vector to obtain the drift projection coordinates of the underwater robot.

8. A marine water quality in-situ detection system based on an underwater robot, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the in-situ marine water quality detection method based on an underwater robot as described in any one of claims 1 to 7.

Images