Modular power management and low power consumption optimization method for deep sea seismic detection system

Abstract

The invention discloses a modularized power supply management and low power consumption optimization method of a deep sea seismic detection system, and relates to the technical field of detector power supply management. Comprising an energy collection module, an energy storage module and the like, a multi-mode composite collector is adopted for energy collection, and a mixing mode is adopted for energy storage; predicting energy based on a deep space-time attention network; dynamically distributing power consumption by using an improved multi-target particle swarm algorithm; when energy is insufficient, a low-power-consumption mode is switched according to a state machine, and module voltage frequency is adjusted; power management is optimized by means of a near-end strategy optimization algorithm in combination with multi-dimensional data; the method further comprises the steps of fault diagnosis, redundant power supply, multi-mode energy recovery and the like. Energy supply stability is improved through multi-modal energy collection and hybrid energy storage, accurate power consumption distribution and low power consumption optimization are realized by using an intelligent algorithm, and fault diagnosis and energy recovery are combined, so that system reliability and energy utilization rate are improved, and working time of a deep sea seismic detection system is prolonged.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of detector power management, in particular to a modular power management and low-power optimization method for a deep-sea seismic detection system. BACKGROUND

[0002] Deep-sea seismic detection is of great significance for studying the internal structure of the earth, predicting earthquake disasters, and exploring marine geological activities. With the development of deep-sea detection technology, deep-sea seismic detection systems are gradually moving towards intelligence and long-term operation, but the power management of the system faces many challenges. The deep-sea environment is complex and changeable, with weak light, large temperature differences, and frequent wave surges. Traditional single energy harvesting methods, such as relying solely on temperature difference or wave energy generation, cannot stably obtain sufficient energy, and cannot meet the long-term operation requirements of the system. At the same time, existing energy storage devices have limitations in energy density, charging and discharging efficiency, and service life, which cannot effectively ensure the stable operation of the system when energy supply is insufficient.

[0003] In terms of power management strategy, traditional deep-sea seismic detection systems mostly use fixed power supply modes, lacking dynamic perception and response to environmental changes and system power consumption requirements. When the system energy harvesting is insufficient, it can only simply turn off some functional modules, and cannot accurately balance function implementation and power consumption control, leading to system performance degradation or premature stop working. Moreover, the existing power management methods do not fully consider the relevance and synergy between system modules, and the power distribution is unreasonable, causing energy waste or insufficient power supply to key modules.

[0004] In addition, during the long-term operation of the deep-sea seismic detection system, the power module is easily affected by seawater corrosion, high pressure, and other harsh environments, with a high probability of failure. However, the existing fault diagnosis and processing mechanism is inefficient and inaccurate, and it is difficult to quickly locate faults and take effective measures. Once the power module fails, it may cause the entire detection system to fail, affecting data acquisition and transmission, making it difficult to continuously and stably carry out deep-sea seismic detection work, and there is an urgent need for more advanced power management and low-power optimization technology. SUMMARY

[0005] The present application proposes a modular power management and low-power optimization method for a deep-sea seismic detection system to solve the problems mentioned in the above prior art.

[0006] In order to achieve the above purpose, the present application adopts the following technical scheme: a modular power management and low-power optimization method for a deep-sea seismic detection system, comprising: The power module is divided into four modules: an energy harvesting module, an energy storage module, a voltage conversion module, and a power distribution module. The energy harvesting module uses a composite energy harvester, integrating a thermoelectric generator, an adaptive wave energy conversion device, and a low-light solar panel. The thermoelectric generator uses thermoelectric materials, and the thermoelectric conversion efficiency formula is... ,in , For Seebeck coefficients, For electrical conductivity, Thermal conductivity, The absolute temperature is used; the adaptive wave energy conversion device can automatically adjust the conversion efficiency according to different wave frequencies and amplitudes. The conversion efficiency formula is: ,in For output power, For input power, For adaptive adjustment coefficients, For wave frequency, The resonant frequency of the device; the low-light solar panel generates electricity in a weak light environment, and its power generation efficiency is affected by light intensity and spectral characteristics, as shown in the formula. ,in For conversion factors, Light intensity, Let be the spectral response function. Let be the spectral distribution function of the light source; the energy storage module adopts a hybrid energy storage approach, combining lithium battery packs and supercapacitors, and is equipped with a battery management system. By real-time monitoring of the voltage, current, and temperature parameters of the batteries and capacitors, a fuzzy control algorithm is used to optimize energy storage allocation, as shown in the formula. ,in Assigning energy storage capacity For energy storage of lithium batteries, For energy storage in supercapacitors, and These are weighting coefficients that are dynamically adjusted based on the system state. Energy prediction steps: A deep spatiotemporal attention network (DSTAN)-based energy prediction model is constructed using historical energy acquisition data, real-time marine environmental parameters, and seismic activity data; combined with deep-sea environmental parameters such as temperature, current velocity, and seismic wave frequency, the prediction is then performed using formulas. Energy harvesting volume prediction, among which To predict energy harvesting volume, For marine environmental parameters, For parameters related to seismic activity, and The coefficients of the corresponding parameters, This is the error term.

[0007] Furthermore, it also includes: Dynamic power allocation step: According to the energy prediction results, the real-time power consumption demand of each module of the system and the task priority, an optimized dynamic power allocation model is established; the power supply power of each module is calculated through formula , wherein is the power allocated to the module, is the power required by the module, is the available energy, is the task priority of the module, is the power required by the first module, is the task priority of the first module, to realize the dynamic allocation of the power supply; Power consumption mode switching step: when the energy harvesting amount is lower than the set threshold, the system automatically enters the low power consumption mode, a low power consumption mode switching strategy based on state machine is introduced, different low power consumption mode combinations are dynamically selected according to different working states and task demands of the system; in addition to shutting down functional modules and adopting pulse power supply mode, dynamic adjustment of voltage and frequency is also carried out on the modules, through formula , wherein is the power consumption, is the load capacitance, is the voltage, is the frequency; the low power consumption judgment formula is set as: if , the mode is switched, wherein is the current harvested energy, is the proportional coefficient, is the energy threshold, is the demand change adjustment coefficient, is the energy demand change amount, is the time interval; Power management optimization step: real-time monitoring of the running state of each module of the power supply system, collection of voltage, current, temperature, fault information data; using the proximal policy optimization algorithm PPO in deep reinforcement learning, combining the experience replay and target network update mechanism, the working parameters and switching strategy of the power supply module are optimized according to the system running state and energy situation; by constructing the reward function , the algorithm learns the power management strategy, wherein is the energy utilization efficiency, is the system reliability, is the system stability, is the corresponding weight coefficient; Fault diagnosis and redundant power supply step: real-time monitoring of the running parameters of each module of the power supply system is realized through the distributed sensor network, the collected data is subjected to feature extraction, and a fault diagnosis model based on graph convolution network GCN is adopted for fault positioning and classification; at the same time, the redundant power supply module is started, and through formula The power supply power of the redundant module is calculated, wherein power is supplied to the redundant module, power is supplied to the fault module, a redundant adjustment coefficient is adjusted, a fault repair expected time is predicted, a redundant power module standby time is calculated, so that the system can operate stably in the case of failure.

[0008] Further, it further comprises: An energy recycling step: collecting waste heat, vibration energy and electromagnetic radiation energy generated by each module during system operation; using thermoelectric recovery modules, piezoelectric recovery modules and electromagnetic induction recovery modules; the thermoelectric recovery module converts waste heat into electrical energy using thermoelectric materials, and the conversion efficiency is affected by the temperature gradient and material characteristics, the formula is wherein is the output power, is the input heat energy, is the thermoelectric conversion enhancement coefficient, is the temperature gradient, is the average temperature; the piezoelectric recovery module converts vibration energy into electrical energy, and the conversion efficiency is related to the vibration frequency and amplitude, the formula is wherein is the output power, is the input vibration power, is the piezoelectric conversion adjustment coefficient, is the vibration frequency, is the resonance frequency of the piezoelectric material; the electromagnetic induction recovery module converts electromagnetic radiation energy into electrical energy, and the conversion efficiency is affected by the magnetic field strength and coil parameters, the formula is wherein is the output power, is the input electromagnetic radiation power, is the electromagnetic conversion enhancement coefficient, is the magnetic field strength, is the reference magnetic field strength; the recovered energy is stored in the energy storage module, and the energy management algorithm is used to optimize the energy recovery and storage process.

[0009] Further, in the energy prediction step, the DSTAN model is optimized, an adaptive feature fusion mechanism is introduced, the weights of each feature are automatically adjusted according to different environmental and seismic activity conditions; at the same time, a transfer learning strategy is used, the existing marine energy collection data and seismic activity data are used to pre-train the model, and then the model is fine-tuned in the target deep sea area, and the prediction error reduction formula is: wherein is the prediction error before optimization, is the feature weight, The error reduction amount brought by feature optimization.

[0010] Further, in the dynamic power distribution step, an elite reservation strategy and a cooperative evolution mechanism are introduced in the IMOPSO algorithm to improve the global search ability of the algorithm; meanwhile, the aging and performance attenuation factors of the power module are considered, and the power distribution formula is modified as , wherein is the aging coefficient, is the service time of the power module.

[0011] Further, in the low-power mode switching step, under the pulse power supply mode, according to the working characteristics and task requirements of the module, an adaptive pulse width modulation (APWM) technology is adopted to dynamically adjust the width and frequency of the pulse; the pulse period is calculated by the formula , wherein is the pulse period, is the equivalent capacitance of the module, is the working voltage, is the average working current, is the task change adjustment coefficient, is the task energy requirement change amount, is the time interval, and the power consumption is reduced under the premise of ensuring the function.

[0012] Further, in the power management optimization step, the PPO algorithm adopts a double network architecture including a policy network and a value network; a priority experience replay mechanism is introduced to improve the learning efficiency of the algorithm; meanwhile, a power management module cooperation strategy is adopted to optimize the operation of the power system, and the convergence time of the algorithm is shortened by the formula: , wherein is the convergence time before optimization, is the optimization strategy weight, is the time shortening amount brought by each optimization strategy.

[0013] Further, the fault diagnosis model adopts an attention mechanism-based graph convolution network (AGCN), which highlights the fault features through the attention mechanism, and establishes a fault knowledge base to classify and summarize the faults, achieving the fault diagnosis accuracy formula: , wherein is the number of correctly diagnosed faults, is the total number of faults, is the fault feature enhancement coefficient, is the key feature score of the th fault.

[0014] Further, in the energy recycling step, the energy recycling device is optimized and designed in an integrated and miniaturized structure; meanwhile, a control algorithm is used to monitor and adjust the energy recycling process in real time, and the recycling resources are dynamically allocated according to different energy sources and recycling efficiencies; the energy recycling efficiency improvement formula is: , wherein is the recycling efficiency after optimization, is the recycling efficiency before optimization, is a recycling efficiency enhancement coefficient, is the optimization effect score of the nth energy recycling mode, is the total recycled energy.

[0015] Further, in the system startup phase, the power module is initialized and configured according to the preset parameters of the deep sea environment, the initial requirements of the seismic detection task and the performance state of the power module, and the initial power supply power is calculated by formula , wherein is the initial power supply power, is the minimum working power of the nth module, is a startup time adjustment coefficient, is the system startup time, is the nominal startup time, is an aging influence coefficient, is the service time of the power module.

[0016] Compared with the prior art, the beneficial effects of the present application are: The system performance is significantly improved from multiple dimensions. In terms of energy harvesting, the multi-modal composite energy harvester integrates multiple harvesting technologies, which can fully utilize the weak light, temperature difference and wave energy in the deep sea, improve the stability and total amount of energy acquisition, and effectively solve the limitations of traditional single harvesting mode. The mixed energy storage mode combines lithium batteries and supercapacitors, taking into account high energy storage and high power output, and cooperating with an intelligent battery management system to optimize energy storage distribution and ensure stable power supply.

[0017] The dynamic power consumption distribution and low power consumption mode switching strategy enable the system to accurately adjust the power supply of each module according to energy supply, task demand and environmental changes, while ensuring key functions and minimizing power consumption. Compared with the traditional fixed power supply mode, the system running time is significantly prolonged, reducing the function limitation or shutdown caused by insufficient energy. The fault diagnosis and redundant power supply mechanism quickly and accurately locates faults and starts redundant power supply through advanced graph convolution network and scientific redundancy calculation, greatly improving the reliability and stability of the system, avoiding interruption of detection work due to power supply failure.

[0018] ​​The energy recycling technology effectively recovers and converts the waste heat and vibration energy generated by the system, further improving the energy utilization rate. Through deep reinforcement learning and other algorithms, the power management strategy is optimized, so that the system can continuously learn and improve itself in the complex and variable deep sea environment, realizing the intelligentization of power management. Overall, the patent technology comprehensively improves the energy utilization efficiency, reliability and working time of the deep sea seismic exploration system, providing strong technical support for deep sea seismic exploration research. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 A schematic block diagram of a modular power management and low-power optimization method for a deep sea seismic exploration system is provided for the present application. Figure 2 A schematic block diagram of a modular power management for a deep sea seismic exploration system is provided for the present application. Figure 3 A schematic block diagram of low-power optimization for a deep sea seismic exploration system is provided for the present application. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0021] In the description of the present application, it should be understood that the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application.

[0022] Moreover, the terms "first", "second", etc. are used herein only to describe different steps in the method, and are not used to denote or imply relative importance or a number of steps indicated. Thus, features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise explicitly and specifically defined. In addition, the terms "mounting", "connecting", "connection" should be broadly understood, for example, it can be fixed connection, or detachable connection, or integral connection; it can be mechanical connection, or electrical connection; it can be direct connection, or indirect connection through intermediate medium, or internal communication of two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances, and the present application will be further described in detail below with reference to the drawings.

[0023] Reference Figure 1 and Figure 3 : Embodiment of a modular power management and low power consumption optimization method of a deep-sea seismic detection system.

[0024] In the implementation of the modular power management and low power consumption optimization method of a deep-sea seismic detection system described in the present patent, detailed and operable technical landing of each step is required, and the following is the embodiment: Power module division step: In the deep-sea seismic detection system, the power system is clearly divided into four core modules. The energy collection module adopts a multi-modal composite energy collector. The improved thermoelectric generator sheet selects a new type of thermoelectric material based on bismuth telluride. In the typical 2-4℃ temperature difference environment of deep sea, according to the formula , , , , , ), the thermal-to-electric conversion efficiency can reach 12%; the self-adaptive wave energy conversion device is equipped with a hydraulic-electromagnetic conversion mechanism and an intelligent adjusting paddle. When the wave frequency is in the range of 0.5-2Hz and the amplitude is 1-3m, according to the formula , , , the conversion efficiency can be up to 35%; the micro-light solar cell panel adopts a perovskite-silicon stacked structure. Under the light intensity of 100-500lux in deep sea, according to the formula , , the power generation efficiency is about 8%. The energy storage module adopts a lithium battery pack of 48V, 100Ah and a super capacitor of 5F, 48V in a hybrid mode. The intelligent battery management system collects the voltage, current and temperature data of the battery and capacitor every 10 seconds, and optimizes the energy storage distribution through the fuzzy control algorithm.

[0025] Energy prediction step: A deep spatio-temporal attention network (DSTAN) model is built, and the input data of the model includes energy collection data in the past 30 days, real-time marine environmental parameters (temperature, salinity, flow rate, light intensity), and seismic activity related data such as seismic wave frequency and magnitude. The model training uses the Adam optimizer with a learning rate of 0.001 and is trained for 500 iterations. In actual prediction, the formula is used, and for a certain deep sea area, when the temperature is 3℃, the flow rate is 0.5m / s, and the seismic wave frequency is 1Hz, the energy collection amount in the next 6 hours is predicted. The actual measurement shows that the prediction error is controlled within 8%. At the same time, a rolling prediction is performed every 2 hours to update the model parameters. Dynamic power allocation step: The dynamic power allocation model based on multi-objective optimization uses an improved multi-objective particle swarm optimization algorithm (IMOPSO) with a particle swarm size of 50, a maximum number of iterations of 100, and an inertia weight linearly decreasing from 0.9 to 0.4. The system real-time collects the power demand of each module, such as the seismic data acquisition module demand power , data transmission module , etc., and combines the energy prediction results and task priority, seismic data acquisition priority , data transmission priority , to calculate the power supply of each module, and finally allocates 120W to the seismic data acquisition module and 80W to the data transmission module.

[0026] Low power mode switching step: The system real-time monitors the energy collection amount, and when , , , , , the conditions are met, the low power mode is triggered. A state machine-based switching strategy is used to close the non-critical system self-checking module, and the adaptive pulse width modulation (APWM) technology is used to power the critical seismic signal acquisition module, while its working voltage is reduced from 12V to 9V and the frequency is reduced from 100Hz to 50Hz, effectively reducing the overall power consumption of the system.

[0027] Power management optimization step: The voltage, current, temperature, fault information, etc. of each module of the power system are collected by the sensor every 5 seconds to build an experience pool. The proximal policy optimization algorithm (PPO) is used for strategy optimization, and the strategy network and value network both use a 3-layer fully connected neural network structure with 128, 64, and 32 neurons respectively. The reward function , , , , after 1000 training iterations, the algorithm converges, and the optimization of the power management strategy is achieved. In addition, in the fault diagnosis and redundant power supply step, the data collected by the distributed sensor network is input into the attention mechanism-based graph convolution network (AGCN) after feature extraction. The model contains 3 layers of graph convolution layers, with the number of convolution kernels being 32, 64, and 128 respectively, and the fault diagnosis accuracy reaches 95%. When a module failure is detected, such as a voltage conversion module failure, , according to the formula , , , , the redundant module is started to provide 60W power. In the energy recycling step, the thermoelectric recycling module uses semiconductor thermoelectric generator sheets. When the waste heat temperature difference is 10℃, according to the formula , , the recovery efficiency is about 15%; the piezoelectric recycling module has a vibration frequency of 10Hz and an amplitude of 0.5m. According to the formula , , , the recovery efficiency reaches 20%; the electromagnetic induction recycling module has a magnetic field strength of 0.1T. According to the formula , , , the recovery efficiency is 18%.

[0028] Beneficial effect representation and explanation:

[0029] From the above table data, it can be seen that the method of the present application is significantly better than the traditional method in many key performance indicators. In terms of energy collection, the multi-modal composite collector effectively solves the problem of unstable energy acquisition in the traditional method. The system running time is doubled, thanks to accurate energy prediction, dynamic power distribution and low-power mode switching, so that limited energy is more reasonably utilized. The power consumption is reduced by 40%, further prolonging the system working time.

[0030] The fault diagnosis accuracy is greatly improved, which ensures the reliability of the system and reduces the data loss and detection interruption caused by faults. The energy utilization rate is increased by 42%, and the energy recycling technology and optimized power management strategy play an important role. These data fully prove the significant advantages and practical application value of the method of the present application in improving the performance of the deep sea seismic exploration system.

[0031] The above is only a preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any skilled person in the art can make equivalent substitutions or changes within the technical scope disclosed by the present application according to the technical solutions and inventive concepts of the present application, which should be covered within the protection scope of the present application.

Claims

1. A modular power management and low-power optimization method for a deep-sea seismic detection system, characterized in that, include: Power module partitioning steps: The power system of the deep-sea seismic detection system is divided into an energy acquisition module, an energy storage module, a voltage conversion module, and a power distribution module; The energy harvesting module employs a composite energy harvester, integrating a thermoelectric generator, an adaptive wave energy conversion device, and a low-light solar panel. The thermoelectric generator utilizes thermoelectric materials, and its thermoelectric conversion efficiency is calculated using the formula: ,in , For Seebeck coefficients, For electrical conductivity, Thermal conductivity, Absolute temperature; The adaptive wave energy conversion device can automatically adjust the conversion efficiency according to different wave frequencies and amplitudes. The conversion efficiency formula is: ,in For output power, For input power, For adaptive adjustment coefficients, For wave frequency, The resonant frequency of the device; the low-light solar panel generates electricity in a weak light environment, and its power generation efficiency is affected by light intensity and spectral characteristics, as shown in the formula. ,in For conversion factors, Light intensity, Let be the spectral response function. Let be the spectral distribution function of the light source; The energy storage module employs a hybrid energy storage approach, combining lithium battery packs and supercapacitors with a battery management system. It optimizes energy allocation by real-time monitoring of battery and capacitor voltage, current, and temperature parameters using a fuzzy control algorithm, as shown in the formula. ,in Assigning energy storage capacity For energy storage of lithium batteries, For energy storage in supercapacitors, and These are weighting coefficients that are dynamically adjusted based on the system state. Energy prediction steps: A deep spatiotemporal attention network (DSTAN)-based energy prediction model is constructed using historical energy acquisition data, real-time marine environmental parameters, and seismic activity data; combined with deep-sea environmental parameters such as temperature, current velocity, and seismic wave frequency, the prediction is then performed using formulas. Energy harvesting volume prediction, among which To predict energy harvesting volume, For marine environmental parameters, For parameters related to seismic activity, and The coefficients of the corresponding parameters, This is the error term.

2. The modular power management and low-power optimization method for a deep-sea seismic detection system according to claim 1, characterized in that, Also includes: Dynamic power allocation steps: Establish an optimized dynamic power allocation model based on energy prediction results, real-time power consumption requirements of each module in the system, and task priorities; Through formula Calculate the power supply of each module, where Allocate power to the module. For the module's required power, For usable energy, Prioritize module tasks. For the first The power requirement of each module For the first The task priority of each module enables dynamic power allocation; Power mode switching steps: When the energy harvesting amount is lower than the set threshold, the system automatically enters the low power mode. A state machine-based low power mode switching strategy is introduced to dynamically select different combinations of low power modes according to different working states and task requirements of the system. In addition to shutting down functional modules and using pulse power supply, the voltage and frequency of the modules are also dynamically adjusted using formulas. While ensuring the basic functions of the system, power consumption is reduced, among which For power consumption, For load capacitance, For voltage, For frequency; the low power consumption judgment formula is set as follows: if Switching modes, among which For current energy harvesting, This is the proportionality coefficient. Energy threshold Adjustment factor for demand changes, This refers to the change in energy demand. For time intervals; Power management optimization steps: Real-time monitoring of the operating status of each module of the power system, and collection of voltage, current, temperature and fault information data; Utilizing the Proximal Policy Optimization (PPO) algorithm from deep reinforcement learning, combined with experience replay and target network update mechanisms, the operating parameters and switching strategies of the power module are optimized based on the system's operating status and energy conditions; a reward function is constructed. The algorithm is guided to learn power management strategies, in which For energy utilization efficiency, For system reliability, For system stability, These are the corresponding weighting coefficients; Fault diagnosis and redundant power supply steps: The operating parameters of each module in the power system are monitored in real time through a distributed sensor network. Features are extracted from the collected data, and a fault diagnosis model based on a graph convolutional network (GCN) is used for fault location and classification. Simultaneously, redundant power supply modules are activated, and the process is performed using the formula... Calculate the power supply of the redundant modules, where For redundant module power, Power of the faulty module. This is a redundancy adjustment factor. The estimated time for fault repair. This provides backup time for redundant power supply modules, enabling the system to operate stably in the event of a fault.

3. The modular power management and low-power optimization method for a deep-sea seismic detection system according to claim 1, characterized in that, Also includes: Energy recovery and reuse steps: Collect waste heat, vibration energy and electromagnetic radiation energy generated by each module during system operation; It employs thermoelectric recovery modules, piezoelectric recovery modules, and electromagnetic induction recovery modules; Thermoelectric recovery modules utilize thermoelectric materials to convert waste heat into electrical energy. The conversion efficiency is affected by the temperature gradient and material properties, as shown in the formula. ,in In order to output electrical energy, To input heat energy, Thermoelectric conversion enhancement coefficient, For temperature gradient, Average temperature; The piezoelectric energy recovery module converts vibration energy into electrical energy. The conversion efficiency is related to the vibration frequency and amplitude, as shown in the formula: ,in For output power, For input vibration power, This is the piezoelectric conversion adjustment coefficient. The vibration frequency, The resonant frequency of the piezoelectric material; the electromagnetic induction recovery module converts electromagnetic radiation energy into electrical energy, and the conversion efficiency is affected by the magnetic field strength and coil parameters, as shown in the formula. ,in For output power, For input electromagnetic radiation power, The electromagnetic conversion enhancement factor, The magnetic field strength, Using the reference magnetic field strength, the recovered energy is stored in the energy storage module, and energy management algorithms are used to optimize the energy recovery and storage process.

4. The modular power management and low-power optimization method for a deep-sea seismic detection system according to claim 1, characterized in that, In the energy prediction step, the DSTAN model is optimized by introducing an adaptive feature fusion mechanism to automatically adjust the weights of each feature according to different environmental and seismic activity conditions. Simultaneously, a transfer learning strategy is employed, using existing ocean energy harvesting data and seismic activity data to pre-train the model, followed by fine-tuning in the target deep-sea region. The prediction error is reduced using the following formula: ,in To optimize the previous prediction error, For feature weights, The amount of error reduction resulting from feature optimization.

5. The modular power management and low-power optimization method for a deep-sea seismic detection system according to claim 2, characterized in that, In the dynamic power allocation step, an elite retention strategy and a co-evolutionary mechanism are introduced into the IMOPSO algorithm to improve the algorithm's global search capability; at the same time, considering the aging and performance degradation factors of the power module, the power allocation formula is modified, and the formula is as follows: ,in The aging factor is... This refers to the usage time of the power module.

6. The modular power management and low-power optimization method for a deep-sea seismic detection system according to claim 2, characterized in that, In the low-power mode switching step, under pulse power supply mode, adaptive pulse width modulation (APWM) technology is used to dynamically adjust the pulse width and frequency according to the module's operating characteristics and task requirements; through the formula Calculate the pulse period, where For pulse period, The equivalent capacitance of the module. Operating voltage This is the average operating current. Adjust the coefficients according to changes in the task. This represents the change in mission energy requirements. The time interval is used to reduce power consumption while ensuring functionality.

7. The modular power management and low-power optimization method for a deep-sea seismic detection system according to claim 2, characterized in that, In the power management optimization step, the PPO algorithm adopts a dual-network architecture, including a policy network and a value network; it improves the algorithm's learning efficiency by introducing a priority experience replay mechanism; and it optimizes the operation of the power system using a collaborative strategy of power management modules. The algorithm's convergence time is shortened by the following formula: ,in To optimize the initial convergence time, To optimize strategy weights, The amount of time reduction brought about by each optimization strategy.

8. The modular power management and low-power optimization method for a deep-sea seismic detection system according to claim 2, characterized in that, The fault diagnosis model employs an attention-based graph convolutional network (AGCN) to highlight fault features. Simultaneously, a fault knowledge base is established to classify and summarize faults, resulting in the following fault diagnosis accuracy formula: ,in The number of correctly diagnosed faults, This represents the total number of faults. The fault characteristic enhancement coefficient, For the first Key feature scores for each fault.

9. A modular power management and low-power optimization method for a deep-sea seismic detection system according to claim 3, characterized in that, In the energy recovery and reuse step, the energy recovery device is optimized and adopts an integrated and miniaturized structure; at the same time, the energy recovery process is monitored and adjusted in real time using control algorithms, and the recovery resources are dynamically allocated according to different energy sources and recovery efficiencies. The formula for improving energy recovery efficiency is: ,in To optimize post-recycling efficiency, To optimize the efficiency of pre-recovery, The factor that enhances the recovery efficiency. For the first The optimization effect score of the energy recovery method, For total recovered energy.

10. A modular power management and low-power optimization method for a deep-sea seismic detection system according to claim 1, characterized in that, During the system startup phase, the power module is initialized and configured based on preset parameters of the deep-sea environment, the initial requirements of the seismic exploration mission, and the performance status of the power module, using formulas... Calculate the initial power supply, where This is the initial power supply. For the first Minimum operating power of each module This is a start-up time adjustment factor. For system startup time, Nominal startup time The aging effect coefficient. This refers to the usage time of the power module.

Images