Search

CN-121978922-A - Multi-scene adaptive solar unmanned aerial vehicle autonomous charge and discharge control system

CN121978922ACN 121978922 ACN121978922 ACN 121978922ACN-121978922-A

Abstract

The invention provides a multi-scene adaptive solar unmanned aerial vehicle autonomous charge and discharge control system, which relates to the technical field of unmanned aerial vehicle charge and discharge control and comprises a regulation and control system, wherein the regulation and control system comprises a multi-mode energy acquisition module, a bionic thermal energy management module, a group cooperative energy sharing module, a neural network driven intelligent scheduling module and an edge calculation and self-adaptive optimization module, and the multi-mode energy acquisition module, the bionic thermal energy management module, the group cooperative energy sharing module, the neural network driven intelligent scheduling module and the edge calculation and self-adaptive optimization module form a closed-loop energy ecological system through a high-speed serial bus and a wireless communication interface so as to realize autonomous charge and discharge regulation and dynamic optimization under multiple scenes.

Inventors

  • YOU FENG
  • YOU CHUN

Assignees

  • 上海捷翔航空技术有限公司

Dates

Publication Date
20260505
Application Date
20260114

Claims (10)

  1. 1. The utility model provides a solar unmanned aerial vehicle autonomous charge and discharge control system of many scene adaptations, includes regulation and control system, its characterized in that: the regulation and control system comprises a multi-mode energy collection module, a bionic thermal energy management module, a group cooperative energy sharing module, a neural network driven intelligent scheduling module and an edge calculation and self-adaptive optimization module, wherein the multi-mode energy collection module, the bionic thermal energy management module, the group cooperative energy sharing module, the neural network driven intelligent scheduling module and the edge calculation and self-adaptive optimization module form a closed-loop energy ecological system to realize autonomous charge and discharge regulation and dynamic optimization under multiple scenes, the multi-mode energy collection module is deployed on an unmanned plane body and wings to collect solar energy, wind energy and temperature difference thermal energy, perform local energy conversion, the bionic thermal energy management module integrates a phase change material layer and a thermoelectric converter to store and release thermal energy, auxiliary charge and discharge, the group cooperative energy sharing module coordinates energy distribution of the unmanned aerial vehicle group through laser energy transmission to form a distributed energy network, the intelligent scheduling module driven by the neural network analyzes task demands and environment data, dynamically distributes energy to a propulsion system, a communication module and a sensor module, the edge calculation and self-adaptive optimization module optimizes a charge and discharge strategy through online reinforcement learning, the multi-modal energy acquisition module, the bionic thermal energy management module, the group cooperative energy sharing module, the intelligent scheduling module driven by the neural network and the edge calculation and self-adaptive optimization module cooperatively operate through layered control flow, the multi-modal energy acquisition module preferentially processes local energy, the intelligent scheduling module driven by the neural network integrates group cooperative data, dynamically adjusting the charging and discharging rhythm.
  2. 2. The multi-scenario adaptive solar unmanned aerial vehicle autonomous charge and discharge control system according to claim 1, wherein a multi-modal energy sensing engine is deployed in the multi-modal energy collection module, a bionic thermal energy regulation engine is deployed in the bionic thermal energy management module, a neural network prediction engine is deployed in the neural network driven intelligent scheduling module, a group cooperative energy sharing module comprises a group cooperative allocation engine, an edge calculation and self-adaptive optimization module comprises a self-adaptive learning engine, and the regulation and control system achieves the aim by the following steps: The method comprises the steps of S1, collecting illumination intensity data, airflow speed data and temperature difference data through a multi-mode energy sensing engine deployed in a multi-mode energy collecting module, and operating a dynamic topology optimization algorithm to generate an energy collecting priority sequence, wherein the multi-mode energy sensing engine utilizes the illumination intensity data, the airflow speed data and the temperature difference data, the illumination intensity data are stored in an annular buffer area in a time sequence, the airflow speed data are analyzed in a vector format, the temperature difference data are generated into characteristic vectors through self-adaptive filtering, the dynamic topology optimization algorithm takes the illumination intensity data, the airflow speed data and the temperature difference data as input, adjusts a solar panel array connection mode through simulation energy collecting competition, and outputs an energy collecting priority sequence, the energy collecting priority sequence is transmitted to an intelligent scheduling module main control unit driven by a neural network through an in-machine high-speed bus, and is encrypted by a dynamic key, the multi-mode energy sensing engine operates in a state machine mode, is sampled once every hundred milliseconds, and is switched to high-frequency sampling and jumps to S2 when the illumination intensity is detected to be suddenly reduced; The method comprises the steps of S2, collecting solar panel temperature data and battery temperature data through a bionic thermal energy regulation engine deployed in a bionic thermal energy management module, operating a thermal energy storage distribution algorithm based on an energy collection priority sequence of S1 to generate a thermal energy release sequence, and broadcasting an energy state outline in combination with a laser energy sharing protocol, wherein the bionic thermal energy regulation engine utilizes the solar panel temperature data and the battery temperature data, the solar panel temperature data are stored in a time sequence, the battery temperature data are analyzed in a key value pair format, the thermal energy storage distribution algorithm takes the energy collection priority sequence, the solar panel temperature data and the battery temperature data as input, optimizes a phase change material layer storage rhythm by calculating a thermal energy storage offset, outputs the thermal energy release sequence, the energy state outline is broadcasted to surrounding unmanned aerial vehicles through a laser energy transmission unit, adopts pulse code compression data, and operates in an event driving mode to preferentially treat overheat abnormality of the solar panel, and jumps to S3 after confirming the abnormality; The method comprises the steps of S3, generating a task priority sequence through a neural network prediction engine deployed in an intelligent scheduling module driven by a neural network, and running a long-period memory network prediction algorithm to detect energy demand abnormality, wherein the neural network prediction engine utilizes task demand data and a heat energy release sequence of S2, the task demand data are stored in a time window code mode, the heat energy release sequence is analyzed in a vector sequence mode, the long-period memory network prediction algorithm takes the task demand data and the heat energy release sequence as input, and generates a task priority sequence through predicting the energy demand of thirty minutes in the future, detects energy demand deviation and outputs abnormal data, and the abnormal data is uploaded to a group collaborative energy sharing module through an encryption channel; The method comprises the steps of S4, collecting unmanned aerial vehicle group energy state records through a group cooperative allocation engine deployed on a group cooperative energy sharing module, operating an energy sharing optimization algorithm based on abnormal data of S3 to generate a shared path sequence and triggering a dynamic charging strategy, wherein the group cooperative allocation engine utilizes the unmanned aerial vehicle group energy state records and the abnormal data, the unmanned aerial vehicle group energy state records are stored by a time sequence directed graph, the abnormal data are encoded by a sparse matrix, the energy sharing optimization algorithm takes the unmanned aerial vehicle group energy state records and the abnormal data as input, and the energy sharing optimization algorithm generates the shared path sequence through simulating energy sharing path diffusion and outputs a charging instruction, the charging instruction is multicast to a target subsystem through an in-machine high-speed bus, and the group cooperative allocation engine operates in an asynchronous mode and jumps to S5 when optimization fails; The self-adaptive learning engine is used for storing the environment feedback data, the battery health status data in a multi-dimensional vector mode, the shared path sequence is integrated in a statistical sequence mode, the on-line reinforcement learning algorithm is used for generating an optimized charge-discharge sequence through an optimized energy allocation strategy, a visualized energy report is output, the optimized charge-discharge sequence is transmitted to the intelligent scheduling module driven by the neural network through an on-machine high-speed bus, a compression format is adopted, the self-adaptive learning engine is operated in a multi-thread pipeline mode, and prompts are sent out through the display module and flow is terminated after the optimized strategy is confirmed.
  3. 3. The multi-scene adaptive solar unmanned aerial vehicle autonomous charge and discharge control system according to claim 1 is characterized in that a dynamic topology optimization algorithm in S1 takes illumination intensity data, airflow speed data and temperature difference data as input, an energy collection priority sequence is generated by dynamically adjusting a solar panel array connection mode, a thermal energy storage distribution algorithm in S2 is driven to analyze overheat abnormality of a solar panel, the specific processing of the dynamic topology optimization algorithm is that photoelectric conversion efficiency of the solar panel array is calculated according to the illumination intensity data, wind power generation potential is evaluated according to the airflow speed data, thermoelectric conversion efficiency is determined according to the temperature difference data, serial connection or parallel connection mode of the solar panel array is adjusted through a competition optimization model, an energy collection priority sequence is stored in a fixed length vector code mode, the energy collection priority sequence is transmitted to a bionic energy regulation engine through an in-plane high-speed bus encryption mode, the dynamic topology optimization algorithm is operated in a state machine mode, and when illumination intensity sudden reduction is detected, switching to high-frequency sampling is conducted, and S2 is notified.
  4. 4. The multi-scene adaptive solar unmanned aerial vehicle autonomous charge and discharge control system according to claim 1, wherein the bionic thermal energy regulation engine in the S2 is composed of a temperature acquisition module, a thermal energy storage analyzer, a laser energy transmission interface and a thermal energy release sequence generator, and the operation mode is that the temperature acquisition module analyzes solar panel temperature data and battery temperature data, the thermal energy storage analyzer takes an energy acquisition priority sequence of the S1, solar panel temperature data and battery temperature data as input, a thermal energy storage distribution algorithm is operated to calculate thermal energy storage offset, the thermal energy release sequence generator compresses thermal energy storage offset data into a fixed-bit long thermal energy release sequence, the laser energy transmission interface broadcasts energy state outlines to surrounding unmanned aerial vehicles, the solar panel temperature data and the battery temperature data are stored in a local buffer memory in a low-rank matrix mode, the thermal energy release sequence is transmitted to a neural network prediction engine of the S3 and a group cooperative distribution engine of the S4 in a pulse coding mode through an in-plane high-speed bus, the bionic thermal energy engine operates in an event circulation mode, and the S3 is triggered when overheat abnormality of the solar panel is detected.
  5. 5. The multi-scenario adaptive solar unmanned aerial vehicle autonomous charge and discharge control system according to claim 1, wherein the thermal energy storage allocation algorithm in S2 takes an energy collection priority sequence, solar panel temperature data and battery temperature data as input, optimizes a phase change material layer storage rhythm by calculating thermal energy storage deviation, outputs a thermal energy release sequence, drives a long-short-period memory network prediction algorithm in S3 to analyze energy demand abnormality, and comprises the specific processes of determining an energy input proportion by the energy collection priority sequence, evaluating heat absorption demand of a phase change material layer by the solar panel temperature data, judging heat preservation demand by the battery temperature data, optimizing the thermal energy storage and release rhythm by a linear programming model, outputting a thermal energy release sequence, storing the thermal energy release sequence by sparse vectors, encrypting and transmitting the thermal energy release sequence to a neural network prediction engine by an in-machine high-speed bus, operating the thermal energy storage allocation algorithm by a feedback loop mode, and dynamically adjusting storage parameters.
  6. 6. The multi-scene adaptive solar unmanned aerial vehicle autonomous charge and discharge control system according to claim 1, wherein the laser energy sharing protocol in S2 is composed of a pulse encoder, a laser transmitter, a photon receiver and an error checking unit, and the working mode is that the pulse encoder maps an energy state outline into an interval coding sequence, the laser transmitter transmits the interval coding sequence, the photon receiver decodes the interval coding sequence, the error checking unit verifies data integrity, the energy state outline is stored in a local cache through sparse vectors, the laser energy sharing protocol transmits the energy state outline through a laser energy transmission unit, supports a dynamic charge strategy of S4 and an optimized charge and discharge sequence of S5, and the laser energy sharing protocol operates in a token scheduling mode and preferentially transmits the high-risk energy state outline to a group cooperative allocation engine.
  7. 7. The multi-scenario adaptive solar unmanned aerial vehicle autonomous charge and discharge control system according to claim 1 is characterized in that the neural network prediction engine in the S3 comprises a task data collector, a long-period memory network analyzer, an encrypted communication interface and an abnormal storage unit, and the working mode is that the task data collector analyzes task demand data of a propulsion system, a communication module and a sensor module to generate a time window code, the long-period memory network analyzer takes task demand data and a heat energy release sequence of the S2 as input, operates the long-period memory network prediction algorithm to calculate an energy demand deviation vector, the encrypted communication interface uploads abnormal data corresponding to the energy demand deviation vector, the abnormal storage unit caches historical energy demand deviation vector, the energy demand deviation vector is stored in a local cache in a sliding window mode, the abnormal data is uploaded to a group cooperative energy sharing module through an encrypted channel in a slicing mode, and the neural network prediction engine drives the long-period memory network prediction algorithm through a timer to trigger the group cooperative allocation engine of the S4 when the energy demand abnormality is detected.
  8. 8. The multi-scene adaptive solar unmanned aerial vehicle autonomous charge and discharge control system according to claim 1, wherein the long-term and short-term memory network prediction algorithm in S3 takes task demand data and a heat energy release sequence as input, detects energy demand abnormality through aggregating task priority sequence deviation, outputs abnormal data, and drives the energy sharing optimization algorithm in S4 to generate a shared path sequence; the method comprises the specific processing steps of extracting a power demand sequence of a propulsion system, a communication module and a sensor module through task demand data, evaluating auxiliary energy supply through a heat energy release sequence, predicting future thirty minutes of energy demand deviation through the long-short-term memory network, outputting abnormal data, storing the abnormal data in a high-dimensional tensor mode, uploading the abnormal data to a group cooperative energy sharing module through an encryption channel, operating the long-short-term memory network prediction algorithm in a pipeline mode, notifying S4 to execute a dynamic charging strategy, wherein a group cooperative allocation engine in the S4 consists of an energy state acquisition unit, a sharing optimizer, a charging controller and a path cache, and the working mode is that the energy state acquisition unit analyzes unmanned aerial vehicle group energy state records, the sharing optimizer generates a shared path sequence through the operation energy sharing optimization algorithm by taking the abnormal data of the S3 and the unmanned aerial vehicle group energy state records as inputs, the charging controller sends a charging instruction to a target subsystem, the path cache stores the historical shared path sequence, the unmanned aerial vehicle group energy state records are stored in a local matrix, the shared path sequence is multicast to the self-adaptive allocation engine in a self-adaption mode of S5 through a high-speed bus in the unmanned aerial vehicle, triggering S5 when the optimization fails.
  9. 9. The multi-scene adaptive solar unmanned aerial vehicle autonomous charge and discharge control system according to claim 1, wherein the energy sharing optimization algorithm in S4 takes unmanned aerial vehicle group energy state record and abnormal data as input, generates a shared path sequence through simulating energy sharing path diffusion, outputs a charging instruction, and drives the online reinforcement learning algorithm in S5 to optimize the charge and discharge sequence; the energy sharing optimization algorithm specifically processes that energy surplus and demand data of each unmanned plane are recorded and extracted according to the energy state of an unmanned plane group, a high priority subsystem is determined according to abnormal data, a laser energy transmission path is simulated through a graph optimization model, a shared path sequence is output, the shared path sequence is stored in a time slicing mode, the shared path sequence is encrypted and multicast to an adaptive learning engine through an internal high-speed bus, the energy sharing optimization algorithm operates in a feedback loop mode, optimization parameters are dynamically adjusted, the adaptive learning engine in S5 consists of an environment feedback acquisition unit, a battery state analyzer, an online reinforcement learning reasoning core and a report generator, the working mode is that the environment feedback acquisition unit analyzes the environment feedback data, the battery state analyzer extracts the battery health state data, the online reinforcement learning reasoning core takes the shared path sequence of S4, the environment feedback data and the battery health state data as input, the online reinforcement learning algorithm is operated to generate an optimized charge-discharge sequence, the report generator generates a visual energy report, the environment feedback data, the battery health state data and the shared path sequence are stored in a local cache in a layered mode, the self-adaptive learning engine operates in a multithread pipeline mode, and triggers the display module after confirming an optimization strategy.
  10. 10. The multi-scene adaptive solar unmanned aerial vehicle autonomous charge and discharge control system according to claim 1 is characterized in that an online reinforcement learning algorithm in S5 takes environment feedback data, battery health state data and a shared path sequence as input, generates an optimized charge and discharge sequence through an optimized energy distribution strategy and outputs a visual energy report, wherein the online reinforcement learning algorithm specifically processes that illumination intensity change, airflow speed change and temperature difference change are extracted through the environment feedback data, battery charge and discharge depth is estimated through battery health state data, group cooperative energy input is determined through the shared path sequence, charge and discharge rhythm is optimized through a depth deterministic strategy gradient model, the optimized charge and discharge sequence is stored in a multi-dimensional vector code mode, the optimized charge and discharge sequence is transmitted to an intelligent scheduling module driven by a neural network through in-machine high-speed bus encryption, and the online reinforcement learning algorithm operates in an incremental update mode and dynamically adjusts learning rate.

Description

Multi-scene adaptive solar unmanned aerial vehicle autonomous charge and discharge control system Technical Field The invention relates to the technical field of unmanned aerial vehicle charge and discharge control, in particular to a solar unmanned aerial vehicle autonomous charge and discharge control system with multiple scene adaptation. Background Along with the rapid development of unmanned aerial vehicle technology, solar unmanned aerial vehicle has been widely used in various scenes such as meteorological monitoring, logistics distribution, ocean search and rescue, polar region scientific investigation and border patrol because of its long-endurance, low energy consumption and environmental friendly characteristics. As a core component of the solar unmanned aerial vehicle, the performance of the energy collection and charge-discharge control system directly influences the task execution efficiency and the cruising ability of the solar unmanned aerial vehicle. Therefore, how to realize high-efficiency energy collection, autonomous charge and discharge management and dynamic optimization in multiple scenes has become an important research direction in the field of unmanned aerial vehicles. At present, energy collection and charge-discharge management of a solar unmanned aerial vehicle mainly depend on the following technical means: the fixed solar panel is collected, the traditional solar unmanned aerial vehicle is generally provided with a fixed solar panel (such as monocrystalline silicon or polycrystalline silicon panel) on a fuselage and a wing, and solar energy is converted into electric energy by utilizing a photovoltaic effect to supply power for a propulsion system, a communication module and a sensor module. The prior system adopts a single-mode energy management scheme, monitors the charge and discharge states of the lithium battery through a simple Battery Management System (BMS), and adjusts energy distribution by combining a preset threshold (such as 20% -80% of electric quantity) so as to prevent overcharge or overdischarge. And the ground station assisted scheduling, namely in a part of task scenes, the unmanned aerial vehicle carries out energy management through remote instructions of the ground station. And the ground station adjusts the flight path or task priority of the unmanned aerial vehicle according to the task demands and weather forecast so as to optimize the solar energy collection efficiency. Backup power supports that solar unmanned aerial vehicles are typically equipped with backup batteries (e.g., lithium polymer batteries) or small fuel cells as auxiliary energy sources to cope with night or severe weather (e.g., cloud cover, sand storm), ensuring task continuity. Although the above-described techniques have achieved certain success in the energy management of solar unmanned aerial vehicles, the following disadvantages still remain: The fixed solar panel collecting mode is limited by a single energy source (solar energy), the efficiency is obviously reduced under low illumination (such as night, illumination intensity <100W/m < 2 >) or severe environment (such as storm and sand dust), and the continuous energy requirement of multi-scene tasks can not be met. This limits the long endurance capabilities of the drone in complex environments. Secondly, the single-mode energy management lacks the integration capability of various energy sources (such as wind energy and temperature difference heat energy), and cannot fully utilize environmental resources. The threshold control strategy of the traditional battery management system is too simple to dynamically adapt to task priority changes (e.g., propulsion systems take precedence over sensor modules), resulting in inefficient energy distribution. Problem three ground station assisted scheduling relies on stable communication links (e.g., 5G or satellite communications), and communication delays (> 200 ms) or interruptions can affect the real-time performance of energy management in remote areas (e.g., ocean, polar) or in high interference environments (e.g., urban high-rise areas). In addition, centralized scheduling of ground stations is difficult to cope with the cooperative demands of multiple unmanned aerial vehicles, and the expansibility of group cooperative tasks is limited. And fourthly, the standby power supply (such as a lithium battery and a fuel cell) increases the weight and the cost of the unmanned aerial vehicle, and reduces the overall energy efficiency ratio. In long-time tasks, the standby power supply has limited capacity (such as 5000 mAh), high-power consumption equipment (such as a high-definition video module, 150W) is difficult to support, and frequent charging and discharging can accelerate battery aging (SOH is reduced to < 0.8), which affects system reliability. Therefore, a multi-scenario adaptive solar unmanned aerial vehicle autonomous charge and discharge control system is needed to solve the a