Solar water pump control method and system based on combination of voltage compensation and LSTM

Abstract

The invention belongs to the technical field of solar water pump control systems, and particularly relates to a solar water pump control method and system based on combination of voltage compensation and LSTM (Long Short Term Memory), which fully considers the influence of factors such as environment illumination change and voltage fluctuation of an energy storage device on the system by combining a voltage compensation mechanism and LSTM prediction control parameters. Higher real-time adaptability and higher control precision are realized; moreover, when the control parameters are determined, the input of the controller at the last time node, the error value of the control system and the output of the solar water pump system are used for calculation, so that the control accuracy and efficiency are improved.

Description

Technical Field

[0001] This invention belongs to the technical field of solar water pump control systems, and particularly relates to a solar water pump control method and system based on voltage compensation and LSTM. Background Technology

[0002] Solar-powered water pump systems, utilizing clean energy, are widely used in agricultural irrigation, landscaping, and water supply in remote areas, offering significant environmental and economic advantages. However, the output voltage of solar panels is greatly affected by variations in sunlight intensity, leading to voltage instability and consequently impacting the normal operation of the water pumps. Under sufficient sunlight, the energy storage device voltage can efficiently drive the pumps, but in insufficient sunlight (such as on cloudy days or at night), the voltage drops, causing the pumps to fail to start or interrupt operation. Traditional solar-powered water pump control systems mostly employ controllers with fixed parameters. This control method cannot adapt to dynamic environmental changes, resulting in lag in pump response, low efficiency, and system interruptions and instability during periods of insufficient sunlight. To address this issue, many improvement methods attempt to optimize controller parameters using intelligent algorithms such as fuzzy control and genetic algorithms. However, these methods typically cannot fully consider deep learning of time-series data and lack real-time adaptability to dynamic environments, limiting their application in complex environments.

[0003] In recent years, methods based on Long Short-Term Memory (LSTM) networks have demonstrated significant advantages in processing time-series data. LSTMs can capture long-term dependencies and are particularly suitable for dynamically changing environments, thus finding widespread application in the field of intelligent control. Compared to traditional controllers, LSTM-based control systems can dynamically predict control parameters based on real-time data, adapting to environmental changes to a certain extent and achieving more precise and flexible control. However, in practice, it has been found that to obtain reliable long-term predictions, LSTMs must be sufficiently trained on historical data containing various lighting conditions, ambient temperatures, and load variations. Insufficient training data leads to a decline in control system performance. Furthermore, the limited computational performance of local devices results in weak model training timeliness, leading to insufficient stability of the control system when the voltage of the energy storage device or the environment changes. Moreover, the control effect of LSTM-based control methods depends entirely on the output of the LSTM, exhibiting strong model dependence and failing to fully utilize prior control parameters. Summary of the Invention

[0004] Based on the above background, this invention proposes a solar water pump control method and system based on voltage compensation and LSTM. The battery management module dynamically monitors the voltage and load current of the solar panels. When sunlight is insufficient, the battery compensation mechanism automatically activates battery discharge to maintain voltage, ensuring the water pump can continue to operate under unstable sunlight conditions. Simultaneously, the LSTM model learns from time-series data such as sunlight intensity, energy storage device voltage, and load current. After predicting the optimal control parameters, it uses prior control results and the output of the control system to comprehensively calculate the final control strategy. This scheme not only solves the impact of unstable energy storage device voltage on the system but also dynamically adjusts the control strategy based on real-time data, thereby improving the stability and efficiency of the water pump under different environmental conditions.

[0005] This invention proposes a solar water pump control system based on voltage compensation and LSTM. The control system includes a solar water pump, a solar panel, and a main board and display screen on the solar water pump for control purposes. The control system also includes a data acquisition module for real-time acquisition of time-series data of the solar water pump system, including light intensity, energy storage device voltage, load current, ambient temperature, and solar panel voltage.

[0006] A data preprocessing module is used to normalize the time-series data and construct the original sequence; The control parameter prediction module is used to input the original sequence into a pre-trained LSTM model and predict the control parameters. , , And the first solar panel voltage U in the next time period ’ .

[0007] The controller input calculation module is used to calculate the input based on the control parameters. , and The controller input is obtained by calculating the control system output, control system error value, and solar water pump system output from the previous time point.

[0008] The control module is used to adjust the speed of the solar water pump according to the input parameters of the controller.

[0009] The energy storage management module is used to adjust the voltage U of the first solar panel. ’ Energy storage management, when U ’ Greater than threshold U up At that time, the solar panels begin to charge and store energy for the energy storage device; when U ’ Less than threshold U low At that time, the energy storage device discharges to replenish the voltage, ensuring the water pump starts and runs, where Uup Greater than U low .

[0010] The controller input calculation module calculates the input based on the control parameters. , and The controller input is calculated from the control system output of the previous time point, the control system error value, and the output of the solar water pump system, as detailed below: In the formula, For controller input, This is the input to the controller at the previous time node. For difference operators, The system error value is equal to the difference between the setpoint and the output of the solar water pump system. This is the output of the solar water pump system; where , .

[0011] The solar water pump control system proposed in this invention also includes an online learning module. The system collects new time-series data through a data buffer, performs periodic batch online learning, and updates the LSTM model parameters. This ensures that the model can adapt to long-term environmental changes and continuously optimize the PID control strategy without manual intervention.

[0012] This invention proposes a solar water pump control method based on voltage compensation and LSTM, specifically including the following steps: Step 1: Collect time-series data of the solar water pump system. The time-series data includes light intensity, energy storage device voltage, load current, ambient temperature, and solar panel voltage U(t). Step 2: Normalize the time-series data and construct the original sequence; Step 3: Input the original sequence into the pre-trained LSTM model to predict and obtain the control parameters. , , And the first solar panel voltage U in the next time period ’ ; Step 4, according to the control parameters , and The controller input is obtained by calculating the control system output, control system error value, and solar water pump system output from the previous time point.

[0013] Step 5: Adjust the speed of the solar water pump according to the input parameters of the controller.

[0014] Step 6, based on the voltage U of the first solar panel’ Energy storage management, when U ’ Greater than threshold U up At that time, the solar panels begin to charge and store energy for the energy storage device; when U ’ Less than threshold U low At that time, the energy storage device discharges to replenish the voltage, ensuring the water pump starts and runs, where U up Greater than U low .

[0015] Specifically, the LSTM model in step 3 adopts a stacked LSTM structure. The first LSTM layer contains 128 units, mapping the original sequence to a 128-dimensional hidden state space and retaining the output of all time steps. The second LSTM layer contains 64 units, further compressing and extracting features, and outputting a 64-dimensional hidden state. The third LSTM layer contains 32 units and is set to return only the output of the last time step, thus condensing the temporal information into a fixed-length vector. Dropout is used after each layer to prevent overfitting. Finally, a fully connected layer maps the output to four dimensions, i.e., the output control parameters. , , And the first solar panel voltage U in the next time period ’ .

[0016] Specifically, the formula for the output layer of the LSTM model is as follows: in It uses the sigmoid activation function, ensuring that the output is in the range [0,1]. It is the hidden state of the last layer of the LSTM; and It is the weight and bias matrix.

[0017] Specifically, in step 4, based on the control parameters , and The controller input is calculated from the control system output of the previous time point, the control system error value, and the output of the solar water pump system, as detailed below: In the formula, For controller input, This is the input to the controller at the previous time node. For difference operators, The system error value is equal to the difference between the setpoint and the output of the solar water pump system. This is the output of the solar water pump system; where , .

[0018] Compared with existing technologies, this invention, by combining a voltage compensation mechanism and LSTM predictive control parameters, fully considers the impact of factors such as changes in ambient light and voltage fluctuations in the energy storage device on the system, resulting in stronger real-time adaptability and higher control accuracy. Furthermore, in determining the control parameters, it uses the input of the controller from the previous time point, the system error value, and the output of the solar water pump system for calculation, improving the accuracy and efficiency of control. This technical solution not only improves the automation control level of the solar water pump system but also significantly enhances the system's operational stability and energy efficiency in complex environments, demonstrating broad application prospects. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is the control system used in this invention; Figure 2 This invention proposes a block diagram of a solar water pump control system based on voltage compensation and LSTM. Figure 3 The graph shows the loss and mean absolute error results of the LSTM model used in this invention. Figure 4 This is a long-term adjustment trend chart of control parameters under the online learning mechanism.

[0021] Figure 5 The present invention proposes a flowchart of a solar water pump control method based on voltage compensation and LSTM. Detailed Implementation

[0022] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will understand that the invention can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0023] The physical structure of the solar water pump control system proposed in this invention is as follows: Figure 1As shown in the diagram, the key components of the system are clearly displayed, including solar panel 1, water inlet 2, water outlet 3, impeller 6, motor 4, mainboard, and display screen 5. The solar panel collects solar energy and converts it into electricity; the water inlet and outlet circulate the water; the impeller and motor drive the water pump; the mainboard integrates an LSTM model and controller for real-time data processing and adaptive control; and the display screen provides system status monitoring.

[0024] like Figure 2 As shown, the present invention proposes a solar water pump control system based on voltage compensation and LSTM. The control system includes a solar water pump, a solar panel 1, and a main board and display screen 5 on the solar water pump for control purposes.

[0025] The control system also includes a data acquisition module for real-time acquisition of time-series data of the solar water pump system, including light intensity, energy storage device voltage, load current, ambient temperature, and solar panel voltage.

[0026] The data preprocessing module is used to normalize the time-series data and construct the original sequence.

[0027] The LSTM model building module uses the original sequences to pre-train the model.

[0028] The control parameter prediction module is used to input the original sequence into a pre-trained LSTM model and predict the control parameters. , , And the first solar panel voltage U in the next time period ’ .

[0029] The controller input calculation module is used to calculate the input based on the control parameters. , and The controller input is obtained by calculating the control system output, control system error value, and solar water pump system output from the previous time point.

[0030] The control module is used to adjust the speed of the solar water pump according to the input parameters of the controller.

[0031] The energy storage management module is used to adjust the voltage U of the first solar panel. ’ Energy storage management, when U ’ Greater than threshold U up At that time, the solar panels begin to charge and store energy for the energy storage device; when U ’ Less than threshold U low At that time, the energy storage device discharges to replenish the voltage, ensuring the water pump starts and runs, where Uup is greater than Ulow.

[0032] The controller input calculation module calculates the input based on the control parameters. , , The controller input is calculated from the control system output of the previous time point, the control system error value, and the output of the solar water pump system, as detailed below: In the formula, For controller input, This is the input to the controller at the previous time node. For difference operators, The system error value is equal to the difference between the setpoint and the output of the solar water pump system. This is the output of the solar water pump system; where , .

[0033] Specifically, in the LSTM model building module, light intensity, energy storage device voltage, load current, ambient temperature, and solar panel voltage are used as inputs. This indicates the selection of control parameters. , , And the first solar panel voltage U in the next time period ’ As output.

[0034] The LSTM model employs a stacked LSTM structure. The first LSTM layer contains 128 units, mapping the original sequence to a 128-dimensional hidden state space and retaining the output of all time steps. The second LSTM layer contains 64 units, further compressing and extracting features, outputting a 64-dimensional hidden state. The third LSTM layer contains 32 units and is configured to return only the output of the last time step, thus condensing the temporal information into a fixed-length vector. Dropout is used after each layer to prevent overfitting. Finally, a fully connected layer maps the output to four dimensions, i.e., the output control parameters. , , And the first solar panel voltage U in the next time period ’ .

[0035] Specifically, the formula for the output layer of the LSTM model is as follows: in It uses the sigmoid activation function, ensuring that the output is in the range [0,1]. It is the hidden state of the last layer of the LSTM; and These are the weight and bias matrices. This is the output of the LSTM model.

[0036] The LSTM network was trained using five datasets, and validated using a sixth dataset. Each dataset contained 5000 data points, with the first 80% used as the training set and the last 20% as the test set. The mean absolute error (MAE) was used to analyze the accuracy of the LSTM network, as shown in the following formula: in This is the loss function value, where N is the number of samples. and The predicted and actual PID parameters are given, and the loss is minimized using the Adam optimizer.

[0037] During training, both training loss and validation loss gradually decrease with increasing epochs, with the validation loss decreasing more rapidly, indicating that the model gradually learns effective control strategies, such as... Figure 3 The model loss and mean absolute error are shown in the graph. The mean absolute error during training and validation gradually decrease, especially the training error, which drops significantly in the first 20 epochs, indicating continuous model optimization. Finally, after 100 epochs, both the training and validation errors approach a stable state, with the validation error less than 0.01, indicating that the model has achieved high accuracy.

[0038] The solar water pump control system proposed in this invention also includes an online learning module. The system collects new time-series data through a data buffer, performs periodic batch online learning, and updates the LSTM model parameters. This ensures that the model can adapt to long-term environmental changes and continuously optimize the PID control strategy without manual intervention.

[0039] like Figure 4 The long-term adjustment trend chart of the control parameters in online learning shows that, based on data from the actual learning cycle, the control parameters... , , The adjustments gradually became very minor. In the early stages, , , There are noticeable changes, indicating that the system underwent significant adjustments in the initial stage. However, in the later stages of the learning cycle, the changes in control parameters are very minor, especially... and The adjustments have almost ceased, indicating that the system has stabilized and the control parameters are close to optimal. Specifically, It gradually increased from 1.1666 to 1.1788. It increased from 0.3263 to 0.3264. The change is minimal, increasing from 0.2009 to 0.20097. These changes demonstrate that the optimization of the control parameters has approached its limit, and further adjustments will have a very limited impact on system performance, indicating that the optimization process is complete.

[0040] like Figure 5 As shown, this invention also proposes a solar water pump control method based on a combination of voltage compensation and LSTM, specifically including the following steps: Step 1: Collect time-series data of the solar water pump system. The time-series data includes light intensity, energy storage device voltage, load current, ambient temperature, and solar panel voltage U(t).

[0041] Step 2: Normalize the time series data and construct the original sequence, then use the original sequence to pre-train the model.

[0042] Step 3: Input the original sequence into the pre-trained LSTM model to predict and obtain the control parameters. , , And the first solar panel voltage U in the next time period ’ .

[0043] Step 4, according to the control parameters , and The controller input is obtained by calculating the control system output, control system error value, and solar water pump system output from the previous time point.

[0044] Step 5: Adjust the speed of the solar water pump according to the input parameters of the controller.

[0045] Step 6, based on the voltage U of the first solar panel ’ Energy storage management, when U ’ Greater than threshold U up At that time, the solar panels begin to charge and store energy for the energy storage device; when U ’ Less than threshold U low At that time, the energy storage device discharges to replenish the voltage, ensuring the water pump starts and runs, where U up Greater than U low .

[0046] Specifically, the LSTM model in step 3 adopts a stacked LSTM structure. The first LSTM layer contains 128 units, mapping the original sequence to a 128-dimensional hidden state space and retaining the output of all time steps. The second LSTM layer contains 64 units, further compressing and extracting features, and outputting a 64-dimensional hidden state. The third LSTM layer contains 32 units and is set to return only the output of the last time step, thus condensing the temporal information into a fixed-length vector. Dropout is used after each layer to prevent overfitting. Finally, a fully connected layer maps the output to four dimensions, i.e., the output control parameters. , , And the first solar panel voltage U in the next time period ’ .

[0047] Specifically, the formula for the output layer of the LSTM model is as follows: in It uses the sigmoid activation function, ensuring that the output is in the range [0,1]. It is the hidden state of the last layer of the LSTM; and It is the weight and bias matrix.

[0048] Specifically, in step 4, based on the control parameters , and The controller input is calculated from the control system output of the previous time point, the control system error value, and the output of the solar water pump system, as detailed below: In the formula, For controller input, This is the input to the controller at the previous time node. For difference operators, The system error value is equal to the difference between the setpoint and the output of the solar water pump system. This is the output of the solar water pump system; where , .

[0049] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0050] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0051] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0052] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0053] In the embodiments provided by this invention, it should be understood that the disclosed apparatus / terminal devices and methods can be implemented in other ways. For example, the apparatus / terminal device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0054] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0055] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0056] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals. The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A solar water pump control system based on voltage compensation combined with LSTM, characterized in that, The control system comprises a solar water pump, a solar panel (1), a mainboard and a display screen (5) arranged on the solar water pump to realize control; The control system further comprises a data acquisition module for acquiring time series data of the solar water pump system in real time, wherein the time series data comprises light intensity, energy storage device voltage, load current, ambient temperature, and solar panel voltage; A data preprocessing module is configured to normalize the time series data and construct an original sequence; The control parameter prediction module is configured to input the original sequence into a pre-trained LSTM model to obtain the control parameter through prediction , , and the first solar panel voltage U of the next time period ’ ; a controller input calculation module for calculating the controller input based on the control parameters 、 and , the control system output of the previous time node, the control system error value, and the output of the solar water pump system A control module is configured to adjust the rotation speed of the solar water pump according to input parameters of the controller; a storage management module for managing the storage according to the first solar panel voltage U ’ When U ’ is greater than a threshold value U up , the solar panel starts to charge the storage device; when U ’ is less than the threshold value U low , the storage device discharges to supplement the voltage to ensure the water pump starts and runs.

2. The solar water pump control system based on voltage compensation combined with LSTM according to claim 1, wherein the controller input is calculated in the control parameter 、 and , the control system output of the previous time node, the control system error value, and the output of the solar water pump system, and the controller input is calculated as follows: wherein, is an input to the controller, is an input to the controller from the previous time node, is a difference operator, is a control system error value equal to the difference between the setpoint and the output of the solar water pumping system, is an output of the solar water pumping system; wherein , .

3. The solar water pump control system based on voltage compensation combined with LSTM according to claim 1, further comprising an online learning module, wherein the system collects new time series data through a data buffer, periodically performs batch online learning, and updates LSTM model parameters.

4. The solar water pump control system based on voltage compensation combined with LSTM according to claim 1, wherein the LSTM model adopts a stacked LSTM structure, the first layer LSTM contains 128 units, maps the original sequence to a 128-dimensional hidden state space, and retains all time step outputs; the second layer LSTM contains 64 units, further compresses and extracts features, and outputs a 64-dimensional hidden state; the third layer LSTM contains 32 units and is set to return only the output of the last time step, thereby condensing the time series information into a fixed-length vector; Dropout is used after each layer to prevent overfitting; and finally, a fully connected layer is used to map the output to four dimensions, i.e. output control parameters 、 、 and the first solar cell panel voltage U of the next time period ’ .

5. The solar water pump control system based on voltage compensation combined with LSTM according to claim 1, U up greater than U low .

6. A solar water pump control method based on voltage compensation combined with LSTM, characterized in that, The control method specifically comprises the following steps: Step 1: acquiring time series data of the solar water pump system, wherein the time series data comprises light intensity, energy storage device voltage, load current, ambient temperature, and solar panel voltage U(t); Step 2: normalizing the time series data and constructing an original sequence, and pre-training the model using the original sequence; Step 3. The original sequence is input into the pre-trained LSTM model, and the control parameters are predicted , , and the first solar panel voltage U of the next time period ’ ; Step 4, controller input is calculated from control parameters , and , previous time node control system output, control system error value, solar water pump system output Step 5: adjusting the rotation speed of the solar water pump according to input parameters of the controller; Step 6, according to the first solar panel voltage U ’ energy storage management is carried out, when U ’ is greater than the threshold value U up , the solar panel starts to charge the energy storage device; when U ’ is less than the threshold value U low , the energy storage device discharges to supplement the voltage to ensure the water pump starts and runs.

7. The solar water pump control method based on voltage compensation combined with LSTM according to claim 6, wherein the controller input is calculated according to the control parameters, the control system output of the last time node, the control system error value and the output of the solar water pump system in step 4, and the specific calculation is as follows: 、 and the control system output of the last time node, the control system error value, the output of the solar water pump system. wherein, is an input to the controller, is an input to the controller from the previous time node, is a difference operator, is a control system error value equal to the difference between a setpoint and an output of the solar water pumping system, is an output of the solar water pumping system; wherein , .

8. The solar water pump control method based on voltage compensation combined with LSTM according to claim 6, further comprising an online learning mechanism, wherein the system collects new time series data through a data buffer, periodically performs batch online learning, and updates LSTM model parameters.

9. The solar water pump control method based on voltage compensation combined with LSTM according to claim 6, wherein the LSTM model adopts a stacked LSTM structure, the first layer LSTM contains 128 units, maps the original sequence to a 128-dimensional hidden state space, and retains all time step outputs; the second layer LSTM contains 64 units, further compresses and extracts features, and outputs a 64-dimensional hidden state; the third layer LSTM contains 32 units and is set to return only the output of the last time step, thereby condensing the time sequence information into a fixed-length vector; Dropout is used after each layer to prevent overfitting; and finally, a fully connected layer is used to map the output to four dimensions, i.e. output control parameters 、 、 and the first solar cell panel voltage U of the next time period ’ .

10. A solar water pump control device, comprising a solar panel (1), an inlet (2), an outlet (3), an impeller (6), a motor (4), a mainboard and a display screen (5), and being configured to execute the solar water pump control method based on voltage compensation combined with LSTM according to any one of claims 6-9.

Images