Vehicle-mounted modular sound wave meteorological intervention cabin and operation method thereof

Abstract

The invention discloses a vehicle-mounted modular sound wave meteorological intervention cabin which comprises a strong sound square cabin, a power square cabin, an intelligent cooperative control system and a vehicle-mounted system. The strong sound square cabin integrates a strong sound generating unit, a posture adjusting unit, an environment sensing unit and the like, and the power square cabin is provided with a diesel-driven air compressor unit and a diesel-driven generator unit to provide independent air sources and electric power. The intelligent cooperative control system comprises a data fusion and processing module, an intelligent decision center, a cooperative control and execution module and an effect evaluation and adaptive learning module. The strong sound square cabin and the power square cabin are designed in a modularized and split mode and are matched with a vehicle-mounted system to achieve rapid transportation and rapid on-site unfolding. Automatic collection and fusion of multi-source meteorological data are achieved through an intelligent cooperative control system, cloud and mist micro physical characteristics are recognized in real time in combination with a CNN-LSTM neural network model, matched sound wave parameters and horn angle instructions are automatically generated, dependence on artificial experience is thoroughly eliminated, and the real-time performance and accuracy of operation are greatly improved.

Description

Technical Field

[0001] This invention relates to the field of acoustic engineering technology, specifically to a vehicle-mounted modular acoustic weather intervention cabin and its operation method. Background Technology

[0002] Fog is an aerosol system composed of a large number of tiny water droplets or ice crystals suspended in the atmosphere. Traditional methods of fog dispersal and rain enhancement mainly rely on the spraying of chemical agents (such as silver iodide, dry ice, etc.) or the use of heating methods. These methods have limitations such as potential environmental pollution, high costs, limited operational range, or dependence on specific meteorological conditions.

[0003] Acoustic wave intervention, as a physical method, involves radiating sound waves of specific frequencies and power into cloud and fog areas. Utilizing the dynamic effects of sound waves on aerosol particles (such as acoustic wakes and resonance), it induces the collision, merging, and growth of tiny cloud or fog droplets, ultimately leading to their sedimentation and thus dissipating fog or inducing precipitation. This method boasts advantages such as being environmentally friendly, having a wide range of applications, and requiring no carrier. However, existing acoustic wave intervention technologies and equipment generally suffer from the following problems:

[0004] Low level of automation: The setting of sound wave parameters (such as frequency and power) and the aiming of the transmitting device (such as pitch angle) rely heavily on human experience and judgment, making it difficult to respond in real time to the ever-changing microphysical characteristics of clouds and fog, resulting in insufficient operational efficiency and accuracy.

[0005] Poor system integration and mobility: The sound system, power system, etc. are often separate, and the deployment is slow, making it difficult to meet the emergency meteorological support needs of rapid response and mobile deployment.

[0006] Weak adaptability: It lacks the ability to intelligently identify and adaptively match different types, altitudes, and densities of clouds and fog, and its intervention strategies are limited.

[0007] Therefore, there is an urgent need for a highly automated and modular acoustic meteorological intervention system that integrates advanced sensing, intelligent decision-making, and precise execution. Summary of the Invention

[0008] The purpose of this invention is to overcome or at least partially solve the above problems by proposing a vehicle-mounted modular acoustic meteorological intervention cabin and its operation method, which can be applied to scenarios such as rain enhancement, fog dispersal, bird control, anti-terrorism and riot control, and shooting down unidentified flying objects. Different observation equipment can be selected for different scenarios. For example, in anti-terrorism and riot control, a visual recognition camera and a thermal imager are provided to observe the dispersal of target groups.

[0009] To achieve the above objectives, the present invention adopts the following technical solution: a vehicle-mounted modular acoustic weather intervention cabin, characterized in that it includes a high-intensity acoustic cabin, a power cabin, an intelligent collaborative control system, and a vehicle-mounted system;

[0010] The high-intensity acoustic cabin includes a high-intensity acoustic generating unit, an attitude adjustment unit, an environmental sensing unit, and an air storage tank. The high-intensity acoustic generating unit is used to generate and radiate low-frequency high-intensity sound waves in a directional manner. The attitude adjustment unit is used to adjust the angle of the acoustic horn and realize the opening and closing of the cabin. The environmental sensing unit is used to detect meteorological and cloud characteristics in real time. The air storage tank is used to stabilize the compressed air pressure.

[0011] The power cabin includes a diesel-powered air compressor unit and a diesel-powered generator unit, which are used to provide independent air and electricity sources, respectively.

[0012] The intelligent collaborative control system includes a data fusion and processing module, an intelligent decision-making center, a collaborative control and execution module, and an effect evaluation and adaptive learning module. The data fusion and processing module is used to receive and process multi-source meteorological data. The intelligent decision-making center includes a trained neural network model, which is used to identify cloud and fog conditions based on the multi-source meteorological data and generate optimized acoustic intervention parameters and horn elevation angle commands. The collaborative control and execution module is used to control the high-intensity sound generation unit and attitude adjustment unit according to the commands, and supports wireless remote control function with a remote control distance of not less than 20 kilometers. The effect evaluation and adaptive learning module is used to optimize the neural network model based on operation feedback data.

[0013] The vehicle-mounted system includes a vehicle chassis and a rotary lock fixing device. The vehicle chassis serves as a mobile platform, while the rotary lock fixing device is used to detachably fix the two large modular cabins and ensure stability.

[0014] In a preferred embodiment, the neural network model adopts a CNN-LSTM hybrid model. The input feature vector includes at least multidimensional parameters such as cloud type, liquid water content, ambient temperature and humidity, average particle size, and wind speed. The output optimization parameters are calculated by a weighted correction formula, and the output optimization parameters include the final sound wave center frequency, the final sound pressure level, and the horn pitch angle.

[0015] The final formula for calculating the center frequency of the sound wave is:

[0016]

[0017] in, The fundamental frequency inferred by the neural network. For wind speed correction, This is a temperature correction term;

[0018] The final sound pressure level calculation formula is:

[0019]

[0020] in, The base sound pressure level inferred by the neural network. This is a distance attenuation factor related to cloud height. , Humidity attenuation factor;

[0021] The formula for calculating the horn's elevation angle is:

[0022]

[0023] in, With the goal of reaching high altitudes, For the height of the work station, The horizontal distance between the site and the cloud projection point. This is the wind direction refraction correction angle.

[0024] In a preferred embodiment, the effect evaluation and adaptive learning module adopts the Actor-Critic reinforcement learning framework, quantifies the operation effect through a reward function, and updates the parameters of the neural network model online or offline based on the reward function. The reward function includes a fog dissipation operation reward function and a rain enhancement operation reward function.

[0025] The reward function for fog dispersal operations is expressed as:

[0026]

[0027] in, , My weighting coefficients, To improve visibility, This represents the reduction in liquid water content.

[0028] The reward function for rain enhancement operations is expressed as:

[0029]

[0030] in, , These are the weighting coefficients. For radar echo enhancement, This represents the increase in precipitation intensity.

[0031] In a preferred embodiment, the high-intensity sound generating unit includes at least an electromagnetic modulation transmitter and an exponential acoustic horn, capable of generating and directionally radiating low-frequency high-intensity sound waves; within 2 seconds after the sound command is issued, the highest sound pressure level at 1 meter from the acoustic horn outlet is ≥160dB, with a frequency range of 25-400Hz, and the sound pressure level at 1 meter from the acoustic horn outlet for each frequency within this frequency range is not lower than 130dB; the high-intensity sound generating unit can also output sound waves at the same frequency as the target aircraft's gyroscope, using the resonance effect to destroy its gyroscope components.

[0032] In a preferred embodiment, the attitude adjustment unit includes at least a hydraulic system and a hollow turntable. The hydraulic system is used to adjust the pitch angle of the acoustic horn and open and close the container. The hollow turntable is used to drive the acoustic horn to rotate infinitely in azimuth.

[0033] In a preferred embodiment, the environmental sensing unit includes at least a meteorological sensor, a millimeter-wave cloud radar, and a laser cloud height meter, for real-time detection of the microphysical properties of target clouds and fog.

[0034] In a preferred embodiment, the power cabin further includes a cryogenic fuel system and sound-absorbing louvers. The cryogenic fuel system can switch between different grades of diesel fuel according to the ambient temperature to ensure normal operation of the equipment in low-temperature environments. The sound-absorbing louvers are integrated into the ventilation openings of the power cabin. The power cabin's energy supply capacity can support the entire intervention cabin to work continuously for no less than 2 hours and can be used normally in coastal environments, at an altitude of 3000 meters, and in an ambient temperature range of -20°C to 40°C.

[0035] A method for operating a vehicle-mounted modular acoustic weather intervention cabin includes the following steps:

[0036] S1. System self-test and deployment: Transport the acoustic and power cabins to the designated work point and deploy them quickly. The system is powered on and performs a self-test, and all units are ready.

[0037] S2. Collect and integrate multi-source meteorological data. The intelligent collaborative control system simultaneously acquires local environmental perception data, regional meteorological forecast data, and remote sensing data.

[0038] S3, Intelligent Decision Generation, Data Fusion and Processing Module preprocesses and standardizes the collected data. The neural network model of the intelligent decision center performs in-depth analysis on the processed data, identifies the characteristics of target clouds and fog, and generates optimized acoustic intervention parameters and horn elevation angle commands.

[0039] S4. Collaborative operation: The collaborative control and execution module controls the high-intensity sound generation unit and attitude adjustment unit to perform operations according to the generated instructions, and emits high-intensity sound waves with specific parameters to the target cloud and fog area.

[0040] S5. Process monitoring and dynamic adjustment: During the operation, monitor changes in cloud and fog conditions and environmental wind field in real time. If the data exceeds the preset threshold for changes in key meteorological elements, return to step S2 and repeat the data collection and decision-making process to achieve dynamic adaptation and adjustment.

[0041] S6. Effect evaluation and model update: After the task is completed, the effect is evaluated based on the comparison of data before and after the task, and the data from this task is used to incrementally learn the neural network model to optimize future decisions.

[0042] In a preferred embodiment, in step S3, the specific process of generating optimized treatment is as follows: after analyzing the preprocessed data, the neural network model, combined with the built-in prior knowledge, calculates the optimal sound wave intervention parameters and horn pitch angle command through forward reasoning.

[0043] In a preferred embodiment, in step S6, the incremental learning triggering conditions include three scenarios: automatic triggering after a single task is completed, triggering when the reward for N consecutive tasks is lower than a preset threshold, and triggering when the cosine similarity between the cloud and fog feature vector detected by the environmental perception unit and the closest record in the historical database is lower than a threshold; during the learning process, task data is stored in the experience playback pool.

[0044] Compared with existing technologies, the high-intensity acoustic and power cabins of this invention adopt a modular design, which, together with the vehicle-mounted system, enables rapid transportation and on-site deployment, adapting to the mobility requirements of different work sites and significantly shortening the preparation time for meteorological operations. Through an intelligent collaborative control system, it achieves automatic acquisition and fusion of multi-source meteorological data, and combines a CNN-LSTM neural network model to identify the microphysical characteristics of clouds and fog in real time, automatically generating matching acoustic parameters and horn angle commands, completely eliminating reliance on manual experience and significantly improving the real-time performance and accuracy of operations. Through multi-dimensional detection by the environmental perception unit, combined with an adaptive learning mechanism, it can intelligently identify different types, heights, and densities of clouds and fog, dynamically adjust intervention parameters, and optimize subsequent strategies through incremental learning, solving the problem of weak adaptability of existing equipment. Attached Figure Description

[0045] Figure 1 A schematic diagram showing the system composition and component connections of a vehicle-mounted modular acoustic meteorological intervention cabin;

[0046] Figure 2 This is a schematic diagram of the modular architecture and functional flow of an intelligent collaborative control system. Detailed Implementation

[0047] The present invention will be further described in detail below with reference to the accompanying drawings.

[0048] This specific embodiment is merely an explanation of the present invention and is not intended to limit the invention. After reading this description, those skilled in the art can make creative modifications to this embodiment as needed, but as long as they are within the scope of the claims of the present invention, they are protected by patent law.

[0049] This invention discloses a vehicle-mounted modular acoustic meteorological intervention cabin and its operation method, which solves the technical problems in the prior art. The overall concept is as follows:

[0050] Example 1

[0051] Please see Figures 1-2 A vehicle-mounted modular acoustic weather intervention cabin includes a high-intensity acoustic cabin module, a power cabin module, an intelligent collaborative control system, and a vehicle-mounted system.

[0052] The high-powered sound-absorbing modular unit module includes:

[0053] The high-intensity sound generating unit includes an electromagnetic modulation generator and an exponential acoustic horn connected to its outlet, used to generate and directionally radiate low-frequency high-intensity sound waves. The generator comprises a magnet, a vibrating assembly, a coil, and a housing. The vibrating assembly includes moving and stationary parts, both with slits to allow airflow. A coil is wound around the moving part; when energized, it is acted upon by a magnetic field, causing it to reciprocate and cut the airflow through the slits, thus generating sound waves.

[0054] Attitude Adjustment Unit: The attitude adjustment unit employs a structural design combining a hydraulic system and a hollow turntable. The hydraulic system uses telescopic movements to push the exponential acoustic horn, achieving pitch adjustment within the range of 0° to 90°, while a separate cylinder drives the opening and closing of the cabin. The hollow turntable, driven by a hydraulic motor, rotates the horn in the horizontal plane within the range of 0° to 180° azimuth. This dual-degree-of-freedom adjustment mechanism works in tandem to achieve precise and stable pointing of dynamic targets in cloud and fog areas.

[0055] Environmental sensing unit: Integrated outside the container, including meteorological sensors (temperature, humidity, pressure, wind) and cloud and fog detection equipment (such as millimeter-wave cloud radar and laser cloud height meter), used to acquire environmental parameters of the work site and macroscopic and microphysical characteristics of target clouds and fog in real time (such as cloud height, thickness, particle spectrum distribution, liquid water content, etc.).

[0056] Air tank: Used to stabilize compressed air pressure.

[0057] The power cabin module includes:

[0058] Diesel-powered air compressor unit: provides a stable and clean high-pressure air source for electromagnetic modulation sound generator.

[0059] Diesel-powered generator set: Provides independent power supply for all electrical equipment in the entire intervention cabin (including sound generators, hydraulic systems, control systems, etc.).

[0060] Low-temperature fuel system: includes a main fuel tank, an auxiliary fuel tank, and a fuel heater, used to switch between different grades of diesel fuel according to different ambient temperatures, ensuring normal start-up and operation of equipment in low-temperature environments.

[0061] Sound Insulation and Heat Dissipation Structure: A specially designed sound-absorbing louver is integrated into the power compartment's ventilation openings. Through its specific blade shape, tilt angle, and internal channel geometry, the louver allows for smooth airflow to ensure heat dissipation. Simultaneously, it causes multiple reflections, interferences, and energy dissipation of broadband noise generated by the power module as it passes through the louver's tortuous path, resulting in significant sound attenuation. This purely structural solution effectively blocks the propagation of mechanical noise to external high-intensity sound-emitting units, preventing noise interference with highly directional acoustic signals. While ensuring continuous and stable heat dissipation for the equipment, it also ensures the purity of the emitted acoustic signals.

[0062] The intelligent collaborative control system is the core embodiment of the intelligence level of this invention. Its hardware core is an industrial-grade server, deployed inside the high-intensity acoustic enclosure, and connected to each unit via wired / wireless networks. Specifically, it includes the following functional modules:

[0063] Data fusion and processing module: Receives and processes real-time data from environmental sensing units, data from external meteorological data sources (such as meteorological satellites and radar networks), and a preset "cloud-fog frequency matching library".

[0064] Regarding the physical mechanism and prior knowledge base construction of the interaction between sound waves and clouds, the "cloud-frequency matching library" is not a simple lookup table; its construction is based on the core dynamic effects of sound waves on aerosol particles.

[0065] The principle of resonance: When the frequency of a sound wave matches the natural vibration frequency of water droplets in clouds, energy transfer is maximized, promoting the vibration, collision, and merging of the droplets. The resonant frequency (f) of a water droplet is approximately inversely proportional to its radius (r), i.e. (where σ is surface tension and ρ is density). The optimal operating frequency range also differs for clouds with different particle size distributions (e.g., warm cloud droplet radius 5-20 μm, fog droplet radius 1-100 μm).

[0066] Sound wake effect: Nonlinear effects (such as sound flow) generated by high-intensity sound waves can cause microparticles to move relative to each other and collide and aggregate.

[0067] Therefore, the "cloud-frequency matching library" is essentially a database based on the mapping relationship between the microphysical properties of clouds and fog and acoustic parameters. Its key fields can be designed as follows:

[0068] Input features (cloud and fog conditions): cloud and fog type (cold cloud / warm cloud / advection fog / radiation fog), number concentration, liquid water content (LWC), average particle size, particle size distribution spectrum, cloud height and thickness.

[0069] Output targets (acoustic parameter benchmarks): recommended center frequency range, basic sound pressure level.

[0070] Intelligent Decision-Making Center (Integrated Neural Network Algorithm): This center incorporates a deep neural network model trained on a large amount of historical operational data and cloud / fog physics models. This model is capable of:

[0071] Feature extraction and situational awareness: Features are extracted from the input multi-source heterogeneous meteorological data to identify and classify the cloud and fog types (such as cold clouds, warm clouds, advection fog, and radiation fog) and their key state parameters in the current operating area in real time.

[0072] Parameter optimization decision-making: Based on the recognition results and combined with prior knowledge from the "cloud-frequency matching library," a neural network model performs multi-objective optimization calculations to dynamically output the optimal combination of acoustic intervention parameters, including center frequency, sound pressure level, modulation method, operation duration, and the corresponding horn elevation angle. Neural networks (such as CNN-LSTM hybrid models) can learn complex nonlinear relationships, and their decisions are more accurate and adaptive than traditional lookup table methods or fixed formulas.

[0073] The role of neural networks (such as CNN-LSTM hybrid models) is to go beyond simple table lookup and perform nonlinear optimization decisions under multi-factor coupling. Its decision-making process can be specifically described as follows:

[0074] Input layer: Receives the multidimensional feature vector X after preprocessing by the data fusion module.

[0075] X = [Cloud / Fog Type Code, LWC, Average Grain Size, Spectral Width, Ambient Temperature, Ambient Humidity, Wind Speed, Wind Direction, Cloud Base Height, Cloud Thickness]

[0076] Hidden layers (feature extraction and relation mapping): The model learns from features X to the ideal output. The complex mapping relationship.

[0077] Output Layer and Weighted Correction: The model does not directly output the "optimal" parameters, but rather outputs a basic parameter set, which is then used to calculate the final instruction via a weighted correction formula. This process can be described as follows:

[0078] Output 1: Final sound wave center frequency:

[0079]

[0080] The neural network infers the fundamental frequency from the prior "matching library" based on the characteristics of clouds and fog.

[0081] Wind speed correction. Strong winds cause sound wave path shifts and energy attenuation, requiring fine-tuning of the frequency to compensate. ( (These are the correction coefficients obtained through training).

[0082] Temperature correction item. The speed of sound is c = 331.4 + 0.6 * T (where T is the temperature in Celsius). Temperature changes affect the wavelength of sound waves, which in turn affects the resonance conditions.

[0083] Output 2: Final sound pressure level:

[0084]

[0085] The basic sound pressure level inferred by the neural network.

[0086] Distance attenuation factor related to cloud height. .

[0087] Humidity attenuation factor: High-intensity sound waves experience additional attenuation when propagating in humid air, which needs to be compensated for in advance.

[0088] Output 3: Horn elevation angle: calculated using simple geometric optics principles.

[0089]

[0090] in The correction angle for sound refraction caused by wind direction.

[0091] (Horn Pitch Angle): This refers to the angle between the transmitting axis of the sound horn and the horizontal plane. By adjusting this angle, the emitted sound beam can be precisely pointed at the target cloud in the air.

[0092] (Cloud height): refers to the altitude of the bottom of the target object (such as a cloud layer) relative to the mean sea level, which is usually obtained in real time by detection equipment such as laser cloud height meter and millimeter-wave radar.

[0093] (Site height): refers to the altitude of the plane of the current operating position of the vehicle-mounted acoustic transmitter relative to the mean sea level, which is obtained through integrated high-precision GPS / BeiDou positioning module and digital elevation model (DEM) data.

[0094] (Horizontal Distance): This refers to the straight-line distance between the work station location and the target cloud's vertical projection point on the ground on the horizontal projection plane. This parameter is calculated by combining positioning system and radar detection data. It is obtained through geometric calculation on the horizontal projection plane by fusing the coordinates of the vehicle-mounted positioning system and the cloud location data detected by radar. The core calculation steps are as follows: Coordinate Acquisition: Station Coordinates (X, Y): The precise geographic coordinates (latitude and longitude, usually converted to Cartesian coordinates such as UTM coordinates for calculation) of the device's current location are obtained through the vehicle's high-precision GNSS positioning module (such as GPS / BeiDou). Cloud Projection Point Coordinates :

[0095] Millimeter-wave radar or lidar detects the spatial position of a target cloud, outputting polar coordinates relative to the radar itself, including slant range, azimuth, and elevation. Using geometric relationships, combined with the radar's installation height and attitude, the planar offset of the cloud's base projection point on the horizontal plane relative to the radar is first calculated. This offset is then superimposed onto the radar's own geographic coordinates (obtained synchronously with the vehicle-mounted GNSS system calibration) to finally obtain the geographic coordinates of the target cloud's projection point. Distance calculation: After obtaining the Cartesian coordinates of the station and the cloud's projection point, the horizontal distance is the straight-line distance between these two points, calculated using the Euclidean distance formula in a two-dimensional plane.

[0096] .

[0097] (Wind Direction Refraction Correction Angle): This is a correction term to compensate for the influence of atmospheric wind field on the sound wave propagation path. As sound waves propagate in the air, the sound speed increases with the wind and decreases with the wind. This sound speed gradient caused by wind speed and direction will cause the sound wave propagation path to be refracted (biased to the side with lower sound speed). This refers to the pitch angle correction calculated based on real-time wind direction and speed data, used to counteract this refraction effect and ensure that the sound wave energy is accurately focused on the target cloud. Its value can be determined using empirical models or lookup tables based on real-time meteorological data.

[0098] Collaborative control and execution module: Based on the instructions of the intelligent decision-making center, it precisely controls the electromagnetic modulation generator to produce a strong sound wave with specified parameters, and synchronously drives the hydraulic action and turntable system to adjust the horn to the target angle.

[0099] Effect Evaluation and Adaptive Learning Module (Reinforcement Learning): During the operation, the system continuously monitors changes in cloud and fog conditions after the operation (through feedback from the sensing unit). This module uses a reinforcement learning algorithm, taking the operation effect (such as the degree of visibility improvement and radar echo enhancement) as a reward signal to fine-tune and optimize the decision-making model of the intelligent decision-making center. This enables the system to continuously improve itself in real-world environments, forming a complete intelligent closed loop of "perception-decision-execution-evaluation-learning".

[0100] (1) Quantification of performance evaluation indicators

[0101] First, the "performance of the assignment" is quantified into a computable metric, which is then used as the reward function (R) for reinforcement learning.

[0102] Fog suppression operation reward function:

[0103]

[0104] in It is the amount of visibility improvement. It is the decrease in liquid water content. , The weighting coefficient indicates that the reward value is positively correlated with the effect.

[0105] Rain enhancement operation reward function:

[0106]

[0107] in For radar echo enhancement, This represents the increase in precipitation intensity. , These are the weighting coefficients.

[0108] (2) The specific process of adaptive learning (reinforcement learning)

[0109] The intelligent collaborative control system can adopt the Actor-Critic reinforcement learning framework, and its specific implementation process is as follows:

[0110] Triggering conditions:

[0111] Learning after each assignment: Automatically triggered after each complete assignment task is completed.

[0112] Periodic assessment of learning: The reward R for N consecutive assignments is lower than a preset threshold. Triggered at any time.

[0113] Encountering a new cloud and fog pattern: When the cloud and fog feature vector detected by the sensing unit has a cosine similarity to the closest record in the historical database that is lower than a threshold, it is determined to be a new pattern and learning is triggered.

[0114] Learning process (adjustments to the neural network model):

[0115] Experience replay: Data (status) from each task Cloud and fog parameters before operation; actions : Acoustic wave parameters used; Rewards Effect evaluation; new status (Cloud and fog parameters after operation) are stored in the experience playback pool.

[0116] Model update:

[0117] Critic Network (Value Assessment) Update: Its goal is to more accurately predict the state. Next, take action The long-term benefits that can be obtained are achieved by updating network parameters by minimizing the temporal difference error.

[0118] Actor network (policy decision-making, i.e., the neural network at the center of intelligent decision-making) updates: Its goal is to adjust the policy to select actions that yield higher rewards. Using the gradient signal provided by the Critic network, the weight parameters of the last few fully connected layers of the Actor network are fine-tuned using policy gradient methods (such as PPO, DDPG), rather than updating the entire network, to ensure learning stability and avoid "catastrophic forgetting".

[0119] Update to the “Cloud-Fog Frequency Matching Library”: Add the {cloud features, final acoustic parameters} pairs from successful jobs (with higher reward R) as new high-quality samples to the matching library and optimize its prior knowledge.

[0120] Criteria for successful optimization:

[0121] Offline validation: On a validation set built using historical data, the average expected reward obtained by the optimized Actor network decision should be significantly higher than that of the unoptimized version.

[0122] Online validation: In subsequent similar assignments, the moving average of the performance evaluation reward R showed a statistically significant improvement.

[0123] Furthermore, the vehicle-mounted system includes a vehicle chassis and a rotary locking device. The vehicle chassis serves as a load-bearing mobile platform, while the rotary locking device is used to detachably secure the two large modular units and ensure stability. The high-intensity acoustic unit and the power unit of this invention adopt a standard modular unit design, adaptable to standard vehicle chassis, facilitating transportation. Inside the high-intensity acoustic unit, an exponential acoustic horn is connected to a hydraulic lifting cylinder via a bracket, enabling stepless adjustment from 0-90°. The industrial server of the intelligent collaborative control system employs a redundant design to ensure reliability.

[0124] During the actual operation, after the system starts, the millimeter-wave cloud radar continuously scans, transmitting cloud information to the intelligent decision-making center in real time. The neural network model of the decision-making center (e.g., a convolutional neural network pre-trained with a large number of historical cloud and fog samples for feature extraction, combined with a long short-term memory network to process time-series data) quickly analyzes and determines that the current cloud is a layered warm cloud, approximately 1500 meters high and 300 meters thick. The model then calls upon the basic resonant frequency range (e.g., 50-200Hz) of the warm cloud from the "cloud-fog-frequency matching library" and performs optimization calculations based on the current wind speed and direction, ultimately deciding on the optimal center frequency of 128Hz, sound pressure level of 145dB, horn elevation angle of 60°, and operation duration of 20 minutes. The control system then drives each unit to execute. After 10 minutes of operation, radar feedback shows that the cloud base height begins to decrease and the echo strengthens. The system determines that the operation is valid and continues execution until completion. All data from this operation is recorded and used for incremental learning of the model.

[0125] The above description of the embodiments is provided to facilitate understanding and use of the present invention by those skilled in the art. It is obvious to those skilled in the art that various modifications can be made to the embodiments, and the general principles described herein can be applied to other embodiments without creative effort. Therefore, the present invention is not limited to the above embodiments. Improvements and modifications made by those skilled in the art based on the disclosure of the present invention without departing from the scope of the present invention should be within the protection scope of the present invention.

Claims

1. A vehicle-mounted modular acoustic meteorological intervention cabin, characterized in that: It includes a high-powered acoustic cabin, a power cabin, an intelligent collaborative control system, and an in-vehicle system; The high-intensity acoustic cabin includes a high-intensity acoustic generating unit, an attitude adjustment unit, an environmental sensing unit, and an air storage tank. The high-intensity acoustic generating unit is used to generate and radiate low-frequency high-intensity sound waves in a directional manner. The attitude adjustment unit is used to adjust the angle of the acoustic horn and realize the opening and closing of the cabin. The environmental sensing unit is used to detect meteorological and cloud characteristics in real time. The air storage tank is used to stabilize the compressed air pressure. The power cabin includes a diesel-powered air compressor unit and a diesel-powered generator unit, which are used to provide independent air and electricity sources, respectively. The intelligent collaborative control system includes a data fusion and processing module, an intelligent decision-making center, a collaborative control and execution module, and an effect evaluation and adaptive learning module. The data fusion and processing module is used to receive and process multi-source meteorological data. The intelligent decision-making center includes a trained neural network model, which is used to identify cloud and fog conditions based on the multi-source meteorological data and generate optimized acoustic intervention parameters and horn elevation angle commands. The collaborative control and execution module is used to control the high-intensity sound generation unit and attitude adjustment unit according to the commands, and supports wireless remote control function with a remote control distance of not less than 20 kilometers. The effect evaluation and adaptive learning module is used to optimize the neural network model based on operation feedback data. The vehicle-mounted system includes a vehicle chassis and a rotary lock fixing device. The vehicle chassis serves as a mobile platform, while the rotary lock fixing device is used to detachably fix the two large modular cabins and ensure stability.

2. The vehicle-mounted modular acoustic weather intervention cabin according to claim 1, characterized in that: The neural network model adopts a CNN-LSTM hybrid model. The input feature vector includes at least multi-dimensional parameters such as cloud type, liquid water content, ambient temperature and humidity, average particle size, and wind speed. The output optimization parameters are calculated by a weighted correction formula and include the final sound wave center frequency, the final sound pressure level, and the horn pitch angle. The final formula for calculating the center frequency of the sound wave is: ； in, The fundamental frequency inferred by the neural network. For wind speed correction, This is a temperature correction term; The final sound pressure level calculation formula is: ； in, The base sound pressure level inferred by the neural network. This is a distance attenuation factor related to cloud height. , Humidity attenuation factor; The formula for calculating the horn's elevation angle is: ； in, With the goal of reaching high altitudes, For the height of the work site, The horizontal distance between the site and the cloud projection point. This is the wind direction refraction correction angle.

3. The vehicle-mounted modular acoustic weather intervention cabin according to claim 2, characterized in that: The effect evaluation and adaptive learning module adopts the Actor-Critic reinforcement learning framework, which quantifies the operation effect through a reward function and updates the parameters of the neural network model online or offline based on the reward function. The reward function includes a fog dissipation operation reward function and a rain enhancement operation reward function. The reward function for fog dispersal operations is expressed as: ； in, , My weighting coefficients, To improve visibility, This represents the reduction in liquid water content. The reward function for rain enhancement operations is expressed as: ； in, , These are the weighting coefficients. For radar echo enhancement, This represents the increase in precipitation intensity.

4. The vehicle-mounted modular acoustic weather intervention cabin according to claim 1, characterized in that: The high-intensity sound generating unit includes at least an electromagnetic modulation transmitter and an exponential acoustic horn, capable of generating and directionally radiating low-frequency high-intensity sound waves; within 2 seconds after the sound generation command is issued, the highest sound pressure level at 1 meter from the acoustic horn outlet is ≥160dB, with a frequency range of 25-400Hz, and the sound pressure level at 1 meter from the acoustic horn outlet for each frequency within this frequency range is not less than 130dB.

5. The vehicle-mounted modular acoustic weather intervention cabin according to claim 1, characterized in that: The attitude adjustment unit includes at least a hydraulic system and a hollow turntable. The hydraulic system is used to adjust the pitch angle of the acoustic horn and open and close the container. The hollow turntable is used to drive the acoustic horn to rotate infinitely in azimuth.

6. The vehicle-mounted modular acoustic weather intervention cabin according to claim 1, characterized in that: The environmental sensing unit includes at least a meteorological sensor, a millimeter-wave cloud radar, and a laser cloud height meter, used to detect the microphysical properties of target clouds and fog in real time.

7. The vehicle-mounted modular acoustic weather intervention cabin according to claim 1, characterized in that: The power cabin also includes a cryogenic fuel system and sound-absorbing louvers. The cryogenic fuel system can switch between different grades of diesel fuel according to the ambient temperature to ensure the normal operation of the equipment in low-temperature environments. The sound-absorbing louvers are integrated into the ventilation openings of the power cabin. The power cabin's energy supply capacity can support the entire intervention cabin to work continuously for no less than 2 hours, and to work continuously for no less than 1 hour in coastal environments, at an altitude of 3000 meters, and in an ambient temperature range of -20℃ to 40℃.

8. A method for operating the vehicle-mounted modular acoustic weather intervention cabin according to any one of claims 1-7, characterized in that, Includes the following steps: S1. System self-test and deployment: Transport the acoustic and power cabins to the designated work point and deploy them quickly. The system is powered on and performs a self-test, and all units are ready. S2. Collect and integrate multi-source meteorological data. The intelligent collaborative control system simultaneously acquires local environmental perception data, regional meteorological forecast data, and remote sensing data. S3, Intelligent Decision Generation, Data Fusion and Processing Module preprocesses and standardizes the collected data. The neural network model of the intelligent decision center performs in-depth analysis on the processed data, identifies the characteristics of target clouds and fog, and generates optimized acoustic intervention parameters and horn elevation angle commands. S4. Collaborative operation: The collaborative control and execution module controls the high-intensity sound generation unit and attitude adjustment unit to perform operations according to the generated instructions, and emits high-intensity sound waves with specific parameters to the target cloud and fog area. S5. Process monitoring and dynamic adjustment: During the operation, monitor changes in cloud and fog conditions and environmental wind field in real time. If the data exceeds the preset threshold for changes in key meteorological elements, return to step S2 and repeat the data collection and decision-making process to achieve dynamic adaptation and adjustment. S6. Effect evaluation and model update: After the task is completed, the effect is evaluated based on the comparison of data before and after the task, and the data from this task is used to incrementally learn the neural network model to optimize future decisions.

9. The operating method of the vehicle-mounted modular acoustic weather intervention cabin according to claim 8, characterized in that: In step S3, the specific process of generating optimized treatment is as follows: After analyzing the preprocessed data, the neural network model combines the built-in prior knowledge and calculates the optimal sound wave intervention parameters and horn pitch angle command through forward reasoning.

10. The operation method of the vehicle-mounted modular acoustic weather intervention cabin according to claim 8, characterized in that: In step S6, the incremental learning triggering conditions include three scenarios: automatic triggering after a single task is completed, triggering when the reward for the effect of N consecutive tasks is lower than a preset threshold, and triggering when the cosine similarity between the cloud and fog feature vector detected by the environmental perception unit and the closest record in the historical database is lower than a threshold; during the learning process, task data is stored in the experience playback pool.

Images