Road sound barrier system and noise reduction method thereof

Abstract

The invention discloses a road sound barrier system and a noise reduction method thereof, and the system comprises an advanced sensing module which is used for obtaining the information of a traffic flow which is about to drive into a target road barrier region in a future set time period in real time; the digital twinning and rehearsing module is internally provided with a digital twinning body corresponding to the physical sound barrier and is used for receiving the traffic flow information and rehearsing a noise field and effects of different adjustment strategies; the intelligent decision-making module integrates a reinforcement learning decision-making model and is used for selecting an optimal noise reduction strategy based on a rehearsal result; and the dynamic execution module is used for receiving an instruction of the optimal noise reduction strategy and adjusting the working state of the physical sound barrier before the traffic flow arrives. According to the method, not only can the current optimal decision be made, but also the influence of the decision on the service life and long-term operation cost of the equipment can be considered, the maximization of comprehensive benefits of the system is realized, and a control method depending on artificial experience or simple rules is exceeded.

Description

Technical Field

[0001] This invention relates to the technical field of road noise reduction, and more particularly to a road noise barrier system and its noise reduction method. Background Technology

[0002] Road noise barriers are key infrastructure for mitigating traffic noise pollution and protecting the acoustic environment of sensitive areas such as residential areas, schools, and hospitals along roads. Existing road noise barriers mainly follow the principles of passive sound insulation and absorption, and their technological development focuses on optimizing materials, structure, and form.

[0003] 1. Traditional static sound barriers and their limitations The most common sound barriers use a structure combining sound-insulating panels (such as metal plates or transparent PC panels) with porous sound-absorbing materials (such as rock wool or glass wool). Once installed, the acoustic performance (such as sound insulation and sound absorption frequency band) and physical form of this type of technical solution remain fixed. Its inherent drawback is that traffic noise is highly dynamic, with its intensity and spectral characteristics fluctuating in real time with traffic flow, vehicle type, and speed. A barrier with fixed performance cannot achieve dynamic matching, resulting in over-performance at low traffic volumes and insufficient protection during high traffic volumes or when special vehicles (such as heavy trucks) pass. Furthermore, its design height and length are determined based on predictive models, making it difficult to adapt to long-term changes in the surrounding environment or traffic flow after construction.

[0004] 2. Improvements and shortcomings of existing smart sound barriers To overcome the shortcomings of static barriers, existing technologies have proposed several intelligent improvement schemes, mainly including: Responsive sound barriers: These barriers monitor real-time noise levels by installing noise sensors and switch between a limited number of operating modes (such as turning on / off additional sound-absorbing structures) based on preset thresholds. This type of solution only reacts to noise that has already arrived, resulting in significant response lag. Furthermore, its simple control logic cannot handle complex spatiotemporal dynamic noise fields.

[0005] Variable structure sound barriers: Some mechanical structures with adjustable louver angles or panel positions have been disclosed in order to change acoustic performance. However, their adjustment strategies mostly rely on simple rule control, lacking precise and forward-looking calculations and optimizations for the overall noise reduction effect. The adjustment is largely blind and often ignores the energy consumption and mechanical losses of the adjustment action itself.

[0006] Simulation-based design optimization: During the barrier design phase, acoustic simulation software (such as Raynoise and SoundPLAN) is used to optimize the shape and position. However, this is a purely offline, one-time design tool and cannot continuously optimize the operation strategy based on real-time data during the barrier operation phase.

[0007] In summary, the core dilemma of existing technologies lies in the contradiction between "static performance" and "dynamic noise," and the disconnect between "passive response" and "active optimization." Sound barrier systems generally lack the ability to proactively perceive traffic flow conditions, accurately predict future noise fields, and make multi-objective collaborative intelligent decisions based on this information. Consequently, their noise reduction efficiency, energy efficiency, and long-term adaptability are not optimal. Summary of the Invention

[0008] The purpose of this section is to outline some aspects of embodiments of the present invention and to briefly describe some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of this application, to avoid obscuring the purpose of these documents; however, such simplifications or omissions should not be construed as limiting the scope of the invention.

[0009] In view of the problems existing in the current road noise barrier system and its noise reduction method, the present invention is proposed.

[0010] Therefore, the purpose of this invention is to provide a road noise barrier system and its noise reduction method, which can not only make the current optimal decision, but also consider the impact of the decision on the equipment life and long-term operating costs, thereby maximizing the overall benefits of the system and surpassing control methods that rely on human experience or simple rules.

[0011] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a road noise barrier system, comprising: The advanced perception module is used to acquire real-time information on traffic flow that will enter the target road barrier area within a set time period in the future; The digital twin and pre-visualization module has a built-in digital twin corresponding to the physical sound barrier, which is used to receive traffic flow information and pre-visualize the noise field and the effects of different adjustment strategies. The intelligent decision-making module integrates a reinforcement learning decision-making model to select the optimal noise reduction strategy based on the pre-simulation results; The dynamic execution module is used to receive instructions from the optimal noise reduction strategy and adjust the working state of the physical sound barrier before the traffic flow arrives.

[0012] In a preferred embodiment of the road noise barrier system described in this invention, the digital twin is constructed in the following manner: Establish a high-precision three-dimensional geometric model of the target road barrier area and its surrounding environment; The positions and real-time status parameters of all adjustable units of the physical sound barrier are mapped in the three-dimensional geometric model. A parameterized vehicle noise source model and a sound wave propagation calculation model are embedded, wherein the vehicle noise source model maps different noise spectrum characteristics according to the vehicle type.

[0013] As a preferred embodiment of the road sound barrier system described in this invention, the reinforcement learning decision model employs a deep deterministic policy gradient, including: Actor Network: Enter current status Output a continuous action vector ; Critics Network: Input Status and actions Assess its expected long-term cumulative rewards; Training environment and process: Using a digital twin as the simulation environment: The training of the model is carried out entirely in a high-fidelity virtual environment constructed from the digital twin; Interaction and Iteration: The intelligent agent (actor network) observes the state of the environment. ; The agent outputs actions ; Digital twin receiving Simulate execution and calculate the state at the next moment. and instant rewards ; This interactive data Store in the experience replay buffer; The critic network uses data in the buffer and updates it through temporal difference learning to more accurately evaluate the value of actions; The actor network updates using a policy gradient method based on critics' guidance to output actions that will receive higher ratings; Training objective: Through millions of such virtual interaction iterations, the actor network will eventually learn a policy function. It can adapt to any complex state This directly maps to rewards that can accumulate over a long period of time. , The optimal action is to maximize the discount factor. .

[0014] As a preferred embodiment of the road noise barrier system described in this invention, the dynamic execution module specifically includes: Multiple distributed adjustable units, each unit integrating an acoustic impedance adjustable surface and a geometrically deformable structure; the acoustic impedance adjustable surface is composed of a micro-perforated plate array driven by piezoelectric ceramics, and the effective aperture of the micro-holes is adjusted by changing the driving voltage; the geometrically deformable structure is composed of a composite material panel with embedded shape memory alloy driving wires, and the panel is induced to produce preset bending or torsional deformation by controlling the current on and off of the driving wires. The local controller receives instructions from the intelligent decision-making module, parses them, and drives the corresponding adjustable unit to perform state adjustments.

[0015] A noise reduction method for road sound barriers includes the following steps: S1: Real-time acquisition of traffic flow information that will enter the target road barrier area within a future set time period, the traffic flow information including at least vehicle type, vehicle speed, vehicle position and estimated arrival time; S2: Input the traffic flow information into a digital twin associated with the target road barrier area, and pre-simulate the three-dimensional dynamic noise field generated by the traffic flow in the digital twin, and simultaneously pre-simulate the noise reduction effect under at least two different sound barrier adjustment strategies; S3: Based on the pre-simulation results, the optimal noise reduction strategy is selected from the at least two adjustment strategies using a reinforcement learning decision model. The reinforcement learning decision model is trained and makes decisions with the goal of maximizing the long-term comprehensive noise reduction benefits. S4: Before the traffic flow actually reaches the target road barrier area, adjust the working state of the sound barrier in advance according to the optimal noise reduction strategy.

[0016] As a preferred embodiment of the noise reduction method for a road sound barrier according to the present invention, in step S1, the traffic flow information is directly obtained from the real-time reported data of connected vehicles through the vehicle-road cooperative communication interface, or is collected by a radar-visual fusion sensing device deployed upstream of the target road barrier area.

[0017] As a preferred embodiment of the noise reduction method for a road sound barrier according to the present invention, wherein: the adjustment of the working state of the sound barrier in step S4 includes at least one of the following methods: adjusting the acoustic impedance distribution of the variable micro-perforated plate array of the sound barrier; driving the variable topology structure unit of the sound barrier to generate a preset deformation; switching the resonant frequency of the local acoustic metamaterial unit.

[0018] As a preferred embodiment of the noise reduction method for road sound barriers according to the present invention, after performing step S4, a feedback and learning step S5 is further included: collecting actual noise reduction effect data through noise monitoring terminals deployed in the protected area, comparing the actual noise reduction effect data with the corresponding pre-simulation results in step S2, generating feedback data, and using the feedback data to update and correct the prediction accuracy of the digital twin and the decision strategy of the reinforcement learning decision model online.

[0019] The beneficial effects of this invention are: This invention can accurately calculate the impact of noise before it actually reaches the sensitive area and deploy the optimal strategy in advance. This is equivalent to giving the sound barrier the ability to "foresee the future," allowing control actions to occur before noise events, achieving true active noise control, and qualitatively improving the timeliness and effectiveness of noise reduction. The system utilizes a reinforcement learning decision-making model for policy selection. This model, trained extensively in a digital twin environment, learns not simply "if-then" rules, but complex strategies that seek the long-term optimal balance among multiple conflicting objectives such as noise reduction efficiency, energy consumption, and equipment health. This enables the system to not only make the best decision in the current situation but also consider the impact of the decision on equipment lifespan and long-term operating costs, maximizing the overall system benefits and surpassing control methods that rely on human experience or simple rules. The system creatively transforms traffic information technology data (vehicle type, speed) from vehicle-road cooperation or intelligent sensing into acoustic physical data through parameterized vehicle noise source models, thereby driving high-precision acoustic simulation. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 This is a schematic diagram of the method steps of a road noise barrier system and its noise reduction method according to the present invention. Detailed Implementation

[0021] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0022] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0023] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0024] Secondly, the present invention is described in detail with reference to the schematic diagrams. When detailing the embodiments of the present invention, for ease of explanation, the cross-sectional views illustrating the device structure may be partially enlarged, not according to the usual scale. Furthermore, the schematic diagrams are merely examples and should not limit the scope of protection of the present invention. In addition, actual fabrication should include three-dimensional spatial dimensions of length, width, and depth.

[0025] Reference Figure 1 A road noise barrier system is provided, comprising: The advanced perception module is used to acquire real-time information on traffic flow that will enter the target road barrier area within a set time period in the future; The digital twin and pre-visualization module has a built-in digital twin corresponding to the physical sound barrier, which is used to receive traffic flow information and pre-visualize the noise field and the effects of different adjustment strategies. The intelligent decision-making module integrates a reinforcement learning decision-making model to select the optimal noise reduction strategy based on the pre-simulation results; The dynamic execution module is used to receive instructions from the optimal noise reduction strategy and adjust the working state of the physical sound barrier before the traffic flow arrives.

[0026] Specifically, the digital twin is constructed in the following manner: Establish a high-precision three-dimensional geometric model of the target road barrier area and its surrounding environment; The positions and real-time status parameters of all adjustable units of the physical sound barrier are mapped in the three-dimensional geometric model. A parameterized vehicle noise source model and a sound wave propagation calculation model are embedded, wherein the vehicle noise source model maps different noise spectrum characteristics according to the vehicle type.

[0027] Specifically, digital twins include: 1. Construction of high-precision 3D geometric models The digital twin is constructed through the following steps and algorithms: First, based on Geographic Information System (GIS) data, laser point cloud scanning data, or oblique photogrammetry data of the target road barrier area, a high-precision three-dimensional geometric model is established, integrating road alignment, pavement material, barrier geometry, and the location of protected buildings or sensitive points. This model has a spatial grid resolution better than 0.5 meters, ensuring accurate characterization of the physical boundaries of sound wave propagation. 2. Physical entity state mapping In the aforementioned three-dimensional geometric model, through a one-to-one coordinate mapping relationship, the position, number, and current real-time state parameters (such as micro-perforated plate unit, deformable structural unit) of each adjustable unit of the physical sound barrier are synchronously mapped to the digital twin. This process establishes a real-time, dynamic data bridge between the physical world and the virtual space. 3. Parametric vehicle noise source model The core innovation lies in incorporating a parameterized vehicle noise source model. This model incorporates vehicle type... (e.g., heavy trucks, medium-sized buses, and cars) are used as the primary index, mapped to their unique noise spectrum characteristic functions. This function describes the speed at a specific velocity. Below, the vehicle noise power level varies with frequency. The distribution of .

[0028] An exemplary model formula is as follows:

[0029] in: This indicates the reference speed of the vehicle model. The reference sound power level below; Simulated noise as a function of vehicle speed The logarithmic growth relationship, The growth coefficient related to vehicle model; This item is used to simulate narrowband peak noise generated by an engine or transmission system; For the center frequency, This is the bandwidth factor; , , , , , , A vehicle acoustic fingerprint database is constructed, obtained through field testing and machine learning fitting. 4. Dynamic sound wave propagation calculation model The implanted acoustic wave propagation calculation model uses the finite-difference time-domain (FDTD) method or the parabolic equation (PE) method to calculate the propagation, reflection, diffraction and absorption of noise in a three-dimensional geometric model.

[0030] Taking the simplified form of the two-dimensional PE method as an example, its core formula is: ; in: In distance and height The complex sound pressure at the location; Reference wavenumber; For refractive index, For reference speed of sound, To account for the actual sound speed distribution under temperature and wind speed gradients; The model was calculated using a parameterized vehicle noise source model. As a dynamic sound source boundary adjustment, and simultaneously using the mapped adjustable unit state of the sound barrier (such as local surface impedance) as a dynamic propagation boundary condition, it is possible to simulate the three-dimensional dynamic noise field under different traffic flows and different barrier states.

[0031] 5. Data relevance and creative expression This approach transforms discrete traffic flow information (vehicle type, speed) into continuous physical sound sources using a parameterized vehicle noise source model. These sound sources, along with a high-precision 3D environment and variable barrier states, are then input into an advanced acoustic simulation model. The final output is a quantified noise reduction effect (e.g., A-weighted sound pressure level at each sensitive point) that can be used for strategy evaluation. Throughout this process, the "vehicle type" data is linked to a spectral coefficient database, driving the sound source model, influencing propagation calculations, and ultimately determining the simulation results. This forms a complete, closed, and logically rigorous technical chain, meeting the requirements of non-isolated data and strong logical correlation, thus constituting a robust and patentable digital twin construction method.

[0032] The reinforcement learning decision model employs a deep deterministic policy gradient, including: Actor Network: Enter current status Output a continuous action vector ; Critics Network: Input Status and actions Assess its expected long-term cumulative rewards; Training environment and process: Using a digital twin as the simulation environment: The training of the model is carried out entirely in a high-fidelity virtual environment constructed from the digital twin; Interaction and Iteration: The intelligent agent (actor network) observes the state of the environment. ; The agent outputs actions ; Digital twin receiving Simulate execution and calculate the state at the next moment. and instant rewards ; This interactive data Store in the experience replay buffer; The critic network uses data in the buffer and updates it through temporal difference learning to more accurately evaluate the value of actions; The actor network updates using a policy gradient method based on critics' guidance to output actions that will receive higher ratings; Training objective: Through millions of such virtual interaction iterations, the actor network will eventually learn a policy function. It can adapt to any complex state This directly maps to rewards that can accumulate over a long period of time. , The optimal action is to maximize the discount factor. .

[0033] The dynamic execution module specifically includes: Multiple distributed adjustable units, each unit integrating an acoustic impedance adjustable surface and a geometrically deformable structure; the acoustic impedance adjustable surface is composed of a micro-perforated plate array driven by piezoelectric ceramics, and the effective aperture of the micro-holes is adjusted by changing the driving voltage; the geometrically deformable structure is composed of a composite material panel with embedded shape memory alloy driving wires, and the panel is induced to produce preset bending or torsional deformation by controlling the current on and off of the driving wires. The local controller receives instructions from the intelligent decision-making module, parses them, and drives the corresponding adjustable unit to perform state adjustments.

[0034] One method for reducing noise using a road sound barrier includes the following steps: S1: Real-time acquisition of traffic flow information that will enter the target road barrier area within a future set time period, the traffic flow information including at least vehicle type, vehicle speed, vehicle position and estimated arrival time; S2: Input the traffic flow information into a digital twin associated with the target road barrier area, and pre-simulate the three-dimensional dynamic noise field generated by the traffic flow in the digital twin, and simultaneously pre-simulate the noise reduction effect under at least two different sound barrier adjustment strategies; S3: Based on the pre-simulation results, the optimal noise reduction strategy is selected from the at least two adjustment strategies using a reinforcement learning decision model. The reinforcement learning decision model is trained and makes decisions with the goal of maximizing the long-term comprehensive noise reduction benefits. S4: Before the traffic flow actually reaches the target road barrier area, adjust the working state of the sound barrier in advance according to the optimal noise reduction strategy.

[0035] Specifically, the traffic flow information in step S1 is obtained directly from the real-time reported data of connected vehicles through the vehicle-road cooperative communication interface, or is collected by the radar-visual fusion sensing device deployed upstream of the target road barrier area. The adjustment of the working state of the sound barrier in step S4 includes at least one of the following methods: adjusting the acoustic impedance distribution of the variable micro-perforated plate array of the sound barrier; driving the variable topology structure unit of the sound barrier to generate a preset deformation; switching the resonant frequency of the local acoustic metamaterial unit; after executing step S4, it also includes a feedback and learning step S5: collecting actual noise reduction effect data through noise monitoring terminals deployed in the protected area, comparing the actual noise reduction effect data with the corresponding pre-simulation results in step S2, generating feedback data, and using the feedback data to update and correct the prediction accuracy of the digital twin and the decision strategy of the reinforcement learning decision model online.

[0036] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A road noise barrier system, characterized in that, include: The advanced perception module is used to acquire real-time information on traffic flow that will enter the target road barrier area within a set time period in the future; The digital twin and pre-visualization module has a built-in digital twin corresponding to the physical sound barrier, which is used to receive traffic flow information and pre-visualize the noise field and the effects of different adjustment strategies. The intelligent decision-making module integrates a reinforcement learning decision-making model to select the optimal noise reduction strategy based on the pre-simulation results; The dynamic execution module is used to receive instructions from the optimal noise reduction strategy and adjust the working state of the physical sound barrier before the traffic flow arrives.

2. The road noise barrier system according to claim 1, characterized in that: The digital twin is constructed in the following manner: Establish a high-precision three-dimensional geometric model of the target road barrier area and its surrounding environment; The positions and real-time status parameters of all adjustable units of the physical sound barrier are mapped in the three-dimensional geometric model. A parameterized vehicle noise source model and a sound wave propagation calculation model are embedded, wherein the vehicle noise source model maps different noise spectrum characteristics according to the vehicle type.

3. A road noise barrier system according to claim 1, characterized in that: The reinforcement learning decision model employs a deep deterministic policy gradient, including: Actor Network: Enter current status Output a continuous action vector ; Critics Network: Input Status and actions Assess its expected long-term cumulative rewards; Training environment and process: Using a digital twin as the simulation environment: The training of the model is carried out entirely in a high-fidelity virtual environment constructed from the digital twin; Interaction and Iteration: The intelligent agent (actor network) observes the state of the environment. ; The agent outputs actions ; Digital twin receiving Simulate execution and calculate the state at the next moment. and instant rewards ; This interactive data Store in the experience replay buffer; The critic network uses data in the buffer and updates it through temporal difference learning to more accurately evaluate the value of actions; The actor network updates using a policy gradient method based on critics' guidance to output actions that will receive higher ratings; Training objective: Through millions of such virtual interaction iterations, the actor network will eventually learn a policy function. It can adapt to any complex state This directly maps to rewards that can accumulate over a long period of time. , The optimal action is to maximize the discount factor. .

4. A road noise barrier system according to claim 3, characterized in that: The dynamic execution module specifically includes: Multiple distributed adjustable units, each unit integrating an acoustic impedance adjustable surface and a geometrically deformable structure; the acoustic impedance adjustable surface is composed of a micro-perforated plate array driven by piezoelectric ceramics, and the effective aperture of the micro-holes is adjusted by changing the driving voltage; the geometrically deformable structure is composed of a composite material panel with embedded shape memory alloy driving wires, and the panel is induced to produce preset bending or torsional deformation by controlling the current on and off of the driving wires. The local controller receives instructions from the intelligent decision-making module, parses them, and drives the corresponding adjustable unit to perform state adjustments.

5. A noise reduction method for road sound barriers, characterized in that: Includes the following steps: S1: Real-time acquisition of traffic flow information that will enter the target road barrier area within a future set time period, the traffic flow information including at least vehicle type, vehicle speed, vehicle position and estimated arrival time; S2: Input the traffic flow information into a digital twin associated with the target road barrier area, and pre-simulate the three-dimensional dynamic noise field generated by the traffic flow in the digital twin, and simultaneously pre-simulate the noise reduction effect under at least two different sound barrier adjustment strategies; S3: Based on the pre-simulation results, the optimal noise reduction strategy is selected from the at least two adjustment strategies using a reinforcement learning decision model. The reinforcement learning decision model is trained and makes decisions with the goal of maximizing the long-term comprehensive noise reduction benefits. S4: Before the traffic flow actually reaches the target road barrier area, adjust the working state of the sound barrier in advance according to the optimal noise reduction strategy.

6. The noise reduction method for a road sound barrier according to claim 5, characterized in that: The traffic flow information mentioned in step S1 is obtained directly from the real-time reported data of connected vehicles through the vehicle-road cooperative communication interface, or collected by the radar-visual fusion sensing device deployed upstream of the target road barrier area.

7. A noise reduction method for a road noise barrier according to claim 5, characterized in that: The adjustment of the working state of the sound barrier in step S4 includes at least one of the following methods: adjusting the acoustic impedance distribution of the variable micro-perforated plate array of the sound barrier; driving the variable topology structure unit of the sound barrier to produce a preset deformation; and switching the resonant frequency of the local acoustic metamaterial unit.

8. A noise reduction method for a road noise barrier according to claim 7, characterized in that: After performing step S4, the process also includes a feedback and learning step S5: collecting actual noise reduction effect data through noise monitoring terminals deployed in the protected area, comparing the actual noise reduction effect data with the corresponding pre-simulation results in step S2, generating feedback data, and using the feedback data to update and correct the prediction accuracy of the digital twin and the decision strategy of the reinforcement learning decision model online.

Images