Weather modification rocket launching device based on artificial intelligence algorithm

Abstract

The application discloses a weather influencing rocket launching device based on an artificial intelligence algorithm and relates to the technical field of artificial intelligence algorithms.The weather influencing rocket launching device comprises a multi-source data acquisition and adaptation module, which acquires meteorological data, 44 / 56 mm rocket parameters, launching frame states and environmental data; an AI launching parameter optimization module, which optimizes a firing angle and propellant parameters by using a hybrid model and adapts a 4-7.5 km shooting height at a medium-high altitude; an AI risk early warning module, which identifies risks such as ignition abnormalities; an AI timing control module, which corrects sowing and self-destruction timing; a launching execution module, which drives a launching frame and ignition; a self-destruction cooperation module, which ensures that the residual debris is less than or equal to 100 g or low-altitude parachute opening; a data storage module, which stores data; and a man-machine interaction module, which displays and operates.The application uses AI to improve the precision of launching parameters, adapts different caliber rockets and medium-high altitude environments, has high catalyst utilization, has strong data storage and iteration capabilities and comprehensively optimizes rain enhancement and hail prevention operation efficiency.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence algorithm technology, and in particular to a rocket launch device for artificial weather modification based on artificial intelligence algorithms. Background Technology

[0002] As weather modification operations become more precise and efficient, traditional rain-enhancing and hail-suppression rocket launch systems, lacking intelligent technology support, struggle to cope with complex and ever-changing operational scenarios and diverse needs. Current systems rely heavily on manual experience to set core parameters, failing to respond in real-time to dynamic changes in meteorological conditions (wind speed, cloud height, air density) and rocket status (projectile weight, propellant remaining amount, flight trajectory). For example, in mid-to-high altitude regions, significant differences in low-altitude drag and gravity compared to plains mean that fixed parameters can easily lead to unexpected rocket altitude deviations, making it difficult to accurately deliver catalysts to target cloud layers at 4-7.5 km, drastically reducing operational efficiency. Furthermore, traditional systems only process multi-source data (meteorology, rocket, launch pad) at a simple acquisition and storage level, lacking in-depth analysis and feature fusion capabilities. This prevents the uncovering of data coupling relationships, causing parameter optimization to lag behind actual operational changes, further limiting operational accuracy.

[0003] Existing launch systems have significant shortcomings in their safety control mechanisms. Timing control relies on mechanical timers or simple logic circuits, which cannot dynamically adjust the timing of critical actions based on real-time rocket flight data. This often results in issues such as premature or delayed dispersal of catalysts and deviations in the timing of self-destruction. These problems not only affect the uniformity of catalyst dispersal but can also lead to safety risks such as abnormal ignition of the launch device, launch pad structural failure, and uncontrollable debris landing points if not addressed promptly due to engine ignition anomalies. Risk warning mechanisms often use fixed thresholds, only able to identify explicit problems such as voltage over-limits and abnormal resistance. They lack the ability to effectively identify implicit risks such as ignition circuit resistance fluctuations and micro-faults in the launch pad locking mechanism, resulting in high false alarm and false negative rates. Regarding debris handling, traditional systems lack intelligent trajectory prediction and dynamic adjustment capabilities, relying solely on preset parachute deployment altitudes or fixed self-destruct energies. This makes it difficult to optimize the process based on real-time wind speed and air density, leaving the risk of debris injuring personnel on the ground and damaging crops or infrastructure.

[0004] Traditional launch systems also have limitations in terms of system adaptability and long-term iteration capabilities. For compatibility with different launcher models, manual replacement of mechanical adapter components is required, resulting in long adjustment cycles and complex operations, making it difficult to quickly respond to the needs of multi-scenario operations. Furthermore, the selection of catalyst dispersal methods (combustion dispersal or explosive dispersal) relies on manual judgment, lacking intelligent analysis of key factors such as cloud density and thickness, easily leading to low catalyst utilization and mismatch between the dispersal range and the target cloud layer. Simultaneously, the system lacks adaptive learning capabilities; when facing new operational areas or new types of rockets (44mm / 56mm caliber), a complete process re-tuning is required, resulting in low iteration efficiency and difficulty in meeting the growing market demand for rain-enhancing and hail-suppressing rockets in the future. Summary of the Invention

[0005] This invention proposes a rocket launch device for artificial weather modification based on artificial intelligence algorithms to solve the problems mentioned in the prior art.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a rocket launch device for artificial weather modification based on artificial intelligence algorithms, comprising the following modules: The multi-source data acquisition and adaptation module deploys various sensing devices and terminals to collect meteorological, rocket core parameters, launch pad status and environmental data in the operation area, and samples them by frequency. The AI ​​launch parameter optimization module uses a hybrid model of deep learning and reinforcement learning. It takes pre-processed data as input and dynamically optimizes the launch elevation angle, propellant combustion control parameters and catalyst seeding trigger threshold for operation needs in medium and high altitude areas. The AI ​​launch risk warning module is based on the autoencoder-Transformer anomaly detection model. It connects to streaming data to build a risk identification network, which identifies multiple types of engine risks. After training with labeled samples, it outputs a warning probability for low risks and triggers a launch prohibition command for high risks. The AI ​​timing control module is equipped with an attention mechanism timing prediction model and integrates a sensor interface. It uses real-time rocket flight data as input to build a timing binding network, dynamically corrects the timing parameters related to seeding and self-destruction, and compares deviations and assigns weights. The launch execution module, in conjunction with the AI ​​timing control and launch parameter optimization modules, integrates ignition, launcher adjustment and functional separate control units, and receives AI commands to complete actions. The self-destruct collaborative control module adopts a federated learning collaborative control model, links related modules to build a collaborative network, and implements differentiated self-destruction and recovery control for rockets of different calibers. The data storage and traceability module adopts a distributed storage architecture to store raw data, training datasets, inference results and model parameters, and supports multi-dimensional classification retrieval. The human-computer interaction module is equipped with an industrial-grade touch terminal, which displays AI output results through data visualization, supports querying the reasoning process, and verifies the rationality of manually modified parameters.

[0007] Furthermore, it also includes an AI weather adaptation and prediction module, which is connected to the multi-source data acquisition and adaptation module and the AI ​​launch parameter optimization module. It adopts an ARIMA-LSTM hybrid time series prediction model, uses real-time meteorological data as input, predicts future weather change trends, captures meteorological abrupt change characteristics in mid-to-high altitude areas through an attention mechanism, and uses the prediction results as incremental input to the AI ​​launch parameter optimization module to dynamically adjust propellant combustion control parameters and launch elevation compensation.

[0008] Furthermore, it also includes an AI launcher compatibility and adaptation module. This module is connected to the multi-source data acquisition and adaptation module and the launch execution module. It adopts a CNN image recognition and mechanical parameter matching model, collects the appearance features of the launcher through image sensors, and combines the launcher model data from the multi-source data acquisition and adaptation module to build a launcher feature map library. The artificial intelligence algorithm automatically extracts the mechanical constraint parameters of different launcher models, generates personalized launcher adjustment schemes, and controls the launcher adjustment unit of the launch execution module to adapt to the mechanical structure. It optimizes the arrow rail angle adjustment step size for Ruida's newly developed variable arrow rail launcher and optimizes the locking mechanism action sequence for Jiangxi 9394 Factory's cage-type automated launcher.

[0009] Furthermore, the AI ​​launch parameter optimization module outputs the optimal launch elevation angle through an artificial intelligence-based launch angle optimization calculation model, using the following formula: ; in The optimal launch elevation angle output by the artificial intelligence model; The target cloud height; It is the acceleration due to gravity; The average wind speed in the work area; The angle between the wind direction and the launch direction; The theoretical flight time of the rocket; This refers to the rocket's average acceleration. For the rocket's launch speed; Altitude correction factor; This is the humidity correction factor; This represents the difference between the actual altitude and the standard altitude. The standard altitude is 1000m.

[0010] Furthermore, the AI ​​timing control module dynamically adjusts the self-destruct delay time through an artificial intelligence self-destruct timing correction model, using the following formula: in The self-destruct delay time after correction of the artificial intelligence model; The baseline self-destruct delay time; This is the speed deviation correction factor; This is the difference between the rocket's actual flight speed and its rated speed. The rated flight speed of the rocket; This is the height deviation correction factor; This is the difference between the rocket's actual flight altitude and its rated altitude. The rated flight altitude of the rocket; This is an air density correction factor; The actual air density in the work area; This refers to standard atmospheric density.

[0011] Furthermore, it also includes an AI fault diagnosis and self-healing module, which is connected to the AI ​​launch risk warning module and the launch execution module. It adopts a CNN-fault tree hybrid diagnostic model, inputting the abnormal data identified by the warning module into the CNN network for fault feature classification, and combining the fault tree model to locate the root cause of the fault. For faults that can heal themselves, the artificial intelligence algorithm generates a self-healing strategy and controls the launch execution module to execute it; for faults that cannot heal themselves, a fault classification report is generated and the launch function is locked, while maintenance suggestions are output.

[0012] Furthermore, the AI ​​timing control module also integrates an AI catalyst seeding optimization unit. This unit employs a hybrid model of reinforcement learning and rule-based reasoning, using cloud density data and rocket flight altitude data from a multi-source data acquisition and adaptation module as input. A DQN network optimizes the catalyst seeding method and duration, while the rule-based reasoning module controls the seeding rate based on the catalyst load, ensuring a cloud density ≥ 0.8 g / m³. 3 At that time, the artificial intelligence algorithm selected explosive dispersal and shortened the delay by 0.2 seconds; cloud density <0.3g / m³ 3 At that time, select combustion spreading and extend the spreading time by 1 second.

[0013] Furthermore, it also includes an AI debris trajectory prediction module, which is connected to the self-destruct collaborative control module and the multi-source data acquisition and adaptation module. It adopts a hybrid model of particle swarm optimization and deep learning. Using the initial velocity of the debris after self-destruction, the falling height, real-time wind speed and direction, and air density as inputs, the LSTM network predicts the change in the falling acceleration of the debris, and the particle swarm algorithm iteratively calculates the trajectory landing point. When the predicted landing point is a densely populated area or the vicinity of important facilities, the artificial intelligence module sends an early warning to the human-computer interaction module, and at the same time feeds back the landing point deviation data to the AI ​​launch parameter optimization module.

[0014] Furthermore, the data storage and traceability module also integrates an AI model incremental update unit. This unit uses a transfer learning algorithm to periodically extract new operational data from the stored data. Through feature transfer, the pre-trained model based on 100,000 rocket data is adapted to the new scenario data. The parameters of the AI ​​launch parameter optimization module and the AI ​​launch risk warning module can be updated without retraining the entire model.

[0015] Furthermore, the human-computer interaction module is also equipped with an AI permission management unit. This unit adopts a hybrid model of biometric recognition and role reasoning. It trains a CNN authentication model through industrial-grade IC card recognition and operator digital certificate data, and constructs a permission reasoning network in combination with operator roles. Administrators have the right to modify the core parameters of the AI ​​module, operators have the right to only perform launch operations and data queries, and maintenance personnel have the right to view fault records and equipment calibration data. The artificial intelligence algorithm records the operator's identity, operation content and operation time for each operation. When an unauthorized operation is detected, the operation function is immediately locked and an alarm message is sent to the administrator.

[0016] Compared with existing technologies, the beneficial effects of this invention are: This invention leverages artificial intelligence algorithms to achieve intelligent upgrades across the entire launch control process, comprehensively addressing the shortcomings in accuracy, safety, and adaptability of traditional systems, and significantly improving the overall efficiency of weather modification operations. At the parameter optimization level, the AI ​​algorithm automatically integrates multi-source data from meteorology, rockets, and the environment to dynamically optimize launch elevation angle, propellant combustion control parameters, and catalyst seeding trigger conditions. Even in complex environments at medium to high altitudes, it ensures precise rocket altitude coverage of target clouds, significantly improving catalyst seeding accuracy and operational efficiency. This avoids the limitations of traditional manual experience-based settings and adapts to the meteorological and geographical characteristics of different regions.

[0017] The level of intelligent safety control has been significantly improved. AI-based timing prediction technology can dynamically correct key timing sequences such as self-destruction separation and catalyst dispersal based on real-time rocket flight data, ensuring coordinated action of all functional units and reducing operational failures caused by timing deviations. Anomaly detection algorithms can accurately identify hidden risks and provide early warnings of potential problems such as abnormal engine ignition and launch pad locking failures, significantly reducing the probability of safety accidents. Debris trajectory prediction functionality, combined with real-time environmental data, predicts impact points, promptly avoiding risks near densely populated areas or important facilities, offering greater safety and reliability compared to traditional fixed-track systems.

[0018] The system's adaptability and iteration capabilities have been significantly enhanced. Through AI image recognition and parameter matching technology, the structural features and mechanical parameters of different launcher models can be automatically identified, generating personalized adjustment schemes. This allows for compatible use of multiple launcher models without manual replacement of adaptable parts, greatly improving operational preparation efficiency. Simultaneously, the AI ​​algorithm possesses adaptive learning capabilities, enabling rapid adaptation to new operational areas or new types of rockets based on historical operational data. This eliminates the need for complete process re-tuning, shortens the iteration cycle, and flexibly addresses future market demand growth and operational scenario expansion needs. Furthermore, the catalyst dispersal method is intelligently decided by the AI ​​algorithm based on cloud conditions, significantly improving utilization and further ensuring the effectiveness of rain enhancement and hail suppression operations, providing stable, efficient, and intelligent technical support for weather modification operations. Attached Figure Description

[0019] Figure 1 This is a schematic block diagram of a rocket launch device for artificial weather modification based on artificial intelligence algorithms proposed in this invention. Figure 2 A graph showing the change in weight of rocket debris over time after its self-destruction. Figure 3 A graph showing the relationship between rocket launch velocity and double-base propellant loading; Figure 4 This is a graph showing the relationship between catalyst utilization and target cloud height. Detailed Implementation

[0020] 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 embodiments of the present invention, and not all embodiments. 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.

[0021] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0022] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified. Furthermore, the terms "installed," "connected," and "linked" should be interpreted broadly; for example, they may refer to a fixed connection, a detachable connection, or an integral connection; they may refer to a mechanical connection or an electrical connection; they may refer to a direct connection or an indirect connection through an intermediate medium; and they may refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances. The invention will now be described in further detail with reference to the accompanying drawings.

[0023] Reference Figures 1 to 4 A rocket launcher for artificial weather modification based on artificial intelligence algorithms includes the following modules: The multi-source data acquisition and adaptation module is used to collect all-dimensional input data required by artificial intelligence algorithms. It deploys a meteorological sensor array, rocket parameter sensing unit, launch pad status monitoring terminal, and environmental sensing equipment. The module collects meteorological data for the operational area, including altitude, wind speed, wind direction, cloud height, and cloud thickness. Key rocket parameters include: missile diameter (44mm / 56mm), missile length (744-1400mm), missile weight (1.2-5.5kg), dual-base propellant load, propellant parameters for the three-stage self-destruct assembly (head self-destruct, middle self-destruct, and tail self-destruct), catalyst load (180g), and A-value per missile. The gI content is 10.8g. The launcher status data includes model (Ruida Civil Explosives JFJZD-01, JFJZD-02, LF-3, LF-6, LF-12, CF-6, CF-12, Ruida New Research Variable Arrow Rail Launcher and Jiangxi 9394 Factory 44mm / 56mm caliber cage-type automated launcher), arrow rail angle, locking status, ignition circuit resistance 0.55-1.0Ω, environmental data including temperature -20℃~50℃, humidity, and air pressure, data sampling frequency 1Hz and real-time preprocessing support to ensure the integrity and temporal continuity of the data features input to the artificial intelligence model; The AI ​​launch parameter optimization module, as the core computing unit, adopts a deep learning-reinforcement learning hybrid model. It takes the preprocessed data from the multi-source data acquisition and adaptation module as input and constructs a CNN-LSTM feature extraction network and a DQN deep Q network decision architecture. The feature extraction network mines the coupled correlation features of meteorological-rocket-environment data. The decision network dynamically optimizes the launch elevation angle (adapting to an 85° launch angle), dual-base propellant combustion control parameters (matching the combustion characteristics of the inner and outer surfaces of the single-hole tubular propellant), and catalyst seeding trigger threshold for the operational needs of medium and high altitude areas. The model is jointly trained with historical rocket operation data of 100,000 rockets and virtual simulation data. The optimized parameters improve the rocket launch speed and flight static stability ≥15%, with a launch altitude coverage of 4-7.5km. The AI ​​launch risk warning module, based on an autoencoder-Transformer anomaly detection model, integrates streaming data from a multi-source data acquisition and adaptation module in real time to construct a risk identification network with multiple feature inputs. The autoencoder extracts normal data distribution features, while the Transformer network captures abnormal fluctuations in time-series data. It can identify risk types such as engine ignition anomalies (based on time-series changes in ignition circuit resistance), launch pad locking failures (based on locking status sensor data), rocket parameter deviations (projectile diameter / length / weight deviating from rated values), and environmental temperature and humidity exceeding limits (exceeding the range of -20℃ to 50℃). The model has been trained with over 5000 labeled anomaly samples. It outputs a warning probability (confidence level 0.7) for low-risk situations and triggers a launch prohibition command for high-risk situations, simultaneously storing anomaly features and identification logs. The AI ​​timing control module, equipped with an attention-based timing prediction model and an integrated acceleration sensor data interface, uses real-time rocket flight data (altitude, velocity, acceleration) as input to construct a Seq2Seq timing binding network. It dynamically corrects the timing of the dispersal function (0.8-1.2s delay after separation to trigger dispersal), the catalyst dispersal ignition timing (12±1s after liftoff to ignite the flare), the self-destruct body separation and ejection timing (simultaneous triggering of dispersal and separation), and the self-destruct delay tube ignition timing (1s delay after separation or 24±2.5s after liftoff to trigger self-destruction). By comparing the deviation between the real-time trajectory and the theoretical trajectory, the model uses an attention weight allocation mechanism to prioritize the correction of key timing parameters, ensuring the coordinated action of the three-explosion self-destruct combination and the dispersal function. The launch execution module, in conjunction with the AI ​​timing control module and the AI ​​launch parameter optimization module, integrates an ignition control unit, a launcher adjustment unit, and a functional component separation control unit. The ignition control unit receives the ignition pulse signal optimized by artificial intelligence and adapts to the ignition requirements of the dual-base propellant igniter. The launcher adjustment unit drives the motor to adjust and lock the arrow track angle according to the launch angle parameters output by AI, adapting to different models of launcher mechanical structures. The functional component separation control unit receives AI timing instructions and controls the ignition of the separation ejection cartridge, realizing the separation of the dispersing functional part from the projectile body and the separation of the self-destructing part from the engine, with an execution action response delay of ≤100ms. The self-destruct collaborative control module adopts a federated learning collaborative control model, linking the AI ​​timing control module and the launch execution module to construct a self-destruct-recovery collaborative network. For 44mm caliber rockets, explosive self-destruction control is adopted, and the detonation energy of the self-destruction cartridge is adjusted through artificial intelligence algorithms to ensure that the weight of the debris after self-destruction is ≤100g. For 56mm caliber two-stage rockets, "explosive self-destruction of the seeding functional unit + low-altitude parachute opening of the first / second stage engines" control is adopted. Artificial intelligence algorithms predict the trajectory of debris falling and trigger parachute opening at an altitude of 500-800m to ensure that the debris falling velocity is ≤8m / s. The model improves the collaborative accuracy through federated training of self-destruction data from multiple operating areas. The data storage and traceability module adopts a distributed storage architecture to store the raw data of the multi-source data acquisition and adaptation module, the training dataset of the AI ​​module (historical operation data, abnormal sample data, virtual simulation data), inference results (optimization parameters, risk level, timing instructions) and model parameters. It supports retrieval by operation time, region and rocket model. The data storage duration is ≥3 years. At the same time, it provides an incremental training data interface for the AI ​​module to realize model iterative updates. The human-machine interaction module is equipped with an industrial-grade touch terminal and uses data visualization algorithms (such as t-SNE feature dimensionality reduction) to display the launch parameter optimization curve, risk warning heat map, timing control timing axis, and self-destruct trajectory prediction map output by the AI ​​model. It supports operators to query the AI ​​model reasoning process (feature importance ranking, risk identification basis). When manually modifying parameters, the AI ​​module must verify the rationality (compare the deviation between the modified parameters and the model's optimal solution). When there is a high-risk anomaly, an audible and visual alarm is triggered to realize human-machine collaborative operation.

[0024] This invention also includes an AI weather adaptation and prediction module, which is connected to a multi-source data acquisition and adaptation module and an AI launch parameter optimization module. It adopts an ARIMA-LSTM hybrid time-series prediction model, using real-time weather data (wind speed, wind direction, cloud movement speed) as input to predict the weather change trend in the next 5-10 minutes. It captures the meteorological change characteristics of mid-to-high altitude areas (such as mountain gusts and low humidity in arid areas) through an attention mechanism. The prediction results are used as incremental inputs to the AI ​​launch parameter optimization module to dynamically adjust the propellant combustion control parameters and launch elevation compensation. When the wind speed increases, the artificial intelligence algorithm automatically improves the propellant combustion efficiency to offset the wind resistance effect. When the cloud height increases, the propellant combustion control time is extended to increase the launch altitude, so that the rocket launch altitude is stabilized in the 4-7.5km operating range and the accuracy of catalyst dissemination is improved.

[0025] This invention also includes an AI launcher compatibility and adaptation module. This module is connected to the multi-source data acquisition and adaptation module and the launch execution module. It adopts a CNN image recognition and mechanical parameter matching model, and collects the appearance features of the launcher (arrow rail structure, locking mechanism, interface type) through an image sensor. Combined with the launcher model data from the multi-source data acquisition and adaptation module, a launcher feature map library is constructed. The artificial intelligence algorithm automatically extracts the mechanical constraint parameters (arrow rail adjustment range, locking force threshold, ignition signal interface protocol) of different launcher models, generates personalized launcher adjustment schemes, and controls the launcher adjustment unit of the launch execution module to adapt to the mechanical structure. It optimizes the arrow rail angle adjustment step size for the Ruida New Research Variable Arrow Rail Launcher and optimizes the locking mechanism action sequence for the Jiangxi 9394 Factory Cage Automated Launcher. It can achieve compatibility and use of multiple launcher models without manual replacement of adaptation parts.

[0026] In this invention, the AI ​​launch parameter optimization module outputs the optimal launch elevation angle through an artificial intelligence firing angle optimization calculation model, as shown in the formula: ; in The optimal launch elevation angle output by the artificial intelligence model, in degrees, is the DQN depth. The core output parameters of a network decision architecture; The target cloud height, in meters, is extracted from meteorological data by a CNN-LSTM feature extraction network, with a value ranging from 4000 to 7500. The acceleration due to gravity is taken as 9.8 m / s². 2 ; The average wind speed in the work area is expressed in m / s. The angle between the wind direction and the launch direction, in degrees; The theoretical flight time of the rocket, in seconds, is calculated by an artificial intelligence model based on the rocket's weight and propellant charge. The average acceleration of the rocket, in m / s².2 Matching the combustion characteristics of the inner and outer surfaces of the dual-base propellant; The launch velocity of the rocket is expressed in m / s. This is an altitude correction factor, ranging from 0.95 to 1.05, dynamically assigned by an artificial intelligence algorithm based on the altitude of the work area. This is a humidity correction factor, with a value ranging from 0.98 to 1.02. This represents the difference between the actual altitude and the standard altitude (1000m), expressed in kilometers. The standard altitude is set at 1000m. This calculation enables the artificial intelligence model to output a precise launch angle by integrating meteorological, altitude, and rocket propulsion characteristics, ensuring that the rocket reaches the target cloud area.

[0027] In this invention, the AI ​​timing control module dynamically adjusts the self-destruct delay time through an artificial intelligence self-destruct timing correction model, using the following formula: in The self-destruct delay time after correction by the artificial intelligence model is expressed in seconds and is the output of the Seq2Seq time series prediction network. The self-destruct delay time is the baseline, in seconds. The preset delay time is 1 second for a 44mm caliber rocket and 24 seconds for a 56mm caliber two-stage rocket. This is the speed deviation correction factor, with a value ranging from 0.8 to 1.2, assigned by an artificial intelligence algorithm based on flight acceleration fluctuations. This is the difference between the rocket's actual flight speed and its rated speed, expressed in m / s. The rated flight speed of the rocket, expressed in m / s; This is the height deviation correction factor, with a value ranging from 0.7 to 1.3; This is the difference between the rocket's actual flight altitude and its rated altitude, expressed in meters (m). The rated flight altitude of the rocket, in meters (m). This is the air density correction factor, with a value ranging from 0.1 to 0.3; The actual air density in the work area, in kg / m³ 3 ; The standard atmospheric density is taken as 1.225 kg / m³. 3 This calculation enables the artificial intelligence model to correct the self-destruct sequence based on the rocket's real-time flight status and atmospheric environment, ensuring that the weight of the debris after self-destruction is ≤100g and the falling speed is ≤8m / s.

[0028] This invention also includes an AI fault diagnosis and self-healing module, which is connected to the AI ​​launch risk warning module and the launch execution module. It employs a CNN-fault tree hybrid diagnostic model, inputting abnormal data identified by the warning module (such as abnormal ignition circuit resistance and sensor data fluctuations) into the CNN network for fault feature classification. The fault tree model is then used to locate the root cause of the fault (sensor failure, launcher mechanical jamming, and propellant loading abnormalities). For self-healable faults (such as temporary resistance fluctuations), the AI ​​algorithm generates a self-healing strategy (adjusting the power supply voltage and cleaning the ignition interface) and controls the launch execution module to execute it. For non-self-healable faults (such as propellant leakage), a fault classification report (confidence level 0.99) is generated, and the launch function is locked. Simultaneously, maintenance suggestions are output. The model has been trained with over 3000 fault samples, achieving a diagnostic accuracy of ≥98%, reducing the time spent on manual fault diagnosis and improving the continuous operation capability of the device.

[0029] In this invention, the AI ​​timing control module also integrates an AI catalyst dispersal optimization unit. This unit employs a reinforcement learning-rule reasoning hybrid model, using cloud density data and rocket flight altitude data from a multi-source data acquisition and adaptation module as input. A DQN network optimizes the catalyst dispersal method (combustion dispersal or explosive dispersal) and dispersal duration. The rule reasoning module controls the dispersal rate based on the catalyst carrying capacity (180g), ensuring a cloud density ≥0.8g / m³. 3 At that time, the artificial intelligence algorithm selected explosive dispersal and shortened the delay by 0.2 seconds to accelerate diffusion; cloud density <0.3g / m³ 3 When choosing to spread the catalyst by burning and extending the spreading time by 1 second, the utilization rate is improved. After more than 10,000 spreading simulation training sessions, the catalyst utilization rate is improved by ≥25%, ensuring the effectiveness of rain enhancement and hail prevention operations.

[0030] This invention also includes an AI debris trajectory prediction module, which is connected to the self-destruct collaborative control module and the multi-source data acquisition and adaptation module. It adopts a particle swarm optimization-deep learning hybrid model, taking the initial velocity of the debris after self-destruction, the falling height (500-800m), real-time wind speed and direction, and air density as inputs. The LSTM network predicts the change in the falling acceleration of the debris, and the particle swarm algorithm iteratively calculates the trajectory landing point. When the predicted landing point is a densely populated area or the vicinity of important facilities, the artificial intelligence module sends an early warning to the human-computer interaction module, and at the same time feeds back the landing point deviation data to the AI ​​launch parameter optimization module. During the next launch, the self-destruct sequence is adjusted to correct the trajectory. The model is trained with more than 5,000 debris trajectory samples, and the prediction error is ≤100m, reducing the safety hazards of debris.

[0031] In this invention, the data storage and traceability module also integrates an AI model incremental update unit. This unit uses a transfer learning algorithm to periodically extract new operational data (launch parameters, meteorological conditions, and operational results) from the stored data. Through feature transfer, the pre-trained model based on data from 100,000 rockets is adapted to new scenario data (such as new high-altitude operational areas). The parameters of the AI ​​launch parameter optimization module and the AI ​​launch risk warning module can be updated without retraining the entire model. The model update cycle is set to 7-30 days. After the update, the anomaly identification accuracy is improved by ≥5%, and the launch angle optimization accuracy is improved by ≥3%, adapting to different operational environment changes and rocket model upgrades.

[0032] In this invention, the human-computer interaction module is also equipped with an AI permission management unit. This unit adopts a hybrid model of biometric recognition and role reasoning. It trains a CNN identity verification model through industrial-grade IC card recognition and operator digital certificate data, and constructs a permission reasoning network by combining operator roles (administrator, operator, and maintenance worker). The administrator has the right to modify the core parameters of the AI ​​module (model weights, optimization thresholds), the operator has the right to only perform launch operations and data queries, and the maintenance worker has the right to view fault records and equipment calibration data. The artificial intelligence algorithm records the operator's identity, operation content, and operation time for each operation. When an unauthorized operation is detected, the operation function is immediately locked and an alarm message is sent to the administrator. The model has been trained with 1000+ user operation samples, with an identity verification accuracy rate of ≥99.5% and a permission allocation error rate of ≤0.1%, ensuring the system's operational security and traceability.

[0033] The following two examples further illustrate the specific implementation of this system: Example 1: Application of a 44mm caliber rain-enhancing and hail-suppressing rocket in a medium-to-high altitude region This embodiment was applied to rain enhancement and hail suppression operations in a mid-to-high altitude mountainous area of ​​a certain county. The average altitude of the operation area is 2000m, covering 200 square kilometers of farmland. The operation aimed to address hail damage caused by severe summer convective weather. A 44mm caliber rain enhancement and hail suppression rocket was used, combined with a launch device based on artificial intelligence algorithms to achieve precise operation. The specific operation is as follows: I. System Module Configuration and Operation Multi-source data acquisition and adaptation module: 15 sets of meteorological sensor arrays are deployed to collect data on wind speed (5 m / s), wind direction (southwest), cloud height (4200 m), and cloud thickness (800 m) in the operational area; 30 sets of rocket parameter sensing units are deployed to collect data on rocket diameter (44 mm), length (750 mm), weight (1.3 kg), propellant load (280 g), self-destruct device (No. 1 + No. 2) propellant parameters (5 g / unit), and catalyst load (180 g, including 10.8 g AgI); 5 launch pad status monitoring terminals are used to collect data on the initial launch pad trajectory angle (75°), normal lock-up status, and ignition circuit resistance (0.7 Ω); 5 sets of environmental sensing devices are used to collect data on ambient temperature (25℃), humidity (60%), and air pressure (80 kPa). The data sampling frequency is 1 Hz. During the preprocessing stage, 3 sets of abnormal wind speed data (an instantaneous wind speed of 12 m / s) are removed to ensure the integrity of the data features input to the AI ​​model.

[0034] The AI ​​launch parameter optimization module employs a deep learning-reinforcement learning hybrid model. It takes preprocessed data as input, extracts meteorological-rocket-environment coupling features through a CNN-LSTM network, and uses a DQN decision network combined with a modified launch angle optimization formula to calculate the optimal launch elevation angle. The formula is: ; in =4200m, =5m / s, =30° is the angle between the wind direction and the launch direction. =28s is the theoretical flight time of the rocket. =15m / s 2 For the rocket's average acceleration, =80m / s is the rocket's launch velocity. =1.02 is the correction factor corresponding to an altitude of 2000m. =1.0 is the correction factor corresponding to 60% humidity. =1000m is the difference between the actual altitude and the standard altitude. =1000m is the standard altitude.

[0035] Substituting the values ​​into the calculation: numerator: 4200 + 5 × cos30° × 28 = 4200 + 121.24 = 4321.24m Denominator: 0.5 × 15 × 28² + 80 × 28 = 5880 + 2240 = 8120m Fraction: 4321.24 ÷ 8120 ≈ 0.532arcsin(0.532) ≈ 32.1° First term: 32.1° × 1.02 = 32.742° Second term: 1.0 × (1000 ÷ 1000) = 1° The final optimal elevation angle was 32.742° + 1° ≈ 33.74°, which was adjusted to 34° to adapt to the operational scenario, ensuring that the rocket could reach a height of 4.3km and accurately cover the target cloud layer.

[0036] AI Launch Risk Warning Module: Based on an autoencoder-Transformer model, it monitors in real-time changes in ignition circuit resistance (0.7Ω±0.05Ω), launcher locking status (pressure sensor feedback locking force ≥500N), and ambient temperature (25℃ within the range of -20℃ to 50℃). Before operation, it detected that the resistance of one ignition circuit momentarily rose to 1.1Ω, exceeding the normal range of 0.55-1.0Ω. The model output a warning probability of 0.85 and a confidence level of 0.7, classifying it as low risk. After the operator cleaned the ignition interface, the resistance returned to 0.7Ω, thus avoiding the risk of ignition anomalies.

[0037] AI timing control module: Equipped with an attention mechanism Seq2Seq model, it takes real-time rocket flight data as input, which is an altitude of 2000m, a velocity of 250m / s, and an acceleration of 10m / s² at 10 seconds after liftoff. 2 Based on this, the dispersal and self-destruct sequence was corrected. At 12 seconds of liftoff, the pyrotechnic delay igniter ignited the warhead pyrotechnic, initiating the dispersal of AgI. At 12.5 seconds of liftoff, the dispersal function unit separation and ignition control module ignited the separation ejection cartridge, separating the dispersal function unit from the warhead. After a 1-second delay, at 13.5 seconds of liftoff, the dispersal function unit detonated, and the catalyst diffused evenly. Simultaneously, the engine ignition and self-destruct separation control module was triggered, igniting the self-destruct separation cartridge and ejecting the first self-destruct component to the bottom of the engine. Simultaneously, the short-delay self-destruct delay tubes of the first and second self-destruct components were ignited. After a 1-second delay, at 14.5 seconds of liftoff, the engine casing self-destructed. The debris weighed 85g.

[0038] Launch execution and self-destruction coordinated control: The launch execution module receives the AI-optimized 34° elevation angle command, drives the launch pad adjustment unit to adjust and lock the rocket track angle, and the ignition control unit outputs a 3A ignition pulse signal to ignite the engine igniter. The dual-base propellant burns inside and outside the launch pad, and the rocket launch velocity reaches 82m / s. The flight static stability is tested to be 16%. The self-destruction coordinated control module receives real-time feedback on the self-destruction body's actions to ensure uniform self-destruction energy. The maximum weight of the debris fragments is 85g, and there are no residues exceeding 100g.

[0039] II. Data Representation and Interpretation Table 1: Comparison of the operational performance of 44mm caliber rockets in a medium-to-high altitude region

[0040] Table 1 shows that this invention has significant advantages in operation at medium to high altitudes. The actual firing altitude is increased to 4.3km, meeting the operational requirements of 4-4.5km. This result is due to the dynamic compensation of altitude and wind speed by the AI ​​model. The accuracy of catalyst dispersal is greatly improved because the AI ​​timing control corrects the dispersal timing and angle, ensuring that AgI covers the target cloud layer. The weight of the debris after self-destruction is far less than 100g, reducing ground safety risks. The ignition anomaly warning rate is increased to 98%, avoiding ignition failure due to abnormal resistance. Operational efficiency is improved because the AI ​​automatically completes parameter optimization and risk detection, reducing manual debugging time and fully adapting to the complex operating environment at medium to high altitudes.

[0041] Example 2: Application of a 56mm caliber two-stage engine rain-enhancing and hail-suppressing rocket in grasslands This embodiment was applied to hail suppression operations on a grassland. The operation area has an average altitude of 1000m and covers 500 square kilometers of pasture. It required addressing spring hail disasters and increasing the catalyst dispersal height. A 56mm caliber two-stage engine rain-enhancing and hail-suppressing rocket was used, combined with a launch device based on artificial intelligence algorithms to achieve efficient operation. The specific operation is as follows: I. System Module Configuration and Operation Multi-source data acquisition and adaptation module: Deploys 20 sets of meteorological sensor arrays to collect data on wind speed (8 m / s), wind direction (northwest), cloud height (6000 m), and cloud thickness (1200 m); 40 sets of rocket parameter sensing units to collect data on rocket diameter (56 mm), length (1350 mm), weight (4.5 kg), first-stage double-base propellant load (450 g), second-stage double-base propellant load (300 g), propellant parameters for the three-stage self-destruct assembly (head + middle + tail) (8 g / unit), and catalyst load (180 g, including 0.8 g AgI); 8 launcher status monitoring terminals to collect data on the initial launcher trajectory angle (80°), normal lock-up status, and ignition circuit resistance (0.8 Ω); and 8 sets of environmental sensing devices to collect data on ambient temperature (15℃), humidity (40%), and air pressure (90 kPa). During data preprocessing, cloud density of 0.7 g / m³ is extracted; this data serves as the basis for selecting the catalyst dispersal method.

[0042] The AI ​​launch parameter optimization module employs a hybrid CNN-LSTM and DQN model to calculate optimal parameters from preprocessed input data. The first-stage rocket aims to penetrate the dense lower atmosphere; the optimized propellant burn time is 8 seconds, and the launch velocity is 95 m / s. For the second-stage rocket, after accumulating kinetic energy, the optimized propellant burn time is 12 seconds, and the average acceleration is 18 m / s². 2 This ensures a final firing altitude of 6.8 km. Simultaneously, based on a cloud density of 0.7 g / m³... 3 The AI ​​model selects the combustion-spreading method and sets the spreading time to 5 seconds to improve catalyst utilization.

[0043] AI Timing Control Module: Adjusts the self-destruct and parachute deployment timings using a self-destruct timing correction formula. The formula is as follows: ; in =24s is the baseline self-destruct time for a 56mm rocket. =1.0 is the speed deviation correction factor. =3m / s is the difference between the actual speed of 280m / s and the rated speed of 277m / s. =277m / s is the rocket's rated flight speed =0.9 is the height deviation correction factor. =300m is the difference between the actual height of 6800m and the rated height of 6500m. =650m is the rocket's rated flight altitude. =0.2 is the air density correction factor. =1.0kg / m 3 This represents the actual air density in the work area. =1.225kg / m 3 This refers to standard atmospheric density.

[0044] Substituting into the calculation, we get: =24×(1+1.0×3 / 277+0.9×300 / 6500)+0.2×1.0 / 1.225≈24×(1+0.0108+0.0415)+0.2×0.905≈24×1.0523+0.181≈25.33s, and the final self-destruct time is set to 25.3s.

[0045] Two-stage engine coordination and self-destruction recovery: After launch, the first-stage rocket engine operates for 8 seconds, propelling the rocket to an altitude of 2500m and a speed of 277m / s. At this point, the AI ​​timing control module triggers the first-stage engine to open its parachute and separate the propellant cartridge. The first-stage rocket debris falls freely to a low altitude of 800m, at which point the first-stage parachute opens, with a measured descent speed of 7.2m / s. Simultaneously, the second-stage rocket engine ignites, accelerating for 12 seconds to an altitude of 6800m and a speed of 350m / s, triggering the combustion and seeding unit to ignite and complete catalyst seeding within 5 seconds. After seeding, there is a 1-second delay before liftoff, and the rocket is in the air for 21 seconds. The seeding unit separates and triggers the nose self-destruct mechanism. At 25.3 seconds, the mid-stage and tail self-destruct mechanisms activate synchronously, with the debris weighing 92g after self-destruction. When the second-stage engine debris falls to a low altitude of 500m, the second-stage parachute opens, with a descent speed of 7.5m / s, and the rocket lands smoothly.

[0046] AI Fault Diagnosis and Self-Healing: During operation, data fluctuations were detected in the timing controller of a second-stage engine. The AI ​​fault diagnosis module located the sensor as having poor contact through a CNN-fault tree model, generated a self-healing strategy to adjust the power supply voltage to 12V, and the data returned to normal, avoiding ignition delay of the second-stage engine.

[0047] II. Data Representation and Interpretation Table 2: Comparison of the performance of 56mm caliber two-stage rockets in grassland operations

[0048] Table 2 data demonstrates the significant advantages of the synergistic effect of the two-stage structure and AI control in this invention. The maximum firing altitude is increased to 6.8 km, covering an operating range of 5-7.5 km; this improvement stems from the optimized energy distribution of the two-stage engines. The average debris fall velocity is far below 8 m / s, thanks to the precise coordination between the low-altitude parachute deployment system and AI timing control, reducing the safety risks associated with debris landing. Catalyst utilization is improved because AI selects combustion and seeding based on cloud density and optimizes the duration. The high success rate of two-stage separation reflects the AI's fault diagnosis and self-healing capabilities. Excellent low-temperature adaptability is achieved because the ignition propellant uses black powder and potassium borate, meeting the operating environment requirements of -20℃ to 50℃, fully adapting to the complex operational needs of grasslands.

[0049] Reference Figure 2 This diagram highlights the high efficiency of the three-stage self-destruct structure of this invention. Traditional self-destruct structures use single-stage detonation, resulting in uneven energy release. Even 2.5 seconds after self-destruction, the debris weight still reaches 110g, exceeding the 100g safety threshold and posing a ground safety hazard. This invention, through a segmented detonation design of the head, middle, and tail self-destruct bodies, combined with an integrated fully sealed assembly, ensures uniform energy release. The debris weight drops to 95g after 1.5 seconds and further to 75g after 2.5 seconds, remaining below the safety threshold throughout. Simultaneously, the shearing structure at the end face of the projectile ensures precise projection of the self-destruct body to the target location, preventing excessive debris residue and completely eliminating the risk of injury to personnel or facilities on the ground, meeting the design requirement of "debris ≤100g after self-destruction" in the documentation.

[0050] Reference Figure 3 This figure illustrates the effect of the propellant design of this invention on improving launch velocity. Traditional particulate propellants have a small combustion surface and low efficiency, resulting in a slow increase in launch velocity with increasing charge, reaching only 76 m / s at a charge of 400g, leading to insufficient rocket flight stability. This invention uses a single-hole tubular propellant with a double-base composition, expanding the combustion surface through simultaneous combustion on both the inner and outer surfaces, significantly improving combustion efficiency. The launch velocity reaches 70 m / s at a charge of 200g and increases to 90 m / s at 400g, a significantly higher growth rate than traditional methods. The higher launch velocity helps the rocket quickly overcome low-altitude drag, and combined with the tail fin design, further enhances flight stability, achieving a static stability of ≥15%, aligning with the design concept of "improving flight stability" outlined in the documentation.

[0051] Reference Figure 4This figure illustrates the optimization effect of the timing control technology of this invention on catalyst utilization. Traditional rockets use a fixed dispersal sequence, which cannot be adjusted according to cloud height. The utilization rate decreases with increasing cloud height, reaching only 55% at 7km, resulting in catalyst waste. This invention, through a timing controller with an acceleration sensor, dynamically sets the dispersal sequence based on the rocket's theoretical trajectory. Dispersal is triggered 0.2 seconds earlier at 3km low altitude and delayed 0.3 seconds at 7km high altitude, ensuring catalyst diffusion in the core region of the target cloud. Utilization reaches 90% at a cloud height of 5km and maintains 85% even at 7km, far exceeding traditional methods. Furthermore, the 44mm rocket supports combustion / explosive dispersal switching, and the 56mm rocket optimizes combustion dispersal duration, further improving utilization and conforming to the working principle of "precise catalyst dispersal" described in the document.

[0052] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A rocket launcher for artificial weather modification based on artificial intelligence algorithms, characterized in that, Includes the following modules: The multi-source data acquisition and adaptation module deploys various sensing devices and terminals to collect meteorological, rocket core parameters, launch pad status and environmental data in the operation area, and samples them by frequency. The AI ​​launch parameter optimization module uses a hybrid model of deep learning and reinforcement learning. It takes pre-processed data as input and dynamically optimizes the launch elevation angle, propellant combustion control parameters and catalyst seeding trigger threshold for operation needs in medium and high altitude areas. The AI ​​launch risk warning module is based on the autoencoder-Transformer anomaly detection model. It connects to streaming data to build a risk identification network, which identifies multiple types of engine risks. After training with labeled samples, it outputs a warning probability for low risks and triggers a launch prohibition command for high risks. The AI ​​timing control module is equipped with an attention mechanism timing prediction model and integrates a sensor interface. It uses real-time rocket flight data as input to build a timing binding network, dynamically corrects the timing parameters related to seeding and self-destruction, and compares deviations and assigns weights. The launch execution module, in conjunction with the AI ​​timing control and launch parameter optimization modules, integrates ignition, launcher adjustment and functional separate control units, and receives AI commands to complete actions. The self-destruct collaborative control module adopts a federated learning collaborative control model, links related modules to build a collaborative network, and implements differentiated self-destruction and recovery control for rockets of different calibers. The data storage and traceability module adopts a distributed storage architecture to store raw data, training datasets, inference results and model parameters, and supports multi-dimensional classification retrieval. The human-computer interaction module is equipped with an industrial-grade touch terminal, which displays AI output results through data visualization, supports querying the reasoning process, and verifies the rationality of manually modified parameters.

2. The artificial weather modification rocket launch device based on artificial intelligence algorithms according to claim 1, characterized in that, It also includes an AI weather adaptation and prediction module, which is connected to the multi-source data acquisition and adaptation module and the AI ​​launch parameter optimization module. It adopts an ARIMA-LSTM hybrid time series prediction model, uses real-time weather data as input, predicts future weather change trends, captures the meteorological change characteristics of mid-to-high altitude areas through an attention mechanism, and uses the prediction results as incremental input to the AI ​​launch parameter optimization module to dynamically adjust the propellant combustion control parameters and launch elevation compensation.

3. The artificial weather modification rocket launch device based on artificial intelligence algorithms according to claim 1, characterized in that, It also includes an AI launcher compatibility and adaptation module, which connects to the multi-source data acquisition and adaptation module and the launch execution module. It adopts a CNN image recognition and mechanical parameter matching model, collects the appearance features of the launcher through image sensors, and combines the launcher model data from the multi-source data acquisition and adaptation module to build a launcher feature map library. The artificial intelligence algorithm automatically extracts the mechanical constraint parameters of different launcher models, generates personalized launcher adjustment schemes, and controls the launcher adjustment unit of the launch execution module to adapt to the mechanical structure. It optimizes the arrow rail angle adjustment step size for Ruida's newly developed variable arrow rail launcher and optimizes the locking mechanism action sequence for Jiangxi 9394 Factory's cage-type automated launcher.

4. The artificial weather modification rocket launch device based on artificial intelligence algorithm according to claim 1, characterized in that, The AI ​​launch parameter optimization module outputs the optimal launch elevation angle through an artificial intelligence-based angle optimization calculation model, using the following formula: ; in The optimal launch elevation angle output by the artificial intelligence model; The target cloud height; It is the acceleration due to gravity; The average wind speed in the work area; The angle between the wind direction and the launch direction; The theoretical flight time of the rocket; This refers to the rocket's average acceleration. For the rocket's launch speed; Altitude correction factor; This is the humidity correction factor; This represents the difference between the actual altitude and the standard altitude. The standard altitude is 1000m.

5. A rocket launcher for artificial weather modification based on an artificial intelligence algorithm according to claim 1, characterized in that, The AI ​​timing control module dynamically adjusts the self-destruct delay time using an artificial intelligence self-destruct timing correction model, with the following formula: in The self-destruct delay time after correction of the artificial intelligence model; The baseline self-destruct delay time; This is the speed deviation correction factor; This is the difference between the rocket's actual flight speed and its rated speed. The rated flight speed of the rocket; This is the height deviation correction factor; This is the difference between the rocket's actual flight altitude and its rated altitude. The rated flight altitude of the rocket; This is an air density correction factor; The actual air density in the work area; This is the standard atmospheric density.

6. The artificial weather modification rocket launch device based on artificial intelligence algorithms according to claim 1, characterized in that, It also includes an AI fault diagnosis and self-healing module, which is connected to the AI ​​launch risk warning module and the launch execution module. It adopts a CNN-fault tree hybrid diagnostic model, inputting the abnormal data identified by the warning module into the CNN network for fault feature classification, and combining the fault tree model to locate the root cause of the fault. For faults that can heal themselves, the artificial intelligence algorithm generates a self-healing strategy and controls the launch execution module to execute it; for faults that cannot heal themselves, a fault classification report is generated and the launch function is locked, while maintenance suggestions are output.

7. The artificial weather modification rocket launch device based on artificial intelligence algorithm according to claim 1, characterized in that, The AI ​​timing control module also integrates an AI catalyst seeding optimization unit. This unit employs a hybrid model combining reinforcement learning and rule-based reasoning. Using cloud density data and rocket flight altitude data from a multi-source data acquisition and adaptation module as input, a DQN network optimizes the catalyst seeding method and duration. The rule-based reasoning module, combined with the catalyst load, controls the seeding rate, ensuring a cloud density ≥ 0.8 g / m³. 3 At that time, the artificial intelligence algorithm selected explosive dispersal and shortened the delay by 0.2 seconds; cloud density <0.3g / m³ 3 At that time, select combustion spreading and extend the spreading time by 1 second.

8. The artificial weather modification rocket launch device based on artificial intelligence algorithm according to claim 1, characterized in that, It also includes an AI debris trajectory prediction module, which is connected to the self-destruct collaborative control module and the multi-source data acquisition and adaptation module. It adopts a hybrid model of particle swarm optimization and deep learning. It uses the initial velocity of the debris after self-destruction, the falling height, the real-time wind speed and direction, and the air density as inputs. The LSTM network predicts the change in the falling acceleration of the debris, and the particle swarm algorithm iteratively calculates the trajectory landing point. When the predicted landing point is a densely populated area or the vicinity of important facilities, the artificial intelligence module sends an early warning to the human-computer interaction module and feeds back the landing point deviation data to the AI ​​launch parameter optimization module.

9. A rocket launcher for artificial weather modification based on an artificial intelligence algorithm according to claim 1, characterized in that, The data storage and traceability module also integrates an AI model incremental update unit. This unit uses a transfer learning algorithm to periodically extract new operational data from the stored data. Through feature transfer, the pre-trained model based on 100,000 rocket data is adapted to the new scenario data. The parameters of the AI ​​launch parameter optimization module and the AI ​​launch risk warning module can be updated without retraining the full model.

10. A rocket launcher for artificial weather modification based on an artificial intelligence algorithm according to claim 1, characterized in that, The human-computer interaction module is also equipped with an AI permission management unit. This unit adopts a hybrid model of biometrics and role-based reasoning. It trains a CNN authentication model through industrial-grade IC card recognition and operator digital certificate data, and constructs a permission reasoning network in combination with operator roles. Administrators have the right to modify the core parameters of the AI ​​module, operators have the right to only perform launch operations and data queries, and maintenance personnel have the right to view fault records and equipment calibration data. The artificial intelligence algorithm records the operator's identity, operation content, and operation time for each operation. When an unauthorized operation is detected, the operation function is immediately locked and an alarm message is sent to the administrator.

Images