Power management method and device based on dynamic load, electronic equipment and medium

Abstract

The invention provides a power management method and device based on a dynamic load, electronic equipment and a medium, and relates to the field of battery management of underwater cruise equipment.The method comprises the steps that a target time period is determined based on the target task type of tasks executed by underwater autonomous cruise equipment in the historical process; inputting the current and the voltage of the target time period into the trained hybrid prediction model for processing, and outputting a multi-scale load prediction curve of a future time period; and determining at least one target power supply strategy of the underwater autonomous cruise equipment in the future time period based on the multi-scale load prediction curve. According to the application, a domestic CPCI mainboard is taken as a core carrier, multi-dimensional data such as CPU / FPGA core voltage and current are acquired in real time through a high-precision sensor, composite characteristic quantities such as a dynamic peak factor (DPF) and a load change acceleration (LCA) are introduced, the limitation of a traditional mean variance index is broken through, and the cruising ability and the response speed of underwater autonomous cruising equipment can be improved.

Description

Technical Field

[0001] The present application relates to the field of battery management for underwater cruising devices. Specifically, it relates to a power management method, device, electronic device, and medium based on dynamic load. Background Art

[0002] For a long time, the power management of underwater autonomous cruising devices mainly relied on a method that combined static threshold control and a single prediction model. This method was usually based on imported hardware design and used general manufacturing processes. This traditional power management mode achieved the conversion between different power supply modes by setting fixed power thresholds. Although this method had simple logic and was easy to implement, in the face of sudden high-load tasks, due to the lack of effective dynamic adjustment ability, it often led to two extreme situations: one was that the power supply was too redundant, resulting in waste; the other was that the instantaneous power supply was insufficient, affecting task execution. In addition, frequently switching the working mode of the battery according to different task requirements would also cause unnecessary power consumption. Therefore, it was difficult for the existing technology to effectively improve the endurance and response speed of underwater autonomous cruising devices. Summary of the Invention

[0003] The purpose of the embodiments of the present application is to provide a power management method, device, electronic device, and medium based on dynamic load, which solves the above problems existing in the existing technology and can improve the endurance and response speed of underwater autonomous cruising devices.

[0004] In a first aspect, a power management method based on dynamic load is provided, which is applied to a processor of a power management system. The processor is integrated on a domestic CPCI motherboard of an underwater autonomous cruising device. The method may include: Determine a target time period based on the target task type of the underwater autonomous cruising device performing tasks in the historical process; Obtain the current and voltage of the target time period; Input the current and voltage of the target time period into a trained hybrid prediction model for processing, and output a multi-scale load prediction curve for a future time period; the hybrid prediction model is used to extract features of the power consumption of the underwater autonomous cruising device in the target time period and predict the hybrid predicted power consumption for a future time period; Based on the multi-scale load prediction curve, determine at least one target power supply strategy for the underwater autonomous cruising device in the future time period.

[0005] In a possible implementation, determining a target time period based on the target task type of the underwater autonomous cruising device performing tasks in the historical process includes: Based on the corresponding relationship between different task types and different time periods in the configuration, determine the target time period corresponding to the target task type; wherein, the task types include burst tasks and periodic tasks; the target time period corresponding to the burst task is a short-term window; the target time period corresponding to the periodic task is a long-term window.

[0006] In a possible implementation, the hybrid prediction model includes a first feature processing module, a second feature processing module, a hybrid power consumption processing module, and an output module.

[0007] In a possible implementation, input the current and voltage of the target time period into the trained hybrid prediction model for processing, and output a multi-scale load prediction curve for the future time period, including: Based on the first current and the first voltage of the short-term window corresponding to the burst task, determine the dynamic peak factor, the load change acceleration, and the high power consumption ratio; Input the dynamic peak factor, the load change acceleration, and the high power consumption ratio into the first feature processing module for processing to obtain the short-term predicted power consumption; Based on the second current and the second voltage of the long-term window corresponding to the periodic task, determine the power consumption mean value and the power consumption variance; Input the power consumption mean value and the power consumption variance into the second feature processing module for processing to obtain the long-term predicted power consumption; Input the short-term predicted power consumption and the long-term predicted power consumption into the hybrid power consumption processing module for processing to obtain the hybrid predicted power consumption; Input the hybrid predicted power consumption into the output module for processing to obtain the multi-scale load prediction curve for the future time period; wherein, the multi-scale load prediction curve is a relationship curve between the future time period and the hybrid predicted power consumption.

[0008] In a possible implementation, input the short-term predicted power consumption and the long-term predicted power consumption into the hybrid power consumption processing module for processing to obtain the hybrid predicted power consumption, including: Based on the configured short-term weight coefficient and long-term weight coefficient, fuse the short-term predicted power consumption and the long-term predicted power consumption to obtain the hybrid predicted power consumption; wherein, the short-term weight coefficient is calculated and determined by using a preset algorithm for the load change acceleration; the sum of the short-term weight coefficient and the long-term weight coefficient is 1.

[0009] In a possible implementation, the preset algorithm is:

[0010] Wherein, is the short-term weight coefficient, S is the load change acceleration, S 0is the configured load threshold, j is the configured adjustment coefficient, and e is the base of the natural logarithm.

[0011] In a possible implementation, based on the multi-scale load prediction curve, determining at least one target power supply strategy for the underwater autonomous cruising device in a future period includes: For any target power consumption interval composed of the mixed predicted power consumptions in the multi-scale load prediction curve, based on the corresponding relationship between different power consumption intervals and different power consumption modes configured, determining the target mode corresponding to the target power consumption interval; the power consumption modes include an extreme energy-saving mode, a balanced mode, and a high-performance mode; Based on the corresponding relationship between different power consumption modes and different power supply strategies, determining the target power supply strategy corresponding to the target mode.

[0012] In a second aspect, a power management device based on dynamic load is provided, which is applied to a processor of a power management system. The processor is integrated on a domestic CPCI motherboard of an underwater autonomous cruising device. The device may include: A determination unit, configured to determine a target period based on the target task type of the underwater autonomous cruising device performing tasks in a historical process; An acquisition unit, configured to acquire the current and voltage in the target period; A processing unit, configured to input the current and voltage in the target period into a trained hybrid prediction model for processing, and output a multi-scale load prediction curve in a future period; the hybrid prediction model is used to extract features of the power consumption of the underwater autonomous cruising device in the target period and predict the hybrid predicted power consumption in a future period; The determination unit is further configured to determine at least one target power supply strategy for the underwater autonomous cruising device in a future period based on the multi-scale load prediction curve.

[0013] In a third aspect, an electronic device is provided. The electronic device includes a processor, a communication interface, a memory, and a communication bus. Among them, the processor, the communication interface, and the memory complete communication with each other through the communication bus; The memory is used to store a computer program; The processor, when executing the program stored on the memory, implements any of the method steps in the first aspect described above.

[0014] In a fourth aspect, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements any of the method steps in the first aspect described above.

[0015] This application provides a power management method based on dynamic load. With a domestic CPCI motherboard as the core carrier, multi-dimensional data such as the core voltage and current of the CPU / FPGA are collected in real time through high-precision sensors. Composite feature quantities such as the dynamic peak factor (DPF) and load change acceleration (LCA) are innovatively introduced to break through the limitations of traditional mean-variance indicators. Based on the feature data, a hybrid prediction model is used to achieve multi-time-scale load prediction. The prediction results are input into the model predictive control (MPC) algorithm. With the prediction results of the next three time steps as constraints, the optimal solution for voltage and frequency regulation is solved, and the domestic PMU chip is driven to execute a multi-mode power supply strategy: when the predicted load is lower than the threshold, it switches to the extreme energy-saving mode and shuts off the power supply to non-core modules; in the stage of sharp load increase, the standby battery unit is automatically enabled and the FPGA main frequency is increased to the safe extreme value. At the same time, the prediction error risk is avoided through the preset confidence interval mechanism. To achieve deep adaptation of domestic hardware, an SPI communication protocol is customized for the power chip, the instruction packing timing is optimized, and the control delay is compressed to the sub-millisecond level. The prediction model parameters are continuously updated through an online learning mechanism. By comparing the actual power consumption data fed back by the PMU with the prediction results, the network weights are dynamically corrected to form a full-process closed-loop control of "perception - prediction - execution - feedback". BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings required for use in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0017] Figure 1 It is a schematic flowchart of a power management method based on dynamic load provided by an embodiment of this application; Figure 2 It is a schematic structural diagram of a power management device based on dynamic load provided by an embodiment of this application; Figure 3 It is a schematic structural diagram of an electronic device provided by an embodiment of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of this application.

[0019] For a long time, the power management of underwater autonomous cruising devices has mainly relied on a method that combines static threshold control and a single prediction model. This method is usually based on imported hardware design and uses general manufacturing processes. This traditional power management mode realizes the conversion between different power supply modes by setting fixed power thresholds (such as CPU load > 70% triggers the high-performance mode). Although this method has simple logic and is easy to implement, in the face of sudden high-load tasks, due to the lack of effective dynamic adjustment capabilities, it often leads to two extreme situations: one is that the power supply is overly redundant, resulting in waste; the other is that the instantaneous power supply is insufficient, affecting task execution. In addition, frequently switching the battery's working mode according to different task requirements will also cause unnecessary power loss (15% - 20% power loss).

[0020] At the hardware level, imported PMU chips (such as LTC2977 of ADI) and standardized packaging schemes are often directly adopted. Their communication protocols have poor compatibility with the SPI bus timing of domestic CPCI motherboards, and the control instruction delay is up to more than 5ms. Moreover, the general packaging process is not optimized for the deep-sea high-pressure environment, and the error rate exceeds 0.5% at a water depth of 1000 meters (10MPa pressure), seriously restricting the reliability of the equipment. In addition, existing security mechanisms mostly rely on software fusing and general encryption algorithms (such as AES). The fusing response delay exceeds 100μs, and the encryption process is not adapted to domestic cryptographic chips, making it difficult to resist high-intensity underwater signal interference and hardware-level attacks. These technical defects together lead to systematic bottlenecks such as short endurance, slow response, high failure rate, and weak security for underwater autonomous cruising devices in the traditional scheme, becoming the core obstacle restricting the large-scale application of underwater autonomous cruisers.

[0021] Therefore, this application provides a power management method based on dynamic load to solve the above problems existing in the prior art and improve the endurance and response speed of underwater autonomous cruising devices.

[0022] The following describes the preferred embodiments of this application in conjunction with the accompanying drawings of the specification. It should be understood that the preferred embodiments described herein are only used to illustrate and explain this application and are not used to limit this application. And without conflict, the embodiments in this application and the features in the embodiments can be combined with each other.

[0023] Figure 1 It is a flowchart of a power management method based on dynamic load provided by an embodiment of this application. This method can be applied to the processor of a power management system, and the processor is integrated on the domestic CPCI motherboard of an underwater autonomous cruising device. As Figure 1 shown, this method may include: Step S110: Determine the target time period based on the target task type of the tasks executed by the underwater autonomous cruising device in the historical process.

[0024] Specifically, first, based on the relevant data of task execution in the historical process, determine the target task type of the task; where the task types include emergency tasks and periodic tasks. After that, based on the corresponding relationship between different task types and different time periods configured, determine the target time period corresponding to the target task type.

[0025] The target time period corresponding to the emergency task is a short-term window; for example: emergency tasks such as obstacle avoidance tasks and emergency detection.

[0026] The target time period corresponding to the periodic task is a long-term window; for example: detection tasks, sleep tasks, etc.

[0027] Generally, the short-term window is set to 60s, and the short-term window is divided into multiple first sub-windows according to the configured first preset interval; the first preset interval is configured to be 10s.

[0028] Generally, the long-term window is set to 6h, and the long-term window is divided into multiple second sub-windows according to the configured second preset interval; the second preset interval is configured to be 1h.

[0029] This method is customized for domestic CPCI motherboards, streamlines the SPI instruction protocol, optimizes the register read / write timing of the PMU chip, and compresses the control delay to the sub-millisecond level.

[0030] Step S120: Obtain the current and voltage of the target time period.

[0031] Specifically, obtain the CPU / FPGA core voltage (0.8 - 1.2V), current (0.5 - 5A), and temperature (-5°C to 60°C) of the underwater autonomous cruising device during task execution in the target time period through the hardware status sensor.

[0032] Since there may be abnormal data during the actual acquisition of voltage, current, and temperature, the median filtering algorithm can be used to replace the abnormal data. The median filtering algorithm can be:

[0033] where P is the abnormal data, is the target data for replacing the abnormal data at time t.

[0034] For example: at time t, there is a phenomenon of a sharp increase in current. At this time, P is the abnormal current. It can be understood that time t is any moment in the target time period.

[0035] In some embodiments, it is also possible to obtain the water depth pressure (0 - 10 MPa) and water temperature (2 - 30 °C) during a target period collected by environmental sensors. The water depth pressure and water temperature during the collected target period can be used to correct the battery hardware performance degradation model. Specifically, first, a standard battery performance degradation model that does not consider external environmental factors is established. This model usually includes variables such as the number of charge and discharge cycles and the service life to predict the battery capacity degradation. Analyze the relationship between the water depth pressure, water temperature and battery performance parameters through statistical methods or machine learning algorithms to determine how they act on the battery performance individually and jointly. Add a correction coefficient determined by the water depth pressure and water temperature to the original degradation model. For example, a function f(Q, T) can be defined, where Q represents the water depth pressure and T represents the water temperature, and the output of this function is the adjustment ratio of the battery performance degradation rate under the current environmental conditions.

[0036] After that, since the domestic CPCI motherboard is equipped with an ADC module, this ADC module can sample the input analog signal at a frequency of 1 kHz and convert it into digital form. Among them, the analog signal refers to the data collected by each sensor.

[0037] Step S130: Input the current and voltage during the target period into the trained hybrid prediction model for processing, and output the multi-scale load prediction curve for the future period.

[0038] Among them, the hybrid prediction model includes a first feature processing module, a second feature processing module, a hybrid power consumption processing module and an output module.

[0039] Specifically, based on the first current and first voltage of the short-term window corresponding to the burst task, determine the dynamic peak factor, load change acceleration and high power consumption ratio; The calculation expression of the dynamic peak factor DPF is: ; where is the peak load power within the window, that is, the maximum value among all sampling points within the window, reflecting the instantaneous highest load, is the average power consumption, is the standard deviation of the load power within the window, quantifying the amplitude of load fluctuations; , , The i-th power consumption, and the i-th power consumption is determined by the first current and first voltage of the corresponding first sub-window; N is the number of first sub-windows of the short-term window, and this number is determined according to the first preset interval and the total duration of the short-term window.

[0040] The calculation expression of the load change acceleration S is: S .

[0041] The calculation expression of the high power consumption ratio HPR is: , where I is an indicator function.

[0042] It should be noted that the dynamic peak factor can reflect the intensity of sudden load increase; the load change acceleration characterizes the severity of fluctuations. If the load change acceleration is large, it indicates that the load has experienced rapid growth or decline in a short period, which may mean that the system needs to quickly adjust to adapt to this change. A high load change acceleration often requires the system to have a high response speed and flexibility to promptly address potential instability factors; the high power consumption ratio is used to identify continuous high-load phases. Identifying continuous high-load phases can help understand which time periods or operations result in the highest energy consumption, thus enabling targeted optimization.

[0043] Input the dynamic peak factor, load change acceleration, and high power consumption ratio into the first feature processing module for processing to obtain the short-term predicted power consumption; among them, the first feature processing module can be an LSTM network module. Specifically, update the data state of the dynamic peak factor, load change acceleration, and high power consumption ratio of each first sub-window through the forget gate, input gate, and output gate mechanisms inside the first feature processing module, so as to learn the dependency relationships in the data corresponding to each first sub-window in the short-term window. After passing through multiple first sub-windows, the first feature processing module can capture the dynamic patterns in the data, such as sudden high peaks, rapid load change trends, and the proportion of high-power consumption periods, etc. Finally, the first feature processing module generates the short-term predicted power consumption representing the short-term window through its output layer.

[0044] Based on the second current and second voltage of the long-term window corresponding to the periodic task, determine the power consumption mean and power consumption variance of the long-term window; specifically, the method for solving the power consumption mean of the long-term window is the same as that of the short-term window.

[0045] Input the power consumption mean and power consumption variance of the long-term window into the second feature processing module for processing to obtain the long-term predicted power consumption; among them, the second feature processing module can be an ARIMA network module, configured with a differencing order d = 1, autoregressive term p = 2, and moving average term q = 1.

[0046] Input the short-term predicted power consumption and long-term predicted power consumption into the hybrid power consumption processing module for processing to obtain the hybrid predicted power consumption; specifically, the algorithm adopted by the hybrid power consumption processing module is:

[0047] where is the hybrid predicted power consumption for the future time period t + 1, For short-term power consumption prediction, k defines the time range of the input data of the LSTM model, that is, the historical data at a preset interval is traced back k sub-windows (the first sub-window or the second sub-window) from the current moment t (example: the preset interval of the sub-window is 10s, and k is 6, indicating that the historical data of 6*10s = 60s will be traced back). For long-term power consumption prediction, m is the number of the second sub-windows within the long-term window. is the short-term weight coefficient. is the long-term weight coefficient.

[0048] Among them, the short-term weight coefficient is calculated and determined by using a preset algorithm for the load change acceleration.

[0049] The preset algorithm can be:

[0050]

[0051] Among them, is the short-term weight coefficient, S is the load change acceleration, S 0 is the configured load threshold, generally set to 0.5, j is the configured adjustment coefficient, generally set to 2, e is the base of the natural logarithm, when the emergency obstacle avoidance task is performed, S↑→α↑, strengthening the short-term weight coefficient.

[0052] The mixed predicted power consumption is input to the output module for processing to obtain a multi-scale load prediction curve for the future time period; among them, the multi-scale load prediction curve is a relationship curve between the future time period and the mixed predicted power consumption; the confidence interval (such as the upper and lower limits of the power consumption corresponding to a 95% confidence level) is marked in the multi-scale load prediction curve.

[0053] This method extracts features through feature extraction in the multi-scale dynamic load prediction technology, that is, uses the sliding window algorithm to extract short-term (60 seconds) load fluctuation features (such as the dynamic peak factor DPF, load change acceleration LCA) and long-term (6 hours) trend features, breaking through the limitations of traditional mean analysis, and through a hybrid prediction method that combines LSTM (capturing time series mutations) and ARIMA (fitting periodic trends), dynamically adjusts the model weights through task priorities (such as the LSTM weight is increased to 70% in the obstacle avoidance task), so that the prediction error is greatly reduced compared with a single model.

[0054] Step S140, based on the multi-scale load prediction curve, determine at least one target power supply strategy for the underwater autonomous cruising device in the future time period.

[0055] Specifically, according to the corresponding relationship between the configured different power consumption modes and different power consumption intervals, determine the target mode of any mixed predicted power consumption interval in the multi-scale load prediction curve, so as to determine the target power supply strategy corresponding to the target mode.

[0056] Among them, the correspondence between different power consumption modes and different power consumption ranges is as follows:

[0057] The above process can also be understood as:

[0058] Among them, is the maximum load power allowed by the system (hardware safety threshold).

[0059] After determining the power supply strategy, the method further includes: based on the Model Predictive Control (MPC) algorithm, optimizing the hybrid predicted power consumption to obtain the target voltage and target frequency.

[0060] The Model Predictive Control (MPC) algorithm is:

[0061] Among them, the constraint conditions for the target voltage V and target frequency f are: ; is the predicted load power for the i-th future sub-window, is the actual load power for the i-th future sub-window, is the short-term weight coefficient, is the long-term weight coefficient.

[0062] The adaptive power closed-loop control system in the above manner provides a multi-mode power supply strategy: switching between extreme energy-saving, balanced, and high-performance modes according to the prediction results, and dynamically adjusting the power supply redundancy in combination with the confidence interval mechanism (such as 95% confidence level); MPC real-time optimization: using the predicted values for the next 3 time steps as constraints to solve the optimal voltage / frequency parameters to ensure that the power supply strategy takes into account both efficiency and safety.

[0063] After that, the target voltage and target frequency are packaged into PMU control instructions and transmitted to the power chip through the SPI bus of the domestic CPCI motherboard.

[0064] In some embodiments, the PMU transmits the actually output voltage, current, and temperature data in real time and compares them with the target voltage and target current.

[0065] If the actual values for 3 consecutive time steps exceed the prediction confidence interval, trigger the model online learning mechanism to update the weights of the first feature processing module.

[0066] The model predictive control (MPC) algorithm uses the prediction results of the next three time steps as constraints to dynamically solve the optimal voltage / frequency solution, enabling the power supply strategy to match the load demand in real time and avoiding "over-power supply" or "power supply lag". The weights of the first feature processing module are dynamically updated through the real-time feedback data of the PMU, forming a "perception-prediction-execution-feedback" closed loop to continuously improve the model prediction accuracy and system robustness.

[0067] The present application can achieve the following technical effects: 1. The traditional fixed-threshold power supply strategy cannot adapt to sudden loads in tasks such as sonar detection and obstacle avoidance (e.g., the power consumption instantaneously increases by 200%), resulting in energy waste or insufficient power supply and restricting the long-endurance ability. The hybrid prediction model of the present application captures the characteristics of sudden loads (such as the dynamic peak factor DPF) through a short-term window (60 seconds) and fits the trend of periodic tasks through a long-term window (6 hours). The prediction error is significantly reduced compared to a single model, accurately predicting future load demands; the present application saves 25% of the total task energy consumption and can still maintain stable power supply (fluctuation < 5%) in high-load scenarios.

[0068] 2. The traditional solution has a high response delay, and tasks such as emergency obstacle avoidance are prone to action failures due to power supply delays. The sliding window of the present application quickly extracts the load change acceleration (LCA) at a 10-second step and combines it with a dynamic weight allocation mechanism (the LSTM weight is increased to 70% in emergency tasks), enabling the hybrid prediction model to focus on short-term mutations and compressing the response delay to the sub-millisecond level; the optimization of domestic PMU instructions further compresses the control delay by streamlining the SPI protocol data packet and optimizing the register read / write timing, ensuring instantaneous execution of power supply instructions. In the obstacle avoidance task, the power supply response delay is reduced throughout the process, successfully avoiding most action failures caused by delays.

[0069] 3. The fixed model parameters of the traditional solution cannot adapt to the load characteristic differences of different task modes (such as scientific research detection, patrol). The online learning mechanism of the present application compares the actual power consumption data and prediction results fed back by the PMU to dynamically update the network weights (such as adaptively adjusting the learning rate), enabling the model to be continuously optimized; the multi-mode strategy stores the power supply parameter templates for different tasks (such as long-term detection, intensive obstacle avoidance), and the system automatically matches the optimal strategy according to the input of the task manager. In cross-task scenario tests, the prediction error can be adaptively optimized and decreased, and it can adapt to new task modes without manual intervention.

[0070] Corresponding to the above method, the embodiment of the present application further provides a power management device based on dynamic loads, as Figure 2 shown. The device includes: A determination unit 210, configured to determine a target time period based on the target task type of the underwater autonomous cruising device performing tasks in the historical process; An acquisition unit 220, configured to acquire the current and voltage during a target period; A processing unit 230, configured to input the current and voltage during the target period into a trained hybrid prediction model for processing, and output a multi-scale load prediction curve for a future period; the hybrid prediction model is used to extract features of the power consumption of the underwater autonomous cruising device during the target period, and predict the hybrid predicted power consumption for a future period; The determination unit 210 is further configured to determine at least one target power supply strategy of the underwater autonomous cruising device for a future period based on the multi-scale load prediction curve.

[0071] The functions of the various functional units of a power management device based on dynamic load provided in the above embodiments of the present application can be implemented by the above method steps. Therefore, the specific working processes and beneficial effects of each unit in a power management device based on dynamic load provided in the embodiments of the present application will not be repeated here.

[0072] Embodiments of the present application further provide an electronic device, as Figure 3 shown, including a processor 310, a communication interface 320, a memory 330, and a communication bus 340. Among them, the processor 310, the communication interface 320, and the memory 330 complete communication with each other through the communication bus 340.

[0073] The memory 330 is used to store a computer program; The processor 310, when executing the program stored on the memory 330, implements the following steps: Determine a target period based on the target task type of the underwater autonomous cruising device performing tasks during a historical process; Acquire the current and voltage during the target period; Input the current and voltage during the target period into a trained hybrid prediction model for processing, and output a multi-scale load prediction curve for a future period; the hybrid prediction model is used to extract features of the power consumption of the underwater autonomous cruising device during the target period, and predict the hybrid predicted power consumption for a future period; Based on the multi-scale load prediction curve, determine at least one target power supply strategy of the underwater autonomous cruising device for a future period.

[0074] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The communication bus can be divided into an address bus, a data bus, a control bus, etc. For the sake of convenience in representation, only a thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0075] The communication interface is used for communication between the above-mentioned electronic device and other devices.

[0076] The memory can include a Random Access Memory (RAM), or can also include a Non-Volatile Memory (NVM), such as at least one disk memory. Optionally, the memory can also be at least one storage device located far from the aforementioned processor.

[0077] The above-mentioned processor can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0078] Since the implementation manners and beneficial effects of each device of the electronic device in the above embodiments can be seen from Figure 1 the steps in the embodiments shown, therefore, the specific working process and beneficial effects of the electronic device provided in the embodiments of the present application will not be repeated here.

[0079] In another embodiment provided by the present application, a computer-readable storage medium is also provided. Instructions are stored in the computer-readable storage medium. When it runs on a computer, it causes the computer to execute any one of the above-mentioned power management methods based on dynamic load.

[0080] In another embodiment provided by the present application, a computer program product containing instructions is also provided. When it runs on a computer, it causes the computer to execute any one of the above-mentioned power management methods based on dynamic load.

[0081] Those skilled in the art should understand that the embodiments in the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the embodiments in the embodiments of the present application can be in the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the embodiments in the embodiments of the present application can be in the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) that contain computer-usable program code.

[0082] The embodiments in the embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments in the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0083] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means realizes the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0084] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for realizing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0085] Unless otherwise defined, the technical terms or scientific terms used in this application shall have the ordinary meanings as understood by those of ordinary skill in the art to which this invention pertains. The terms "first", "second" and similar words used in this application do not denote any order, quantity or importance, but are only used to distinguish different components. Words such as "comprising" or "including" mean that the elements or items appearing before this word cover the elements or items listed after this word and their equivalents, without excluding other elements or items. Words such as "connected", "coupled" or "linked" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Upper", "lower", "left", "right", etc. are only used to indicate relative positional relationships, and when the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0086] Although the preferred embodiments in the embodiments of this application have been described, those skilled in the art can make additional changes and modifications once they know the basic creative concepts. Therefore, the embodiments of this application are intended to be construed as including the preferred embodiments and all changes and modifications falling within the scope of the embodiments of this application.

[0087] Obviously, those skilled in the art can make various changes and modifications to the embodiments in the embodiments of this application without departing from the spirit and scope of the embodiments in the embodiments of this application. Thus, if these modifications and variations of the embodiments in the embodiments of this application fall within the scope of the embodiments of this application and their equivalent technologies, the embodiments of this application also intend to include these changes and modifications.

Claims

1. A power management method based on dynamic load, characterized in that: A processor applied to a power management system, the processor being integrated on a domestic CPCI mainboard of an underwater autonomous cruise device, the method comprising: Determine the target time period based on the target task type of the underwater autonomous cruise equipment in the historical process of performing tasks; Obtain the current and voltage during the target period; The current and voltage of the target period are input into the trained hybrid prediction model for processing, and a multi-scale load prediction curve for a future period is output; the hybrid prediction model is used to extract features of the power consumption of the underwater autonomous cruising equipment in the target period, and predict the hybrid predicted power consumption in the future period; Based on the multi-scale load prediction curve, at least one target power supply strategy for the underwater autonomous cruising device in a future period is determined.

2. The method according to claim 1, characterized in that Based on the target task type of the underwater autonomous cruise equipment in the historical process of performing tasks, the target period is determined, including: Based on the correspondence between different configured task types and different time periods, the target time period corresponding to the target task type is determined; wherein the task types include burst tasks and periodic tasks; the target time period corresponding to the burst task is a short-term window; and the target time period corresponding to the periodic task is a long-term window.

3. The method according to claim 2, characterized in that The hybrid prediction model includes a first feature processing module, a second feature processing module, a hybrid power consumption processing module and an output module.

4. The method according to claim 3, characterized in that The current and voltage of the target period are input into the trained hybrid prediction model for processing, and a multi-scale load prediction curve for the future period is output, including: Determine a dynamic peak factor, a load change acceleration, and a high power consumption ratio based on a first current and a first voltage in a short-term window corresponding to the burst task; Inputting the dynamic peak factor, the load change acceleration and the high power consumption ratio into the first feature processing module for processing to obtain short-term predicted power consumption; Determine a power consumption mean and a power consumption variance based on a second current and a second voltage in a long-term window corresponding to the periodic task; Inputting the power consumption mean and the power consumption variance into the second feature processing module for processing to obtain long-term predicted power consumption; Inputting the short-term predicted power consumption and the long-term predicted power consumption into the mixed power consumption processing module for processing to obtain mixed predicted power consumption; The hybrid predicted power consumption is input into the output module for processing to obtain a multi-scale load prediction curve for the future period; wherein the multi-scale load prediction curve is a relationship curve between the future period and the hybrid predicted power consumption.

5. The method according to claim 4, characterized in that Inputting the short-term predicted power consumption and the long-term predicted power consumption into the mixed power consumption processing module for processing to obtain mixed predicted power consumption includes: Based on the configured short-term weight coefficient and long-term weight coefficient, the short-term predicted power consumption and the long-term predicted power consumption are merged to obtain the mixed predicted power consumption; wherein, the short-term weight coefficient is determined by calculating the load change acceleration using a preset algorithm; the sum of the short-term weight coefficient and the long-term weight coefficient is 1.

6. The method according to claim 5, characterized in that The preset algorithm is: in, is the short-term weight coefficient, S is the load change acceleration, S0 is the configured load threshold, j is the configured adjustment coefficient, and e is the base of the natural logarithm.

7. The method according to claim 1, characterized in that Determining at least one target power supply strategy for the underwater autonomous cruising device in a future period based on the multi-scale load prediction curve includes: For any target power consumption interval of each mixed predicted power consumption composition in the multi-scale load prediction curve, based on the correspondence between different configured power consumption intervals and different power consumption modes, determine the target mode corresponding to the target power consumption interval; the power consumption mode includes an extreme energy saving mode, a balanced mode and a high performance mode; Based on the correspondence between different power consumption modes and different power supply strategies, a target power supply strategy corresponding to the target mode is determined.

8. A power management device based on dynamic load, characterized in that: A processor used in a power management system, the processor being integrated on a domestic CPCI mainboard of an underwater autonomous cruise device, the device comprising: A determination unit, for determining a target period based on a target task type of a task performed by the underwater autonomous cruise device in a historical process; An acquisition unit, used for acquiring the current and voltage in a target period; A processing unit, used to input the current and voltage of the target period into a trained hybrid prediction model for processing, and output a multi-scale load prediction curve for a future period; the hybrid prediction model is used to extract features of the power consumption of the underwater autonomous cruising equipment in the target period, and predict the hybrid predicted power consumption in the future period; The determination unit is further used to determine at least one target power supply strategy for the underwater autonomous cruising device in a future period based on the multi-scale load prediction curve.

9. An electronic device, characterized in that: The electronic device comprises a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; Memory, used to store computer programs; A processor, for implementing the method steps described in any one of claims 1 to 7 when executing a program stored in a memory.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method steps described in any one of claims 1 to 7 are implemented.

Images