Multi-beam underwater measurement method and system

Abstract

The invention belongs to the technical field of underwater measurement, and provides a multi-beam underwater measurement method and system. The method comprises the following steps: performing low-resolution detection on a target water area by adopting a single-beam mode, and obtaining the water depth and terrain complexity of the target water area based on obtained first detection data; screening to obtain a plurality of interference signal identification models based on water depth and terrain complexity, taking each interference signal identification model as a base model, and integrating all the base models into an integrated model; performing high-resolution detection on the target water area by adopting a multi-beam mode, performing interference signal identification on the obtained second detection data by using an integrated model, and attenuating or removing the identified interference signal to obtain third detection data; and measuring underwater topographic data of the target water area based on the processed third detection data. According to the invention, multi-beam interference signals can be accurately identified, and the accuracy of underwater topographic data obtained through detection is ensured.

Description

Technical Field

[0001] The present invention relates to the field of underwater measurement technology, and in particular to a multi-beam underwater measurement method and system. Background Art

[0002] Underwater measurement technology plays a vital role in many fields such as marine resource exploration, underwater engineering construction, and marine scientific research. Traditional underwater measurement technology mainly uses single-beam ultrasonic measurement methods, and its working principle is based on the principle of acoustic ranging, that is, a single ultrasonic transducer transmits sound waves to underwater targets. The sound waves are reflected after encountering the target during the propagation process. The transducer receives the reflected signal and calculates the distance of the target based on the propagation time of the sound wave and the known sound speed using the formula d=v×t / 2 (where d is the target distance, v is the sound speed, and t is the round-trip time of the sound wave). The position of the target is then determined by combining the attitude and position information of the measuring device.

[0003] However, with the continuous improvement of the requirements for underwater measurements in related fields, the limitations of the single-beam measurement method have become increasingly prominent. First, its measurement efficiency is extremely low. Since only one point can be measured each time, it takes a lot of time and manpower to measure in large underwater areas, and the coverage is very limited, which makes it difficult to meet the needs of large-scale projects to quickly obtain data. Secondly, there are serious deficiencies in measurement accuracy. The water environment is complex and changeable. Factors such as temperature, salinity, and flow rate will have a significant impact on the speed of sound, which will lead to high measurement errors. For example, in sea areas with large temperature changes, the speed of sound may change significantly, resulting in a large deviation in the target distance calculated based on a fixed speed of sound. Furthermore, the single-beam measurement method has poor real-time performance, and its data processing process is relatively cumbersome. From signal acquisition to the final generation of usable measurement data, multiple processing steps are required, resulting in slow data processing speed, which cannot provide timely support for on-site decision-making, and it is difficult to meet the current urgent needs for large-scale, high-precision real-time measurement.

[0004] In recent years, in order to overcome the defects of the single-beam measurement method, multi-beam ultrasonic measurement technology has emerged and gradually become a research hotspot. By emitting multiple ultrasonic beams at the same time, this technology can obtain data from multiple points in one measurement, greatly improving the measurement efficiency. Moreover, multiple beams measure the target area from different angles, and data can be compared and corrected with each other, thereby significantly improving the measurement accuracy. However, multi-beam ultrasonic measurement technology faces a key problem in practical applications, that is, multi-beam signals are prone to mutual interference. Since multiple beams propagate simultaneously in space, their emission and reception processes affect each other, resulting in a large amount of interference components mixed in the received signal, which seriously reduces the measurement accuracy and cannot give full play to the advantages of multi-beam measurement technology, limiting the wide application of this technology in practical scenarios.

[0005] Therefore, providing a multi-beam underwater measurement method that can effectively solve the multi-beam signal interference problem, further improve the measurement accuracy and efficiency, and meet the real-time requirements has important practical significance and broad application prospects. Summary of the invention

[0006] In response to the above technical problems, the present invention provides a multi-beam underwater measurement method, system, electronic device, computer storage medium and computer program product to solve the problems of low measurement accuracy, complex equipment and high cost existing in the prior art.

[0007] The present invention discloses a multi-beam underwater measurement method, which comprises the following steps: using a single-beam mode to perform low-resolution detection on a target water area, and obtaining the water depth and terrain complexity of the target water area based on first detection data obtained; S2, screening and obtaining a plurality of interference signal recognition models based on the water depth and the terrain complexity, using each of the interference signal recognition models as a base model, and integrating all the base models into an integrated model; using a multi-beam mode to perform high-resolution detection on the target water area, using the integrated model to identify interference signals from the second detection data obtained, attenuating or removing the identified interference signals, and obtaining third detection data; and determining the underwater terrain data of the target water area based on the processed third detection data.

[0008] Optionally, the first detection data includes distance data and position data, and the water depth and terrain complexity of the target water area are obtained based on the first detection data, including: calculating the average value of each distance data and using it as the water depth; performing three-dimensional projection on the corresponding distance data according to each position data to obtain a three-dimensional projection map, and using a classifier to classify the three-dimensional projection map to obtain the terrain complexity.

[0009] Optionally, a number of interference signal recognition models are screened based on the water depth and the terrain complexity, each of the interference signal recognition models is used as a base model, and all the base models are integrated into an integrated model, including: matching and analyzing the water depth and the terrain complexity with the standard water depth and standard terrain complexity corresponding to each interference signal recognition model in the model library, to obtain a first number of interference signal recognition models that meet the matching threshold value, and use them as base models; and randomly screening a second number of interference signal recognition models that do not meet the matching threshold value, and also use them as base models; wherein the second number is lower than the first number; and integrating all the screened base models into an integrated model.

[0010] Optionally, the method of using a classifier to classify the three-dimensional projection image to obtain the terrain complexity includes: using a classifier to classify the three-dimensional projection image to obtain the terrain complexity and its classification confidence; then the second quantity is determined in the following manner: performing a comparative calculation based on the classification confidence and preset control data to obtain the second quantity; wherein, in the control data, the second quantity is negatively correlated with the classification confidence.

[0011] Optionally, before determining the underwater terrain data of the target waters based on the processed third detection data, the method also includes: determining the recognition confidence of the integrated model for the interference signal in the second detection data; if the recognition confidence is lower than the confidence threshold, adding an interference signal mark to the corresponding third detection data, and associating the recognition confidence of the interference signal and the corresponding second detection data with the third detection data.

[0012] The present invention also discloses a multi-beam underwater measurement system, which includes an ultrasonic transmitting module, an ultrasonic receiving module, and a signal processing module; the ultrasonic transmitting module adopts a linear array transducer, which is composed of a plurality of ultrasonic transducer units; the ultrasonic receiving module includes a plurality of receiving transducers for receiving reflected ultrasonic signals; the signal processing module includes a filtering unit, an amplifying unit, and a digital signal processing unit; the filtering unit is used to eliminate noise and extract effective signals; the amplifying unit performs gain adjustment on the signal to ensure that the signal strength is moderate; the digital signal processing unit processes the ultrasonic signals processed by the filtering unit and the amplifying unit to obtain underwater terrain data of the target waters; the filtering unit has a processor and a memory, and the processor is used to call and execute a computer program in the memory to implement a multi-beam underwater measurement method as described above.

[0013] Optionally, the system further comprises a control module, which adopts an embedded system to achieve high-precision timing control and data processing to coordinate the work of the ultrasonic transmitting module, the ultrasonic receiving module and the signal processing module.

[0014] The present invention also discloses an electronic device, comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor executes the computer program to implement the method described in any of the preceding items.

[0015] The present invention also discloses a computer storage medium, wherein the computer storage medium stores a computer program, and the computer program is executed by a processor to implement the method as described in any of the above items.

[0016] The present invention also discloses a computer program product, which includes computer code. When the computer code is executed by a processor of an electronic device, the method described in any of the above items is implemented.

[0017] The beneficial effect of the present invention is at least that: the multi-beam underwater measurement scheme proposed in the present invention first uses a single beam to quickly determine the water depth and terrain characteristics of the target water area, thereby screening the base model and constructing a suitable integrated model to achieve accurate identification of interference signals with different characteristics, thereby solving the signal interference problem of multi-beams, allowing the advantages of multi-beam technology to be brought into play, ensuring the accuracy of the underwater terrain data obtained by detection, and providing accurate data for on-site decision-making in a timely manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0019] Figure 1 It is a flow chart of a multi-beam underwater measurement method disclosed in an embodiment of the present invention.

[0020] Figure 2 It is a structural schematic diagram of a multi-beam underwater measurement system disclosed in an embodiment of the present invention.

[0021] Figure 3 It is a structural schematic diagram of a signal processing module disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The following is a description of the implementation of the present application by specific specific embodiments. People familiar with the technology can easily understand other advantages and effects of the present application from the contents disclosed in this specification. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0023] In addition, the technical features involved in the different embodiments of the present application described below can be combined with each other as long as they do not conflict with each other.

[0024] like Figure 1 , Figure 2As shown, in response to the above technical problems, an embodiment of the present invention discloses a multi-beam underwater measurement method, which includes the following steps: S1, using a single-beam mode to perform low-resolution detection of the target water area, and deriving the water depth and terrain complexity of the target water area based on the first detection data obtained.

[0025] In this step, the multi-beam underwater measurement device of the present invention can switch between single-beam mode and multi-beam mode. Single-beam measurement is relatively simple and can preliminarily reflect the general situation of the water area, so the present invention first uses the single-beam mode to carry out low-resolution detection of the target water area. That is, the multi-beam measurement device only emits one beam for measurement each time, and the interval between adjacent detection positions is significantly larger than the multi-beam measurement mode, that is, sparse sampling measurement is performed on the target water area. In this way, the first detection data of the target water area is obtained.

[0026] Based on the first detection data obtained, the water depth information of the target water area is calculated. At the same time, the terrain complexity of the water area is evaluated based on the water depth information, for example, whether it is flat terrain or complex terrain such as undulations and gullies. The terrain complexity can be expressed in the form of terrain complexity level, terrain complexity value, etc. The higher the level or value, the more complex the corresponding terrain.

[0027] S2, screening and obtaining a plurality of interference signal recognition models based on the water depth and the terrain complexity, taking each of the interference signal recognition models as a base model, and integrating all the base models into an integrated model.

[0028] In this step, different water depths and terrain complexities will cause interference signals with different characteristics to be generated during multi-beam measurement. Therefore, the present invention pre-constructs and trains multiple interference signal recognition models for different water depths and terrain complexities to form a model library. The characteristics of interference signals at different water depths and terrain complexities and the requirements for interference signal recognition models are as follows: 1) When the beam propagates in shallow water, due to the thin water layer, the path of the sound wave from emission to reception is relatively short, and the beams are more likely to interact with each other. At the same time, the bottom reflection in shallow water is strong, and the reflected wave and the direct wave may produce complex interference between different beams, resulting in the interference signal showing strong periodicity and regularity, and the signal strength is relatively large.

[0029] In view of the strong and periodic characteristics of interference signals in shallow water areas, it is appropriate to adopt a model that can capture the periodicity and regularity of the signal. For example, the frequency domain analysis model based on Fourier transform can effectively extract the frequency characteristics of the interference signal and identify the interference signal by detecting specific frequency components in the spectrum. In addition, some simple linear models, such as the autoregressive model (AR), can also model and predict shallow water interference signals with certain regularity.

[0030] 2) When the beam propagates in deep water, the increased water depth makes the sound wave propagation path longer, and the interference signal between beams is more significantly affected by factors such as water absorption and scattering. This will cause the intensity of the interference signal to gradually weaken, and the attenuation and distortion of the signal will be more obvious. In addition, the environment in deep water is relatively stable and the noise level is relatively low. However, since the sound waves will be affected by hydrological conditions such as thermocline and halocline during long-distance propagation, the interference signal between beams may have complex situations such as frequency offset and phase change, and its spectral characteristics will change with the change of water depth.

[0031] The complex changes of interference signals in deep water areas require more powerful models to handle. The convolutional neural network (CNN) in deep learning can automatically extract the features of interference signals through convolutional layers and pooling layers, and has good adaptability to complex features such as frequency offset and phase change of signals. At the same time, the recurrent neural network (RNN) and its variant, the long short-term memory network (LSTM), can process time series information and have better recognition capabilities for interference signals in deep water areas that change with time and depth.

[0032] 3) In a flat underwater terrain environment, the interference signal between beams is relatively stable and uniform. Due to the consistency of the terrain, the reflection and scattering of sound waves are relatively regular. The interference signal is mainly generated by the characteristics of the water body itself and the inherent interaction between beams. Its spatial distribution is relatively uniform, the amplitude and phase changes of the signal are relatively small, and the spectrum characteristics are relatively stable.

[0033] For relatively simple and stable interference signals in flat terrain, traditional machine learning models, such as support vector machines (SVMs) and decision trees, can usually achieve good recognition results. These models can accurately identify interference signals by learning a small number of features, because the features of interference signals in flat terrain are relatively easy to extract and describe.

[0034] 4) When the underwater terrain is complex, such as mountains, canyons, reefs, etc., the beam will encounter irregular reflective surfaces and scatterers during propagation. This will cause the interference signal between beams to become very complex, and the beams at different locations will be interfered to different degrees, and the spatial distribution of the interference signal will show obvious non-uniformity. For example, near mountains, the beam may experience strong reflection and diffraction, resulting in a complex multipath effect, causing the interference signal to contain multiple reflected waves from different paths. The amplitude and phase of the signal will change dramatically, and the spectrum will become more complex, and multiple peaks and band broadening may occur.

[0035] Interference signals in complex terrain require models with more flexibility and powerful representation capabilities. Deep learning models such as deep neural networks (DNNs) can learn complex feature representations through nonlinear transformations of multiple layers of neurons, and can better handle the spatial inhomogeneity and complex spectral characteristics of interference signals. In addition, some methods based on model fusion, such as models that combine CNN and RNN, can also simultaneously utilize CNN's ability to extract spatial features and RNN's ability to process time series information, thereby more comprehensively identifying inter-beam interference signals in complex terrain.

[0036] Based on different water depths and different terrain complexities, appropriate algorithms are selected to build models (for example, models adapted to different water depths are preliminarily integrated with models adapted to different terrain complexities to obtain a composite model, i.e., an interference signal recognition model), and measured multi-beam detection data adapted to the corresponding water depth and terrain complexity are selected, and interference signals in the measured multi-beam detection data are identified based on manual or machine annotation (i.e., training labels are formed), thereby forming training data. These training data are used to train the corresponding interference signal recognition model, and all trained interference signal recognition models are associated with the corresponding water depth and terrain complexity (i.e., the subsequent standard water depth and standard terrain complexity), and then put into the aforementioned model library together.

[0037] Then, based on the water depth and terrain complexity obtained in step S1, suitable interference signal recognition models are screened out from the pre-established model library, and these screened out models are called base models. All the screened out base models are integrated to form an integrated model. By targeted screening and integration of interference signal recognition models, the final integrated model can better adapt to the specific conditions of the target waters and improve the recognition capability of multi-beam measurement interference signals in the waters.

[0038] S3, using a multi-beam mode to perform high-resolution detection of the target waters, using the integrated model to identify interference signals from the second detection data obtained, attenuating or removing the identified interference signals, and obtaining third detection data.

[0039] In this step, the multi-beam mode is used to perform high-resolution detection of the target waters. At this time, the device will emit multiple beams at the same time to obtain a large amount of more detailed second detection data. Although the multi-beam mode can improve the measurement efficiency and accuracy, interference signals will be generated between the beams, affecting the measurement results. This is because the acoustic wave beam emitted by the transducer is not a completely ideal narrow beam, and there are certain side lobes and beam widths. When multiple beams work at the same time, the side lobes of adjacent beams may overlap, resulting in interference signals from other beams mixed into the received signal.

[0040] Therefore, the present invention further uses the integrated model constructed in step S2 to process the second detection data to identify the interference signals therein. Once the interference signals are identified, corresponding measures are taken to attenuate or directly remove them, thereby obtaining purified third detection data.

[0041] The multi-beam underwater measurement system of the present invention can be installed in a hoisting manner, that is, installation carriers (such as high towers) are set on both sides of the water area, and steel cables are laid between the two installation carriers. The multi-beam underwater measurement system is installed on the steel cables through the positioning host, and then the multi-beam underwater measurement system is driven to move on the steel cables through the transmission mechanism to achieve detection of different water areas. Reference stations can also be laid out at appropriate locations around the water area.

[0042] S4, determining the underwater topographic data of the target water area based on the processed third detection data.

[0043] In this step, the underwater topographic data of the target water area is determined based on the third detection data obtained after processing in step S3, for example, detailed information such as water depth and topographic undulations at different locations is calculated, and finally complete and accurate underwater topographic data of the target water area is formed to meet the needs of marine resource exploration, underwater engineering construction, marine scientific research and other fields for high-precision underwater topographic data.

[0044] The multi-beam underwater measurement scheme proposed in the present invention first uses a single beam to quickly determine the water depth and terrain characteristics of the target water area, thereby screening the base model and constructing a suitable integrated model to achieve accurate identification of interference signals with different characteristics, thereby solving the signal interference problem of multi-beams, allowing the advantages of multi-beam technology to be brought into play, ensuring the accuracy of the underwater terrain data obtained by detection, and providing accurate data for on-site decision-making in a timely manner.

[0045] Optionally, the first detection data includes distance data and position data, and the water depth and terrain complexity of the target water area are obtained based on the first detection data, including: calculating the average value of each distance data and using it as the water depth; performing three-dimensional projection on the corresponding distance data according to each position data to obtain a three-dimensional projection map, and using a classifier to classify the three-dimensional projection map to obtain the terrain complexity.

[0046] In this embodiment, after low-resolution detection of the target water area using a single-beam mode, first detection data including distance data and position data is obtained. The distance data refers to the distance between the transducer and the reflection point calculated based on the propagation time of the sound wave after the sound wave emitted by the single-beam transducer encounters an underwater object or the bottom of the water and is reflected back. Since the water depth at different measurement points may vary to a certain extent, the average value of each distance data is calculated and used to more accurately characterize the overall water depth of the target water area.

[0047] When performing low-resolution detection and obtaining each distance data, the underwater position corresponding to the distance data, i.e., the position data (such as longitude and latitude), can also be determined based on the direction corresponding to the single beam (i.e., the transmission direction angle). These position information are combined with the corresponding distance data and projected in three-dimensional space to obtain a three-dimensional projection map of the target waters, which intuitively shows the depth of different positions in the target waters.

[0048] Then, the classifier is used to classify the 3D projection map. According to the undulations and shapes of the terrain in the 3D projection map, the classifier divides the terrain complexity of the target water area into different categories, such as simple terrain (such as a flat bottom), medium-complex terrain (with some small undulations or obstacles), and complex terrain (such as the presence of high mountains and deep valleys), and then derives the terrain complexity of the target water area.

[0049] By performing three-dimensional projection and classification of the target waters, the terrain complexity of the target waters can be accurately assessed, providing an important basis for the subsequent construction of a suitable interference signal identification model, enabling the measurement method to better adapt to different terrain conditions and improve the accuracy and reliability of the measurement.

[0050] Next, taking the random forest classifier as an example, we will explain how to classify the three-dimensional projection map of the target water area to obtain the terrain complexity. The specific process is as follows: 1. Initialization: Initialize the experience replay pool D to store training related data.

[0051] Use randomly generated values ​​as the initial parameter values ​​of the current random forest classifier .

[0052] Set hyperparameters such as the number of decision trees n.

[0053] 2. Data preprocessing - feature extraction: For each three-dimensional projection image s∈S in the data set: calculate the height difference h: find the depth of the highest and lowest points in the projection image, and subtract them to get the height difference h.

[0054] Calculate the slope change p: Analyze the slopes at different positions of the projection map through an algorithm to obtain the slope change index p.

[0055] Calculate terrain roughness r: Use relevant algorithms to obtain the terrain roughness index r.

[0056] Group (h,p,r) into feature vector fs.

[0057] 3. Model training: The first layer of loop: initialize the training round number parameter epoch, and loop to the set total number of rounds E.

[0058] Observe the current training status, such as loss value, as the initial state .

[0059] The second loop: initialize the batch parameter batch and loop to the set total number of batches B.

[0060] A batch of projection images and corresponding labels are randomly selected from the dataset to form a training subset.

[0061] Train a random forest classifier based on the feature vector.

[0062] Make predictions on the training subset and get the predicted category .

[0063] According to the predicted category and the true label In contrast, the loss value loss is calculated as the reward value (the reward for correct prediction is high, and the reward for error is low).

[0064] Observe the state after training, such as the updated loss value, as the next state .

[0065] The four-tuple ( ,training action,loss, ) into the experience replay pool D.

[0066] Randomly select sample quadruplets from the experience replay pool for gradient descent and update the parameters of the current random forest classifier .

[0067] Determine whether the current batch value is a multiple of a specific value C. If so, perform additional optimization adjustments on the classifier.

[0068] The second cycle ends.

[0069] The first cycle ends.

[0070] 4. Classification prediction: 3D projection map of the target water area to be classified :According to the above feature extraction method, calculate the height difference, slope change, and terrain roughness to form a feature vector .

[0071] Will Input to the trained random forest classifier.

[0072] The classifier outputs the terrain complexity category (simple, medium or complex, or the corresponding numerical value) corresponding to the target projection map through internal decision tree calculation and voting mechanism.

[0073] Optionally, a number of interference signal recognition models are screened based on the water depth and the terrain complexity, each of the interference signal recognition models is used as a base model, and all the base models are integrated into an integrated model, including: matching and analyzing the water depth and the terrain complexity with the standard water depth and standard terrain complexity corresponding to each interference signal recognition model in the model library, to obtain a first number of interference signal recognition models that meet the matching threshold value, and use them as base models; and randomly screening a second number of interference signal recognition models that do not meet the matching threshold value, and also use them as base models; wherein the second number is lower than the first number; and integrating all the screened base models into an integrated model.

[0074] In this embodiment, a plurality of different interference signal recognition models are stored in the model library, and each model corresponds to parameters such as standard water depth and standard terrain complexity range. The water depth and terrain complexity of the target water area actually measured in step S1 are compared and analyzed with the standard water depth and standard terrain complexity corresponding to each interference signal recognition model in the model library. For example, model A is suitable for shallow water areas (standard water depth is 0-20 meters) and areas with relatively simple terrain (standard terrain complexity is low). If the water depth of the target water area is 15 meters and the terrain complexity is evaluated to be low, it meets the matching conditions of the model. A matching threshold value is set, and when the actual water depth and terrain complexity match the standard parameters of the model to a degree that reaches or exceeds this threshold value, the model is screened out. The number of models that meet the conditions is recorded as the first number, and these screened models are used as base models.

[0075] Through this matching method, the interference signal recognition models that are theoretically most suitable for the actual environmental conditions of the target waters can be preferentially selected, because the models that have a high degree of matching with the water depth and terrain complexity of the target waters are more likely to accurately identify the interference signals generated during multi-beam measurement in the waters.

[0076] Since the aforementioned step S1 roughly estimates the water depth and terrain complexity of the target water area through low-resolution detection, the actual water depth and terrain complexity of the local target water area may be quite different from the rough detection result, for example, the coefficient sampling may cross the underwater deep ditch area. Therefore, when performing multi-beam detection on the area with a large difference, the integrated model constructed only based on the first number of interference signal recognition models that meet the matching threshold value may not accurately identify the interference signal in the second detection data.

[0077] Therefore, in addition to the above-mentioned models that meet the matching threshold value, the present invention is also configured to randomly select a second number of models (the second number is lower than the first number) from the interference signal recognition models that do not meet the matching threshold value as base models. For example, if the first number is 10, the second number can be set to 3. Adding a small number of models that do not meet the matching threshold value to the base model set can increase the diversity of the model, thereby effectively dealing with the identification of interference signals in the second detection data corresponding to the aforementioned waters that are significantly different from the rough detection results, that is, making the integrated model have a stronger generalization ability, and can cope with some unexpected interference situations, thereby improving the reliability and comprehensiveness of the overall interference signal identification.

[0078] Finally, since a single interference signal recognition model may have limitations, by integrating multiple base models into an integrated model, the integrated model can analyze and judge the interference signal from multiple angles, reducing the risk of misjudgment or omission of a single model, thereby more effectively processing the interference signals generated in multi-beam measurements and improving the accuracy and reliability of the entire underwater measurement.

[0079] The first number of base models selected by matching analysis and the second number of base models randomly selected are all integrated together to form an integrated model. The integrated model can be constructed by, for example, a voting method, a weighted average method, etc., to synthesize the output results of multiple base models to obtain a more accurate interference signal identification result, which will not be described in detail. Of course, stacking methods, boosting methods, etc. can also be used, which will not be described in detail.

[0080] It should be noted that after obtaining multiple base models as mentioned above, these base models should be combined in multiple ways, and then each model combination should be tested and evaluated using test data to select the optimal model combination and integrate the multiple base models corresponding to the optimal model combination into an integrated model.

[0081] Optionally, the method of using a classifier to classify the three-dimensional projection image to obtain the terrain complexity includes: using a classifier to classify the three-dimensional projection image to obtain the terrain complexity and its classification confidence; then the second quantity is determined in the following manner: performing a comparative calculation based on the classification confidence and preset control data to obtain the second quantity; wherein, in the control data, the second quantity is negatively correlated with the classification confidence.

[0082] In this embodiment, when a classifier is used to classify the three-dimensional projection map of the target water area to determine the terrain complexity, the classifier not only gives a category of terrain complexity (such as simple, medium or complex), but also outputs a classification confidence associated with the classification result. The classification confidence reflects the degree of certainty of the classifier in its own judgment results.

[0083] There are many ways for classifiers to calculate confidence, such as probability-based methods. For a multi-classification problem, the classifier can output the probability of each category. The category with the highest probability is the classification result, and the probability value can be used as the classification confidence. For example, when the classifier judges that the complexity of a certain water area is "complex", it gives a confidence of 0.8, which means that the classifier is 80% sure that the judgment is correct. Generally speaking, the higher the confidence, the more reliable the classification result; the lower the confidence, the greater the uncertainty of the classification result.

[0084] Multiple groups of control data between the second quantity and the classification confidence are pre-constructed, and these control data show a negative correlation. The higher the classification confidence, the lower the second quantity calculated based on the control data, that is, the water depth and terrain complexity of the target water area determined above are relatively more accurate and reliable. At this time, a smaller number of interference signal recognition models that do not meet the matching threshold value are set to be added to avoid the integrated model being too large and to reduce the interference signal recognition results of these models from having too much impact on the final recognition result; conversely, the lower the classification confidence, the higher the second quantity, that is, a larger number of interference signal recognition models of the second quantity that do not meet the matching threshold value are set to be added to effectively deal with the accurate recognition of interference signals in the case of a large gap between the above-mentioned and rough detection results.

[0085] Optionally, before determining the underwater terrain data of the target waters based on the processed third detection data, the method also includes: determining the recognition confidence of the integrated model for the interference signal in the second detection data; if the recognition confidence is lower than the confidence threshold, adding an interference signal mark to the corresponding third detection data, and associating the recognition confidence of the interference signal and the corresponding second detection data with the third detection data.

[0086] In this embodiment, in the multi-beam underwater measurement method, after using the integrated model to identify the interference signal of the second detection data, it is necessary to determine the confidence of the integrated model in the identification result of the interference signal. The integrated model can calculate the identification confidence based on the voting situation of different base models, the probability value of the model output, etc. Taking the weighted voting method as an example, the calculation method of the identification confidence is explained: the weight corresponding to each base model is set to ,and .

[0087] Assume that the basic model set for judging the signal as an interference signal is S, then the recognition confidence For example, there are 3 base models with weights of Among them, base model 1 and base model 2 determine that the signal is an interference signal, and the recognition confidence C=0.4+0.3=0.7.

[0088] The obtained recognition confidence is compared with a preset confidence threshold. If the recognition confidence is lower than this threshold, it means that the recognition result of the integrated model for the interference signal is not very reliable and there is a large uncertainty. At this time, an interference signal mark is added to the corresponding third detection data. The function of this mark is to identify the interference signal that may not be accurately identified or accurately processed in the data. For example, this mark can be represented by a specific symbol, value or label so that these data with potential problems can be easily identified in the subsequent determination of underwater terrain data of the target waters or other processing processes. In addition to adding the interference signal mark, it is also necessary to associate the recognition confidence of the interference signal, the corresponding second detection data and the third detection data. This association can be achieved by establishing a data structure or a database table, storing these three data items (recognition confidence, second detection data, third detection data) together, and establishing a corresponding relationship between them.

[0089] With such a configuration, in the subsequent analysis or processing of the underwater topographic data of the target waters based on the third detection data, analysts can see the mark and retrieve and identify the confidence and original second detection data for further inspection or correction, so as to take more appropriate processing measures and improve the accuracy and reliability of the underwater topographic data measurement.

[0090] like Figure 2 , Figure 3 As shown, an embodiment of the present invention further discloses a multi-beam underwater measurement system, which includes an ultrasonic transmitting module 1, an ultrasonic receiving module 2, and a signal processing module 3.

[0091] The ultrasonic transmitting module 1 adopts a linear array transducer, which is composed of a plurality of ultrasonic transducer units.

[0092] The ultrasonic receiving module 2 includes a plurality of receiving transducers for receiving reflected ultrasonic signals.

[0093] The signal processing module 3 includes a filtering unit 301, an amplifying unit 302 and a digital signal processing unit 303; the filtering unit 301 is used to eliminate noise and extract effective signals; the amplifying unit 302 performs gain adjustment on the signal to ensure that the signal strength is moderate; the digital signal processing unit 303 processes the ultrasonic signals processed by the filtering unit 301 and the amplifying unit 302 to obtain underwater topographic data of the target water area; the filtering unit has a processor and a memory, and the processor is used to call and execute a computer program in the memory to implement a multi-beam underwater measurement method as described above.

[0094] Optionally, the system further includes a control module 4, which uses an embedded system (such as ARM or FPGA) to achieve high-precision timing control and data processing to coordinate the work of the ultrasonic transmitting module 1, the ultrasonic receiving module 2 and the signal processing module 3 to ensure the synchronization and accuracy of the measurement process.

[0095] An embodiment of the present invention further discloses an electronic device, comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor executes the computer program to implement the method described in the above embodiment.

[0096] An embodiment of the present invention further discloses a computer storage medium, wherein the computer storage medium stores a computer program, and the computer program is executed by a processor to implement the method described in the above embodiment.

[0097] An embodiment of the present invention further discloses a computer program product, which includes computer code. When the computer code is executed by a processor of an electronic device, the method described in the above embodiment is implemented.

[0098] The computer storage media mentioned above include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the above. Alternatively, the computer readable storage medium may be a machine readable signal medium. More specific examples of machine readable storage media may include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above.

[0099] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0100] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A multi-beam underwater measurement method, characterized in that: The method comprises the following steps: performing low-resolution detection on the target water area in a single-beam mode, and obtaining the water depth and terrain complexity of the target water area based on the first detection data obtained; screening and obtaining a plurality of interference signal recognition models based on the water depth and the terrain complexity, using each of the interference signal recognition models as a base model, and integrating all the base models into an integrated model; performing high-resolution detection on the target water area in a multi-beam mode, using the integrated model to identify the interference signal of the second detection data obtained, attenuating or removing the identified interference signal, and obtaining the third detection data; and determining the underwater terrain data of the target water area based on the processed third detection data.

2. A multi-beam underwater measurement method according to claim 1, characterized in that: The first detection data includes distance data and position data, and the water depth and terrain complexity of the target water area are obtained based on the first detection data, including: calculating the average value of each distance data and using it as the water depth; performing three-dimensional projection on the corresponding distance data according to each position data to obtain a three-dimensional projection map, and using a classifier to classify the three-dimensional projection map to obtain the terrain complexity.

3. A multi-beam underwater measurement method according to claim 2, characterized in that: Based on the water depth and the terrain complexity, several interference signal recognition models are obtained by screening, each of the interference signal recognition models is used as a base model, and all the base models are integrated into an integrated model, including: matching and analyzing the water depth and the terrain complexity with the standard water depth and standard terrain complexity corresponding to each interference signal recognition model in the model library, and obtaining a first number of interference signal recognition models that meet the matching threshold value, and using them as base models; and randomly screening a second number of interference signal recognition models that do not meet the matching threshold value, and also using them as base models; wherein the second number is lower than the first number; and integrating all the base models obtained by screening into an integrated model.

4. A multi-beam underwater measurement method according to claim 3, characterized in that: The method of using a classifier to classify the three-dimensional projection image to obtain the terrain complexity includes: using a classifier to classify the three-dimensional projection image to obtain the terrain complexity and its classification confidence; then the second quantity is determined in the following manner: performing a comparative calculation based on the classification confidence and preset control data to obtain the second quantity; wherein, in the control data, the second quantity is negatively correlated with the classification confidence.

5. A multi-beam underwater measurement method according to claim 1, characterized in that: Before determining the underwater terrain data of the target waters based on the processed third detection data, the method also includes: determining the recognition confidence of the integrated model for the interference signal in the second detection data; if the recognition confidence is lower than the confidence threshold, adding an interference signal mark to the corresponding third detection data, and associating the recognition confidence of the interference signal and the corresponding second detection data with the third detection data.

6. A multi-beam underwater measurement system, characterized in that: The system includes an ultrasonic transmitting module, an ultrasonic receiving module, and a signal processing module; the ultrasonic transmitting module adopts a linear array transducer, which is composed of a plurality of ultrasonic transducer units; the ultrasonic receiving module includes a plurality of receiving transducers for receiving reflected ultrasonic signals; the signal processing module includes a filtering unit, an amplifying unit, and a digital signal processing unit; the filtering unit is used to eliminate noise and extract effective signals; the amplifying unit performs gain adjustment on the signal to ensure that the signal strength is moderate; the digital signal processing unit processes the ultrasonic signals processed by the filtering unit and the amplifying unit to obtain underwater topographic data of the target waters; The filtering unit has a processor and a memory, and the processor is used to call and execute a computer program in the memory to implement a multi-beam underwater measurement method as described in any one of claims 1-5.

7. A multi-beam underwater measurement system according to claim 6, characterized in that: The system also includes a control module, which adopts an embedded system to achieve high-precision timing control and data processing to coordinate the work of the ultrasonic transmitting module, the ultrasonic receiving module and the signal processing module.

8. An electronic device comprising: At least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor executes the computer program to implement the method according to any one of claims 1 to 5.

9. A computer storage medium storing a computer program, characterized in that: The computer program is executed by a processor to implement the method according to any one of claims 1 to 5.

10. A computer program product, characterized in that: The computer program product includes computer codes, and when the computer codes are executed by a processor of an electronic device, the method according to any one of claims 1 to 5 is implemented.

Images