A biomimetic seal whisker type anti-hydrostatic pressure underwater multi-modal perception sensor

Abstract

The application discloses a bionic seal whisker type anti-hydrostatic pressure underwater multi-modal sensing sensor and belongs to the technical field of underwater robot environment sensing. The sensor comprises a non-sealed open shell, a substrate at the bottom of the shell, a flexible circuit board arranged on the substrate, and a magnetically sensitive elastomer layer formed integrally; the upper surface of the elastomer layer is provided with array-distributed through holes, and a downward ellipsoidal protrusion is formed below each hole on the lower surface. The sensor comprises a fluid sensing unit array corresponding to the through holes one by one, each unit is designed based on a bionic seal whisker touch rod structure, and the root of the bionic seal whisker touch rod is vertically fixed at the center of a cross cantilever beam. A magnetic field detection element is arranged on the flexible circuit board and covered by a support layer, the support layer is located below the ellipsoidal protrusion and in contact with the ellipsoidal protrusion, and the protrusion and the magnetic field detection element jointly constitute a touch sensing unit. The sensor balances the internal and external water pressure through the open structure, realizes multi-modal environment sensing, has high sensitivity and strong deep water adaptability.

Description

Technical Field

[0001] This invention relates to the field of underwater robot environmental perception technology, specifically to a biomimetic seal whisker-shaped anti-hydrostatic pressure underwater multimodal perception sensor. Background Technology

[0002] As ocean exploration expands into the deep sea, autonomous underwater vehicles (AUVs) and remotely operated vehicles (ROVs) place extremely high demands on environmental perception capabilities when performing complex tasks such as subsea pipeline inspection, biological capture, and deep-sea archaeology. Traditional underwater perception methods mainly rely on acoustics (sonar) and vision (cameras), forming the foundation of current underwater robot environmental perception technology. However, acoustics and vision have blind spots in murky waters, close-range operations, or confined spaces, making the development of more refined and instinctive perception methods an urgent need in this field.

[0003] In the prior art, several technical solutions have been developed in this field aimed at achieving fine near-field sensing. For example, underwater flow velocity sensing typically employs Doppler current meters (DVL), while tactile or pressure sensing relies on traditional sensors protected by sealed oil-filled structures or pressure-resistant housings. These prior art solutions attempt to simulate biological sensing capabilities in specific aspects, and their technical essence lies in protecting the core sensitive components from damage by the high-pressure underwater environment through physical encapsulation, thereby enabling the measurement of a single physical quantity (such as flow velocity or contact pressure).

[0004] The aforementioned existing technical solutions have significant drawbacks. First, in the deep-sea environment, the enormous hydrostatic pressure (an increase of one atmosphere for every 10 meters of descent) directly affects the sensor's sensitive elements, causing a significant drift in the reference signal. This makes it difficult for the sensor to distinguish between depth pressure and the contact force of an object, resulting in severe hydrostatic interference. Second, to withstand pressure, sensors are often made very thick and heavy, leading to large size, high power consumption, inability to be integrated into the end effector of a robotic arm, and reduced sensitivity. More importantly, existing sensors typically can only sense either flow velocity or contact force, failing to simultaneously detect subtle changes in water flow (wake tracking) and object contact detection.

[0005] Achieving multimodal fusion of hydrostatic pressure resistance, high-sensitivity flow field perception, and tactile perception in a miniaturized structure is a key technical challenge in this field. Summary of the Invention

[0006] To address the aforementioned issues, this invention proposes a biomimetic seal whisker-shaped underwater multimodal sensing sensor resistant to hydrostatic pressure. By eliminating hydrostatic interference through an open shell design, and integrating a fluid sensing unit array based on biomimetic whiskers with a magnetic tactile sensing unit into a micro-package, it achieves pressure resistance, multimodal fusion, and wideband sensing.

[0007] The technical solution adopted in this invention is as follows:

[0008] In a first aspect, the present invention proposes a biomimetic seal whisker-shaped underwater multimodal sensing sensor resistant to hydrostatic pressure, comprising:

[0009] The shell is open and not sealed. The side walls of the shell are provided with flow channels that allow external water to enter and exit freely, and the top is provided with an opening that corresponds to the position of the internal array.

[0010] The base is fixed to the bottom of the outer casing;

[0011] A flexible circuit board is disposed on the substrate;

[0012] The magnetically sensitive elastomer layer is placed inside the shell and is an integrally formed flat plate structure. The upper surface of the flat plate structure has an array of through holes, and the lower surface of the flat plate structure forms a downward ellipsoidal protrusion directly below each of the through holes.

[0013] A fluid sensing unit array is arranged in a one-to-one correspondence with the through holes. Each fluid sensing unit includes a biomimetic seal whisker-shaped contact rod, a cross cantilever beam, and a pressure-sensitive sensing element disposed on the cross cantilever beam. The cross cantilever beam is fixed in a through hole of the magnetic elastomer layer. The root of the biomimetic seal whisker-shaped contact rod is vertically fixed to the central intersection of the cross cantilever beam and extends outward through the corresponding opening at the top of the outer shell. The flow field information is calculated by detecting the deformation signal generated by the cross cantilever beam due to the impact of water flow.

[0014] A magnetic field detection element is disposed on a flexible circuit board and covered by a support layer. The support layer is located below and in contact with the ellipsoidal protrusion. The ellipsoidal protrusion and the magnetic field detection element together constitute a tactile sensing unit, which is used to convert the deformation of the magnetically sensitive elastomer caused by external contact pressure into a detectable magnetic field change signal.

[0015] Preferably, the biomimetic seal whisker rod is a columnar body extending along the axial direction, with an elliptical cross-section perpendicular to the axial direction. The ratio of the major axis to the minor axis of the ellipse is (4-4.5):1. Furthermore, the lengths of both the major and minor axes of the ellipse exhibit periodic sinusoidal changes along the axial direction, and there is a phase difference between the periodic changes of the major axis and the periodic changes of the minor axis.

[0016] Preferably, the cross cantilever beam includes a first beam arm and a second beam arm extending along the X-axis direction, and a third beam arm and a fourth beam arm extending along the Y-axis direction; pressure-sensitive sensing elements are respectively disposed on the surface of each beam arm and the four pressure-sensitive sensing elements on each cross cantilever beam are connected to form a Wheatstone full-bridge circuit for outputting a differential voltage signal characterizing the stress state of the cross cantilever beam.

[0017] Preferably, the magnetically sensitive elastomer layer is formed by mixing and solidifying a flexible polymer matrix material with hard magnetic particles uniformly dispersed within the matrix material; the magnetically sensitive elastomer layer generates an initial static magnetic field when not subjected to external force, and undergoes elastic deformation when subjected to pressure transmitted from the biomimetic seal whisker touch rod or direct contact pressure, thereby changing the spatial distribution density and magnetic moment direction of the internal magnetic particles.

[0018] Preferably, the magnetic field detection element is a miniature 3D Hall sensor.

[0019] Preferably, the length of the unsealed open housing is 4-6 mm, the width is 4-6 mm, and the height is 2-4 mm.

[0020] Secondly, this invention proposes a method for fabricating the aforementioned biomimetic seal whisker-shaped hydrostatic pressure-resistant underwater multimodal sensing sensor, comprising the following steps:

[0021] (1) Establish a three-dimensional digital model of the sensor and divide the model into a magnetically sensitive elastomer layer, a tough structure part and the remaining rigid support part according to the material properties; the tough structure part includes a cross cantilever beam and a biomimetic seal whisker touch rod;

[0022] (2) Prepare magnetic printing paste, tough resin paste and ordinary rigid resin paste. The magnetic printing paste is made by incorporating hard magnetic microparticles into elastic photosensitive resin and then performing vacuum degassing treatment. The mass fraction of magnetic powder in the hard magnetic microparticles is 10%-30%.

[0023] (3) The slurry prepared in step (2) is molded and manufactured using multi-material photopolymerization 3D printing technology;

[0024] (4) Integrate pressure-sensitive sensing elements on the surface of the cross-shaped cantilever beam and complete the circuit connection;

[0025] (5) The magnetically sensitive elastomer layer is subjected to pulse magnetization treatment and encapsulated with waterproof insulating glue to complete the assembly of the open shell and prepare a complete sensor structure.

[0026] Further, step (3) includes:

[0027] (3.1) Print the remaining rigid support parts, including the substrate, open shell and support layer, using a common rigid resin paste;

[0028] (3.2) Install and fix the flexible circuit board with the magnetic field detection element soldered on it in the reserved slot of the substrate;

[0029] (3.3) Switch to magnetic printing paste, and print and cure the magnetic elastomer layer on the flexible circuit board and magnetic field detection element in an integrated manner;

[0030] (3.4) Switch to tough resin slurry, and directly print and cure at the corresponding through holes of the magnetic elastomer layer to form a cross cantilever beam and a biomimetic seal whisker touch rod.

[0031] Thirdly, the present invention proposes a multimodal measurement method for the above-mentioned biomimetic seal whisker-shaped hydrostatic pressure-resistant underwater multimodal sensing sensor, including a fluid field measurement process and a tactile measurement process that can be executed independently or synchronously.

[0032] The fluid field measurement process is executed based on the fluid sensing unit array, including: synchronously acquiring multi-channel analog voltage signals output by pressure-sensitive sensing elements on each cross cantilever beam in the fluid sensing unit array, and performing analog-to-digital conversion; extracting dual-modal features from the converted time-series signals of each channel, wherein the dual-modal features include the static time-domain mean and dynamic vibration features of each channel time-series signal; and concatenating the dual-modal features to form a hybrid feature vector, which is then input into a pre-trained fluid field measurement model to predict the water flow velocity vector.

[0033] The tactile measurement process is executed based on the tactile sensing unit and includes: real-time acquisition of the three-dimensional orthogonal magnetic field components output by the magnetic field detection element, which characterize the changes in the synthetic magnetic field caused by the deformation of all ellipsoidal protrusions on the magnetically sensitive elastomer layer; inputting the time-series signal of the three-dimensional orthogonal magnetic field components into the pre-trained tactile measurement model to predict the three-dimensional net contact force vector and spatial orientation acting on the sensor.

[0034] Furthermore, the pre-trained fluid field measurement model is a multilayer perceptron model, and the training data comes from fluid field measurement sample data collected by the sensor under the impact of water flow with different flow velocities and directions in a controllable water tank; the fluid field measurement sample data includes multi-channel analog voltage signals and corresponding real water flow velocity vector data synchronously collected by each pressure-sensitive sensing element.

[0035] Furthermore, the pre-trained tactile measurement model is a convolutional neural network or a recurrent neural network model. The training data comes from tactile measurement sample data collected by calibration experiments on contact forces at different positions, sizes and directions on the sensor surface. The tactile measurement sample data includes the time-series signal of the three-dimensional orthogonal magnetic field components collected by the magnetic field detection element, as well as the real three-dimensional net contact force vector data and calibration information of the contact force application position.

[0036] The beneficial effects of this invention are:

[0037] This invention integrates an upper-layer fluid sensing unit array based on biomimetic whiskers with a lower-layer magnetic tactile sensing unit into a single miniature open-cell body, constructing a heterogeneous and complementary fluid-tactile physical sensing system. The open-cell design automatically balances internal and external hydrostatic pressure, completely eliminating static interference from deep-sea pressure on the sensitive elements and giving the sensor the potential for full-ocean-depth operation. Simultaneously, this integrated, array-based miniature package allows the sensor to be easily integrated into the end effector or skin of underwater robotic arms, laying the foundation for achieving high spatial resolution and precise environmental perception.

[0038] This invention fully utilizes the natural complementarity of two types of sensing elements in terms of frequency domain response and force modes. In the frequency domain, the upper layer primarily responds to low-frequency bending deformation caused by fluid drag, suitable for capturing steady-state flow fields; while the lower layer is extremely sensitive to fluid-induced high-frequency micro-tremors and turbulent pulsations, thus achieving full-band coverage from low-frequency steady-state flow to high-frequency turbulence at the physical level, overcoming the limitations of single-sensor principles in measuring over a wide flow velocity range. In terms of mechanical modes, this design achieves decoupling from the source of complex interference through physical structure: water flow mainly generates lateral shear force and bending moment on the beard (sensed by the upper layer), while solid contact or collision generates significant vertical pressure on the sensor substrate (sensed by the lower layer). By simultaneously monitoring and analyzing the response and proportional relationship of these two types of physical signals, the system can effectively distinguish between strong water flow impact and physical collision at the hardware source, achieving intelligent environmental adaptability with touch-and-sensitivity and perception-and-discrimination capabilities.

[0039] This invention achieves high sensitivity, strong anti-interference capability, wide bandwidth response, and multimodal information fusion capability on a miniaturized platform, providing a reliable and comprehensive environmental perception solution for underwater robots to perform precision operations. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the overall structure of a biomimetic seal whisker-shaped underwater multimodal sensing sensor resistant to hydrostatic pressure.

[0041] Figure 2 This is an exploded schematic diagram of a biomimetic seal whisker-shaped underwater multimodal sensing sensor resistant to hydrostatic pressure.

[0042] Figure 3 This is a cross-sectional view of a biomimetic seal whisker-shaped underwater multimodal sensing sensor resistant to hydrostatic pressure.

[0043] Figure 4 This is a structural diagram of an open-shell enclosure;

[0044] Figure 5 This is a structural diagram of the magnetically sensitive elastomer layer;

[0045] Figure 6 This is a structural diagram of the fluid sensing unit (pressure-sensitive sensing element is not shown).

[0046] Figure 7 This is a schematic diagram of the beard fluid sensing principle;

[0047] Figure 8 This is a schematic diagram of the sensing principle of a magnetic elastomer;

[0048] In the diagram: 1-Open shell, 2-Bionic seal whisker touch bar, 3-Pressure-sensitive sensing element, 4-Cross cantilever beam, 5-Magnetic-sensitive elastomer layer, 6-Support layer, 7-Magnetic field detection element, 8-Flexible circuit board, 9-Substrate, 10-Flow channel, 11-Through hole, 12-Ellipsoidal protrusion. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can also refer to the internal connection of two components. "A plurality of" or "several" means two or more. Those skilled in the art can understand the specific meaning of the above terms in this invention in light of the specific circumstances.

[0051] This invention provides a biomimetic seal whisker-shaped underwater multimodal sensing sensor resistant to hydrostatic pressure. For example... Figure 1 , Figure 2 and Figure 3 As shown, the sensor adopts a miniaturized and arrayed design, uses an unsealed open structure to adapt to deep-sea hydrostatic pressure, and integrates a biomimetic fluid sensing and magnetic tactile sensing dual-modal unit in a miniaturized package to achieve synchronous and highly sensitive measurement of underwater flow field and contact force.

[0052] Specifically, the sensor's overall structure is in the form of a miniature, flat cube. Its exterior is a non-sealed, open housing 1, as shown... Figure 4As shown, the outer casing has multiple uniformly distributed flow channels 10 on its sidewalls, and an opening at the top corresponding to the position of the internal array. When the sensor is placed underwater, external water can freely enter and exit the sensor's internal cavity through these channels, thereby instantly balancing the internal and external hydrostatic pressure and eliminating interference from water depth on the sensitive element. In a preferred embodiment, the outer casing is approximately 4.9 mm long, 4.9 mm wide, and 2.82 mm high, with the flow channels 10 approximately 0.4 mm wide, achieving a high degree of integration and miniaturization. A base 9 is fixed to the bottom of the outer casing as a supporting foundation. A flexible circuit board 8 is mounted on the base 9 for laying out circuits and installing electronic components.

[0053] The core sensing structure layer of the sensor is a single-piece molded magnetically sensitive elastomer layer 5, such as... Figure 5 As shown, the entire structure is flat. The magnetically sensitive elastomer layer 5 is formed by mixing and solidifying a flexible polymer matrix material with uniformly dispersed hard magnetic microparticles (such as neodymium iron boron microparticles), thereby enabling it to generate an initial static magnetic field. Multiple through holes 11 are arranged in a 3×3 array on the upper surface of the flat plate. Specifically, on the lower surface of the flat plate, directly below each through hole 11, a downward-facing, convex ellipsoidal protrusion 12 is integrally formed.

[0054] The sensor's fluid sensing function is achieved through an array of fluid sensing units, each corresponding to one of the aforementioned through holes 11. For example... Figure 6 As shown, each fluid sensing unit mainly includes a biomimetic seal whisker-shaped contact rod 2, a cross-shaped cantilever beam 4, and a pressure-sensitive sensing element 3. The biomimetic seal whisker-shaped contact rod 2 is a columnar body extending axially, with an elliptical cross-section perpendicular to the axial direction. The ratio of the major axis to the minor axis of this ellipse is designed within the range of (4-4.5):1 to mimic the hydrodynamic characteristics of a seal whisker. More importantly, the lengths of both the major and minor axes of this elliptical cross-section exhibit periodic sinusoidal changes along the axial direction of the contact rod, and there is a specific phase difference between the period of change of the major axis and the period of change of the minor axis, preferably 180 degrees. This unique configuration effectively suppresses its own vortex-induced vibration and improves the signal-to-noise ratio. The cross-shaped cantilever beam 4 is fixed within a through hole 11 in the magnetically sensitive elastomer layer 5, and includes a first and second beam arm extending along the X-axis, and a third and fourth beam arm extending along the Y-axis. The root of the biomimetic seal whisker-shaped contact rod 2 is vertically fixed at the central intersection of the cross cantilever beam 4 and extends outward through the corresponding opening at the top of the outer casing. Pressure-sensitive sensing elements 3 are respectively disposed on the surface of each beam arm, and the four pressure-sensitive sensing elements 3 on each cross cantilever beam 4 are connected to form a Wheatstone full-bridge circuit, used to output a differential voltage signal characterizing the bending state of the cross cantilever beam 4 under stress in the flow field. Figure 7As shown, during operation, the impact of water flow on the biomimetic seal whisker contact rod 2 causes the cross cantilever beam 4 to bend and deform. By calculating the change in the output signal of the Wheatstone bridge, the flow field information can be perceived. In a preferred embodiment, the pressure-sensitive sensing element 3 is a pressure-sensitive resistor. The water flow velocity vector can be obtained through a pre-established mapping model between the characteristics of multiple voltage signals and the water flow velocity vector.

[0055] The tactile sensing function of the sensor is accomplished by a unit composed of ellipsoidal protrusions 12 on the magnetic elastomer layer 5 and the corresponding magnetic field detection element 7. The magnetic field detection element 7 is mounted on the flexible circuit board 8 and covered and fixed by the upper support layer 6; the ellipsoidal protrusions 12 on the lower surface of the magnetic elastomer layer 5 face the magnetic field detection element 7 and are in contact with it or maintain a small gap (contact is possible under slight force). In this embodiment, a 3D Hall sensor is integrated at the geometric center directly below the array of nine elastomer protrusions embedded with mixed magnetic particles. This Hall sensor has a built-in temperature compensation mechanism to overcome the performance drift of the sensitive element caused by changes in ambient temperature, thereby ensuring the accuracy and stability of magnetic field measurement throughout the entire operating temperature range; this sensor can be implemented using existing technology, the principle of which will not be elaborated in this invention. Figure 8 As shown, when the sensor is subjected to axial pressure transmitted from the biomimetic seal whisker-shaped contact rod 2 or direct contact pressure from an external object, the force is transmitted through the protrusions, causing one or more protrusions in the magnetically sensitive elastomer to undergo local compression or shear deformation. This deformation alters the spatial distribution density and magnetic moment direction of the internal hard magnetic particles, causing the magnetic field vectors generated by the nine magnetic particles to superimpose in space, thereby modulating the spatial magnetic field vector and resulting in a change in the composite magnetic flux density vector within the sensitive area of ​​the Hall sensor at the center. The 3D Hall sensor located directly below can detect the changes in the components of this magnetic field in the three orthogonal directions of X, Y, and Z in real time. Through a pre-established mapping model between the three-dimensional orthogonal magnetic field component signals and the contact pressure, the three-dimensional net contact force vector acting on the sensor can be output, while simultaneously classifying and identifying the coarse spatial orientation of the contact. This architecture utilizes a single sensor to achieve centralized acquisition of the contact state of the entire array area, and senses the overall force situation by monitoring changes in the composite magnetic field vector.

[0056] The aforementioned biomimetic seal whisker-shaped hydrostatic pressure-resistant underwater multimodal sensing sensor was fabricated using multi-material photopolymerization 3D printing technology based on digital models. This technology enables integrated, high-precision molding of the rigid support, flexible magnetic sensing functional layer, and tough sensing structure, thereby ensuring the sensor's miniaturization, high consistency, and excellent hydrostatic pressure resistance.

[0057] In one specific embodiment of the present invention, the specific steps are as follows:

[0058] S1. Establish a three-dimensional digital model of the sensor, and based on the material properties and functions of each component, divide the model into three main parts in the software: magnetic elastomer layer 5, tough structure part (including cross cantilever beam 4 and biomimetic seal whisker touch rod 2) and other rigid support parts (including open shell 1, base 9 and support layer 6).

[0059] S2, prepares three different printing pastes with different functions.

[0060] Ordinary rigid resin slurry is used to mold the shell, base, and other structures to ensure the stability of the overall frame.

[0061] Tough resin slurry is used to manufacture the cross cantilever beam 4 and the biomimetic seal whisker touch rod 2, which need to withstand repeated bending deformation.

[0062] The magnetic printing paste is used to manufacture the magnetic elastomer layer 5. First, neodymium iron boron (NdFeB) magnetic powder with an average particle size of 5-20 micrometers is uniformly mixed into the elastic photosensitive resin (Shore hardness A 40) at a mass fraction of 20 wt%. Before incorporation, the magnetic powder is usually surface-treated with a silane coupling agent to improve its compatibility and dispersibility with the resin matrix. During preparation, the paste is thoroughly mixed in a planetary mixer and subjected to vacuum degassing to obtain a homogeneous, bubble-free paste.

[0063] S3 employs a multi-material photopolymer 3D printing system for sequential execution:

[0064] The first step involves printing the remaining rigid support components that form the sensor skeleton using a common rigid resin paste. These include the bottom substrate 9, the open housing frame 1 and the flow channels 10 on its side walls, and the built-in support layer 6 structure for covering and fixing the circuitry. Flexible circuit boards 8, pre-soldered with magnetic field detection elements 7 (i.e., miniature 3D Hall sensors), are manually installed or fixed in the precision slots reserved in the printed substrate 9.

[0065] The second step involves switching the printing tank to one filled with magnetic printing paste. Integrated printing and photopolymerization are then performed on the substrate 9, above which the circuit board has been mounted, directly forming a complete magnetic elastomer layer 5. This process molds the complex three-dimensional structure of this layer in a single step: including a flat plate with an array of through-holes 11 on its upper surface and an ellipsoidal structure protruding downwards directly below each hole. During printing, it is crucial to ensure good bonding between the elastomer and the underlying circuit board, support layer 6, and the surrounding rigid shell.

[0066] The third step involves switching the printing material back to a tough resin slurry. On the upper surface of the cured magnetic elastomer layer 5, a cross-shaped cantilever beam 4 and its connected biomimetic whisker-like appendages 2 are precisely printed and cured directly at each through-hole 11. Utilizing the chemical bonding of the resin monomers during the photocuring process, a strong interface bond is formed between the root of the tough structure and the underlying magnetic elastomer layer 5 plate, eliminating the need for additional adhesives. This ensures an efficient force transmission path from the whiskers and cantilever beam to the magnetic elastomer.

[0067] S4, After the main structure printing is completed, subsequent integration and processing are required:

[0068] Pressure-sensitive sensing elements 3 are integrated on the surface of each beam arm of the cross-shaped cantilever beam 4 and connected to form a Wheatstone full-bridge circuit, which is then connected to the corresponding lines on the flexible circuit board 8 via leads. Next, the magnetically sensitive elastomer layer 5 in the overall sensor undergoes pulsed magnetization, typically in a pulsed magnetic field with an intensity of 1.5T or higher, to ensure that the isotropic magnetic powder acquires consistent permanent magnet characteristics, thereby generating an initial static magnetic field. Finally, waterproof insulating adhesive is used to reliably encapsulate all circuit connection points, interfaces, and housing seams, completing the assembly of the entire sensor and producing a structurally complete and fully functional sensor product.

[0069] Based on this sensor, this invention proposes a measurement method based on a deep neural network, which can execute fluid field measurement process and tactile measurement process independently or synchronously.

[0070] The fluid field measurement process is executed based on the fluid sensing unit array to achieve wide-domain flow field calculation from low-speed laminar flow to high-speed turbulent flow. Specifically, the system first performs a bridge network on 36 piezoresistors at the roots of 9 whiskers in a 3×3 array to acquire 18 analog voltage signals characterizing the lateral and longitudinal forces on each whisker, and performs synchronous digital sampling. To take into account the fluid dynamics characteristics under different flow velocities, the processor performs dual-modal feature extraction on the original time-series signals within the sliding window: on the one hand, it calculates the time-domain mean of each channel signal, which mainly reflects the steady-state drag force and bending moment generated by the water flow on the elliptical cross-section whiskers under low-speed flow fields; on the other hand, it calculates the root mean square (RMS) or standard deviation of each channel signal within the sliding time window (as a dynamic vibration feature), which mainly reflects the energy intensity of fluid-induced vibration (FIV) and vortex-induced vibration (VIV) under high-speed flow fields. The above 18 static mean features and 18 dynamic vibration features are concatenated to construct a hybrid feature input vector with a dimension of 36. Subsequently, this feature vector is input into a pre-trained fluid field measurement model for inference. This model, trained using extensive flue experiment data, establishes a nonlinear mapping relationship from resistance-voltage characteristics to velocity vectors. It not only adaptively fits the nonlinear characteristics of the drag coefficient of the seal whisker elliptical cross-section varying with the angle of attack, but more importantly, it utilizes the spatial distribution information of a 3×3 array and learns the gradient changes in signal intensity within the array to automatically compensate for the wake occlusion and flow field coupling interference caused by upstream whiskers on downstream whiskers. Finally, the model outputs the orthogonal components of the velocity vector in the sensor coordinate system, thereby calculating the precise magnitude and direction of the water flow velocity, achieving high-precision end-to-end measurement of nonlinear, strongly coupled, and complex flow field environments.

[0071] In this embodiment, the pre-trained fluid field measurement model is a multilayer perceptron network. The training data comes from the fluid field measurement sample data collected by the sensor under the impact of water flow with different velocities and directions in a controllable water tank. The fluid field measurement sample data includes multi-channel analog voltage signals synchronously collected by each pressure-sensitive sensing element 3 when the sensor is impacted by water flow with different velocities and directions in the controllable water tank, as well as the real water flow velocity vector data corresponding to each signal, which is measured by a high-precision calibration instrument.

[0072] The tactile measurement process is executed based on the tactile sensing unit and includes: real-time acquisition of three-dimensional orthogonal magnetic field components output by the magnetic field detection element 7, which characterize the changes in the synthetic magnetic field caused by the deformation of all ellipsoidal protrusions 12 on the magnetically sensitive elastomer layer 5; inputting the time-series signal of the three-dimensional orthogonal magnetic field components into a pre-trained tactile measurement model, which infers and outputs the three-dimensional net contact force vector acting on the sensor based on the signal, and simultaneously classifies and identifies the coarse spatial orientation of the contact. Under this design, the model's input only needs to receive the three orthogonal component signals output by a single Hall sensor; in order to extract effective information from the dimensionality-reduced mixed signal, the model uses the time-series signal of the three-dimensional orthogonal magnetic field components to identify the dynamic pattern of magnetic field changes.

[0073] In this embodiment, the pre-trained tactile measurement model is a convolutional neural network or a recurrent neural network. The training data comes from tactile measurement sample data collected during calibration experiments on contact forces of different positions, sizes, and directions on the sensor surface. The tactile measurement sample data includes the time-series signals of the three-dimensional orthogonal magnetic field components collected by the magnetic field detection element 7 when a three-dimensional contact force of known size and direction is applied at different positions on the sensor surface; the real three-dimensional net contact force vector data measured by the force sensor corresponding to each set of magnetic field signals; and the calibration information of the contact force application position. The model is trained with a large amount of experimental data to learn the mapping law of different positions and sizes of pressure on the central magnetic field, thereby achieving two main functions: first, resultant force calculation, that is, outputting the three-dimensional net contact force vector acting on the entire array; second, orientation classification, based on the deflection characteristics of the magnetic field vector, performing probability classification on the coarse area where contact occurs (such as the left, right, or center of the array), thereby achieving spatially resolvable tactile perception with minimal hardware cost.

[0074] This invention integrates the upper fluid sensing unit and the lower tactile sensing unit into the same sensor body, constructing a flow-tactile heterogeneous complementary physical sensing system. It fully utilizes the natural complementarity of the two types of sensitive elements in frequency domain response and force mode: On the one hand, by utilizing the frequency domain complementarity of low-frequency flow field and high-frequency micro-vibration, the pressure-sensitive sensing element 3 of the upper fluid sensing unit mainly responds to low-frequency bending deformation caused by fluid drag force, which is suitable for capturing steady-state flow velocity and large-scale eddies; while the magnetically sensitive elastomer protrusion structure of the lower tactile sensing unit, with its high elastic modulus and the high-frequency response characteristics of the magnetic field detection element 7, has extremely high mechanical sensitivity to fluid-induced high-frequency micro-vibration and transient pressure fluctuations in the turbulent boundary layer. Thus, at the physical level, it achieves full-band coverage of low-frequency steady-state flow and high-frequency turbulence, effectively solving the physical bottleneck of response lag in high-frequency turbulence measurement by a single resistance strain gauge. On the other hand, by utilizing the modal complementarity of shear moment and vertical pressure, this invention addresses the challenge of a single sensor failing to distinguish between strong water flow impact and physical collision in complex underwater environments. It decouples the components through the difference in force exerted on the physical structure: water flow primarily generates lateral shear force and bending moment on the whiskers (sensed by the upper fluid sensing unit), while solid contact or collision generates significant vertical pressure and impact force on the base (sensed by the lower tactile sensing unit). By monitoring the proportional relationship between the responses of these two physical quantities, the system can effectively eliminate collision interference at the hardware level, achieving intelligent environmental adaptability with the ability to sense and distinguish between touch and objects.

[0075] The above-described embodiments are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. Those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

Claims

1. A multimodal measurement method for a biomimetic seal whisker-shaped underwater multimodal sensing sensor resistant to hydrostatic pressure, characterized in that the sensor... include: The unsealed open shell (1) has a flow channel (10) on the side wall that allows external water to enter and exit freely, and an opening on the top that corresponds to the position of the internal array; The base (9) is fixed to the bottom of the outer shell; Flexible circuit board (8) is disposed on the substrate; The magnetically sensitive elastomer layer (5) is placed inside the shell and is an integrally formed flat plate structure. The upper surface of the flat plate structure has an array of through holes (11), and the lower surface of the flat plate structure forms a downward ellipsoidal protrusion (12) directly below each through hole. The fluid sensing unit array is set up one-to-one with the through hole. Each fluid sensing unit includes a biomimetic seal whisker touch rod (2), a cross cantilever beam (4), and a pressure-sensitive sensing element (3) set on the cross cantilever beam. The cross cantilever beam is fixed in a through hole of the magnetic elastomer layer. The root of the biomimetic seal whisker touch rod is vertically fixed at the center intersection of the cross cantilever beam and extends outward through the corresponding opening at the top of the shell. The flow field information is calculated by detecting the deformation signal generated by the cross cantilever beam due to the impact of water flow. The magnetic field detection element (7) is disposed on the flexible circuit board and covered by the support layer (6). The support layer is located below the ellipsoidal protrusion and in contact with the ellipsoidal protrusion. The ellipsoidal protrusion and the magnetic field detection element together constitute a tactile sensing unit, which is used to convert the deformation of the magnetic elastic body caused by external contact pressure into a detectable magnetic field change signal. The multimodal measurement method includes a fluid field measurement process and a tactile measurement process that can be executed synchronously; The fluid field measurement process is executed based on the fluid sensing unit array, including: synchronously acquiring multi-channel analog voltage signals output by pressure-sensitive sensing elements on each cross cantilever beam in the fluid sensing unit array, and performing analog-to-digital conversion; extracting dual-modal features from the converted time-series signals of each channel, wherein the dual-modal features include the static time-domain mean and dynamic vibration features of each channel time-series signal; and concatenating the dual-modal features to form a hybrid feature vector, which is then input into a pre-trained fluid field measurement model to predict the water flow velocity vector. The tactile measurement process is executed based on the tactile sensing unit and includes: real-time acquisition of the three-dimensional orthogonal magnetic field components output by the magnetic field detection element, which characterize the changes in the synthetic magnetic field caused by the deformation of all ellipsoidal protrusions on the magnetically sensitive elastomer layer; inputting the time-series signal of the three-dimensional orthogonal magnetic field components into the pre-trained tactile measurement model to predict the three-dimensional net contact force vector and spatial orientation acting on the sensor.

2. The multimodal measurement method of the biomimetic seal whisker-type hydrostatic pressure-resistant underwater multimodal sensing sensor according to claim 1, characterized in that, The biomimetic seal whisker-shaped stylus is a columnar body extending along the axial direction, with an elliptical cross-section perpendicular to the axial direction. The ratio of the major axis to the minor axis of the ellipse is (4-4.5):

1. Furthermore, the lengths of both the major and minor axes of the ellipse exhibit periodic sinusoidal changes along the axial direction, and there is a phase difference between the periodic changes of the major and minor axes.

3. The multimodal measurement method of the biomimetic seal whisker-type hydrostatic pressure-resistant underwater multimodal sensing sensor according to claim 1, characterized in that, The cross cantilever beam includes a first beam arm and a second beam arm extending along the X-axis, and a third beam arm and a fourth beam arm extending along the Y-axis; pressure-sensitive sensing elements are respectively disposed on the surface of each beam arm, and the four pressure-sensitive sensing elements on each cross cantilever beam are connected to form a Wheatstone full-bridge circuit for outputting a differential voltage signal characterizing the stress state of the cross cantilever beam.

4. The multimodal measurement method of the biomimetic seal whisker-type hydrostatic pressure-resistant underwater multimodal sensing sensor according to claim 1, characterized in that, The magnetically sensitive elastomer layer is formed by mixing and solidifying a flexible polymer matrix material with hard magnetic particles uniformly dispersed within the matrix material. When the magnetically sensitive elastomer layer is not subjected to external force, it generates an initial static magnetic field. When subjected to pressure transmitted from the biomimetic seal whisker touch rod or direct contact pressure, it undergoes elastic deformation, changing the spatial distribution density and magnetic moment direction of the internal magnetic particles.

5. The multimodal measurement method of the biomimetic seal whisker-type hydrostatic pressure-resistant underwater multimodal sensing sensor according to claim 4, characterized in that, The magnetic field detection element is a miniature 3D Hall sensor.

6. The multimodal measurement method of the biomimetic seal whisker-type hydrostatic pressure-resistant underwater multimodal sensing sensor according to claim 1, characterized in that, The pre-trained fluid field measurement model is a multilayer perceptron model. The training data comes from fluid field measurement sample data collected by the sensor under the impact of water flow with different velocities and directions in a controllable water tank. The fluid field measurement sample data includes multi-channel analog voltage signals and corresponding real water flow velocity vector data synchronously collected by each pressure-sensitive sensing element.

7. The multimodal measurement method of the biomimetic seal whisker-type hydrostatic pressure-resistant underwater multimodal sensing sensor according to claim 1, characterized in that, The pre-trained tactile measurement model is a convolutional neural network or a recurrent neural network model. The training data comes from tactile measurement sample data collected from calibration experiments of contact forces at different positions, sizes and directions on the sensor surface. The tactile measurement sample data includes the time-series signal of the three-dimensional orthogonal magnetic field components collected by the magnetic field detection element, as well as the real three-dimensional net contact force vector data and calibration information of the contact force application position.

Images